Selling to agents: what AI assistants say about Circle today, and what I'm thinking we do about it

ToArina, JKFromAndyDateSep 24, 2026ReWhy assistants don't recommend us, where the wrong facts come from, and an idea for how we could fix it. I'd love your reaction.

Arina, JK, I wanted to put everything I've learned about how AI assistants see Circle in one place, along with what I think we could do about it and roughly in what order. I'm sharing it to get your take, not to hand you a plan. You may already be on the same page with a lot of this. But I've been sitting with this research for a week and a half now, and the thing I keep coming back to is how much of the fix is small. Not easy necessarily, but small. A lot of it is a few hours of work, not weeks.

Here's the short version. When someone asks ChatGPT, Claude, Perplexity or Gemini what to build a coaching business on, we come up when they type the word "community," and for almost nothing else. When the buyer names us, the assistant knows what we do and gets it mostly right. What it gets wrong is our prices, which it has from 2024, and everything we shipped in 2026, which it can't find. And when I went looking for where the wrong facts come from, most of them come from us. Our own pages, our own machine-readable description, our own blog posts.

My point isn't that the site is bad. A lot of it is fine and some of it is very good. My point is that we're currently the source of our own worst positioning, and that's the part we can fix fastest. That's why I'm writing to the two of you and not to engineering.

What assistants say about us today

I had Quinn, my AI research assistant, run three passes over the last two weeks. The first tested what assistants recommend across 28 things buyers say they want to build and 20 real buying situations, each asked cold (what the model remembers) and live (what it finds when it searches). The second fetched all 215 reachable pages on circle.so the way an AI crawler fetches them, without JavaScript, and gave every page a verdict. The third swept the 149 English blog posts neither of the first two had read. The full tables are in the appendix, and they're worth a read, because if we do this you'd be writing to these terms. Here's what came out of it.

16 of 20
buying situations where the assistant gets our capabilities right, once we're named
9 of 20
situations where we don't come up at all unless the buyer names us
2 of 8
buyer questions where circle.so itself was among the pages the assistant read
11 to 14
words a crawler gets from each of our 29 release posts on community.circle.so (checked Sep 23)
56 of 149
blog posts say "community platform"; 23 carry a stale plan or price word
75 · 75 · 14 · 10 · 8
page verdicts: keep, fix, rewrite, merge, retire

Assistants recommend us when the buyer says "community," and almost never otherwise. Across 28 terms buyers use, we're strong for paid community and, oddly, for "Discord alternative." For courses, memberships, newsletters, coaching and selling digital products, we're either missing or filed under "community-first," which is a smaller box inside "all-in-one." Kajabi owns "digital business platform" and "all-in-one" in the pages assistants read. SamCart calls itself a digital business platform too.

The ones I find most interesting are where the assistant already thinks of us and then can't find a page to point to. Mastermind, challenge, accelerator, signature program, paid club. For each of those, the model puts us first or second from memory, then it searches the web, finds nothing on circle.so with that word on it, and drops us from the answer. Mighty owns mastermind with a single page. For signature program and paid club, nobody ranks at all. Those are probably the cheapest wins we have, because one page on our site would change the answer the next time somebody asks.

When the buyer names us, the assistant mostly knows what we do. Right or nearly right in sixteen of twenty situations. What it gets wrong is the same every time: the 2024 four-tier ladder with 4 percent fees, a retired Enterprise tier at $360 or $419 or $499, a $39 or $49 Basic that has never existed. And nothing we shipped in 2026 shows up anywhere. Public RSVP, public webinars, AI-built sales pages, post-checkout upsells, Forms as application intake, the migration service, SSO by plan, AI agents on Plus. All of it is sitting in one-paragraph entries on our product updates page, which is a changelog, and nobody asking a buying question ever gets pointed to a changelog.

The wrong facts assistants repeat about us mostly come from us. This is the one that surprised me. Our llms.txt, the file on the site written specifically for machines to read, literally says "Circle is community-first, not course-first or marketing-first." On circle.so/pricing, the only $49 a crawler can see is the "Customizable profile fields" add-on at $49/month, sitting in the plan comparison table where the column headers don't come through, and Capterra and GetApp still list us "from $49," which was our Basic plan price back in 2023. Put those two together and every assistant thinks we have a $49 tier today. Our own comparison pages and listicles carry wrong fees and retired plan names. And when the assistant reads our top-ranked "best coaching platforms" post, every sentence it uses about us comes from that post, and it still recommends someone else, because the post describes us in retention words like engagement and community rather than in selling words.

Buyers don't call it a community. Kit's 2026 creator survey has coaching and consulting at 31.6 percent of what creators sell, courses and digital products at 23.1, newsletters at 16.4, and communities at 5.1. The category we're named for is five percent of what creators sell. In our own customer stories nobody calls their thing a community either. They name it: Part-Time YouTuber Academy, The Lab, London Writers' Salon, PMM Camp, Notion Mastery. We're the ones who call it a community. Our customers don't.

Where we stand, at a glance

Strong other people's pages recommend us without being askedFragile we only show up when our own pages happen to rankOpportunity the assistant already thinks of us but can't find a page of ours to point toAbsent we don't come up, and the assistant doesn't think of us either
By who's asking
SituationStatusWhat's going on
When the buyer names usStrongWhen somebody asks about Circle by name, the assistant gets our features right or nearly right in sixteen of twenty situations. Where it's wrong, it's our prices and anything we shipped this year.
When the buyer doesn't name usFragileThe assistant thinks of us first or second in nine of twenty situations from memory. But once it searches the web we disappear in nine of twenty, because the pages that rank are niche tools and roundups we aren't in.
Assistants that recommend platformsFragileWe get recommended when the buyer says "community." For everything else we're either missing or filed under "community-first," and the pages the assistant reads about us are written by our competitors.
Agents that operate Circle for a customerOpportunityWe have a good MCP server, and it isn't listed in the two places agents look first, the Claude connector directory and the official MCP registry. Getting listed is two forms.
Agents that build apps for peopleAbsentNone of the app builders people use (Lovable, Bolt, Replit) have us in their stack. Our SDK hasn't been touched since 2024, and an agent can't try us without a $199 plan. The one thing nobody in our category has done yet is let an agent set up an account for someone.

The full grid, all twenty business types, is in appendix A. The six rows I'd point you at are the ones marked Opportunity: mastermind, challenge, accelerator, signature program, paid club, and the one that doesn't have a name yet, where someone asks an AI to set the whole business up for them. On each of those the assistant already thinks of us first or second and has nothing on our site to point to.

Why it happens: what assistants read about us

An assistant reads two kinds of pages about us, other people's and ours, and right now both are working against us.

Start with other people's pages. The pages an assistant pulls up about Circle are mostly written by our competitors and their affiliates. Across the first benchmark, circle.so came up on two of eight buyer questions. The domains that came up most were group.app, schoolmaker.com and mightynetworks.com. Mighty ranks first for "Skool vs Circle" and "Circle alternatives." SchoolMaker owns "Circle pricing." GroupApp owns "Circle review." A buyer researching us reads our competitors describing us. The reason is money. Skool and SchoolMaker pay their affiliates 40 percent recurring, and until June we paid $120 once. We moved to $100 plus 10 to 20 percent recurring in June, but the pages assistants read haven't caught up, because nobody has told the reviewers. That one's on Justine and me, and I'll come back to it further down.

Now our own pages. The crawl found that a crawler can read almost every page on circle.so that matters, and on most of them what it reads is the old story. Blog posts, compare pages, feature pages and the product updates page all come through in full. And what they say is community platform: the machine-readable description says community-first, the plan comparison table on /pricing puts a $49/month add-on price where a crawler reads it as a plan price, thirteen pages still say "co-pilot" for a product we renamed (as of Sep 23: /compare and five of the head-to-head compare pages, /platform/creators, /platform/courses, /platform/community-managers, /membership, two blog posts and llms.txt), eight business-type pages share a template whose structured data asks "How does Circle compare to other community platforms?", and not one feature or compare page names anything shipped in 2026. Two things a crawler can't read at all. The help center returns a 403 and a "Verifying you are a human" page to every AI crawler that was tried. And our release notes: circle.so/product-updates itself comes through fine, about 950 words linking 29 releases, but every one of those links goes to a post on community.circle.so that renders through JavaScript, so a crawler gets 11 to 14 words, the title and one line. So every "can Circle do X" question gets answered from what the model remembered from 2024.

Everything in this section is checkable in a minute. Open the page, view the source, and search for the phrase. What's in the source is what an assistant sees. Or paste a release note URL into ChatGPT and ask it what shipped. If a fact in this memo doesn't match what you find, tell me, because the site moves. Two things already changed between the Sep 14 crawl and my re-check on Sep 23: "Content co-pilot" is off the pricing page and off most of the platform pages, and the $49 on the pricing page is now the only one on the core pages.

The count: 182 verdicts across 215 pages

GroupPagesKeepFixRewriteMergeRetire
Core commercial pages: home, pricing, platform, AI, MCP, Plus, migration35918350
Feature pages and standalone pages421618134
The comparison layer: compare hub, head-to-heads, alternatives posts321614101
Blog listicles, business-type how-tos, AI posts341314412
Customer stories, event hub, localized pages, structural files and subdomains392111511
All182757514108

Rows are page-level verdicts from the five review tables in the appendix. Localized pages, the event hub and careers are counted as sets, so the page count is below 215.

The good news is that a person and an assistant want the same thing from a page. It should say what the product does today, in the words a buyer would use, with a number next to it. That's really the whole idea.

The work, sorted by return on time

I've sorted everything into five buckets, because I think the effort is the thing that gets lost when a list like this lands on someone's desk. Bugs, then the basic blocking and tackling we should have done already, then mistakes on our own pages that are easy to correct, then the real builds, then the engineering and product work that runs alongside. The first three buckets are hours each. The fourth is days each and carries most of the upside. Every row has a score from one to five, where a five is something I don't think needs any debate, and a link to the full write-up, which has the specifics down to the exact line to change. In the Detail column, "card" means one of the eighteen site-fix write-ups in appendix B, "change" means one of the five changes the research ranked highest, also in B, and "grid" and "rulings" point at the tables in this memo.

If I had to pick one thing, it'd be rewriting llms.txt, which is about an hour. The first three buckets together are maybe two or three weeks of work between two people, and once they're done we've stopped feeding the assistants wrong facts ourselves. The term pages and the fact pages in bucket four are where I think new trials actually come from. Everything before them is clearing the ground so those pages get believed when they go up.

Bucket one: things that are just broken

All of it is hours, and none of it needs a decision from anyone.
#What to doWhich wrong answers it fixesEffortOwnerScoreDetail
1Fix the canonical tag on /mcp and list the page. Point the canonical at /mcp (it points at /chat today), add it to the sitemap and llms.txt, link api.circle.so/mcp, add "since April 2026" and the two access modes.One line of markup tells search engines our MCP page is a duplicate of the chat page. It hides the page that proves we're ahead of Kajabi and Mighty on this.MinutesJK, web eng5card 3
2Fix the bugs a visitor can see today. Two "Builder block error" messages on /membership. The /plus/branded-app H1 renders as "Earn more with your own." The Analytics Manager card on /ai carries the Notification Manager description. Six of seven featured creators on /discover are captioned "Productivity expert." A "Launch your course" heading on /platform/podcasters. /compare says 14 competitors in the subhead and 17 in the body.The membership page is a money page and it's broken in public. It's probably the cheapest credibility fix we have.HoursJK, web eng5card 6
3One retire, redirect and cleanup ticket. Lorem ipsum on /academy. A Portuguese sales form under an English "Book a Demo" title at /demo-video. An internal note published on /migration/courses ("This one might be worth discussing internally"). /v3, /affiliate-faqs, the two year-review stubs, the two 2023 Copilot posts. Three copy-paste errors in customer stories. 301s for the eight URLs agents guess (/customers, /case-studies, /compare/kajabi, /knowledge-base, /help, /fees, /about, /changelog all 404).Old pages are what the models end up quoting. And right now every URL an agent would guess for our pricing or our help center returns a 404, which is a dead end for every "what does Circle cost" question.A dayJK, web eng5card 6
4Noindex the nineteen event-hub transcripts from 2021 to 2023.About 190,000 words that say "community" fifty to two hundred times a page. It's the biggest block of text on the site after the blog, and all of it says the old positioning.An hourWeb eng5card 6

Bucket two: basic blocking and tackling

This is the basic stuff any company that wants assistants to recommend it has to have in place. Hours to a day each.
#What to doWhich wrong answers it fixesEffortOwnerScoreDetail
5Rewrite llms.txt and regenerate llms-full.txt. Open with the digital business definition, name the business types, prices and fees. Add the 2026 capabilities with plan gating, MCP, Discover, the platform pages, all compare pages, all 25 stories, product updates. Regenerate the full file from product, pricing, compare and platform pages first, blog last.This is the file written for machines, and it says "Circle is community-first, not course-first or marketing-first." It also says 18,000 communities where the site says 20,000, and links a retired Copilot page. It's an hour of writing to fix what I think is the single worst sentence on the site.An hour to writeJK; web eng for the full file5card 2
6Fix the plan comparison table on /pricing, and the shared plan block. Plan availability as text instead of icons, monthly prices next to annual, the add-ons in their own clearly labeled table, a fee example at three revenue levels, product and offer structured data, a markdown copy of the page. ("Content co-pilot" was in the shared block on Sep 14; it's off /pricing as of Sep 23 and still on thirteen other pages, which is item 13.)The only $49 a crawler sees on circle.so/pricing is the "Customizable profile fields" add-on at $49/month in the plan comparison table, and with the column headers lost it reads like a plan price. Capterra and GetApp still say "from $49." Easy to check: view the source of circle.so/pricing and search for $49. And third parties invent $99 and $129 monthly prices because we don't publish a monthly figure.HoursJK, web eng5card 1
7Fix the review-site profiles and categories. Capterra and GetApp show "from $49"; GetApp files us under HR software. Add G2 categories for coaching, course, membership and newsletter, not only community.The models use G2's categories to decide what a product is. If we're not in the coaching and courses categories, we don't exist for those questions.HoursJK5change 3
8Publish three fact pages. Pricing and fees explained, with total-cost math at $5K, $20K and $40K a month. What shipped in 2026, with plan gating. What Circle is, in plain facts (customer count, plans, one paragraph per business type).Every one of these missing pages showed up as a wrong answer in the testing. Nothing on our site corrects the hidden-fee story, and when you ask "what is Circle" you get the crypto company.Hours to a day eachJK5card 14
9Make product updates something an assistant can actually read. The fix, in order. First, the quick steps: give each release its own page on circle.so with the full release note in the HTML (not a link to the community post), add dated article markup, and add the product-updates page and each release to llms.txt and the sitemap. Second, rewrite the entries one per capability, in the buyer's words, with the plan it's on, the date, and links to the feature page and help article. Third, put a "new in 2026" block on the payments, events, website builder, email hub, workflows, forms and AI pages so the feature pages themselves carry the facts. The longer-term product fix, serving community post bodies as real HTML, is item 29.It's the only first-party record of what we shipped this year, and a crawler can't read it. Checked Sep 23: circle.so/product-updates links 29 releases and comes through at about 950 words, but each release post on community.circle.so gives a crawler 11 to 14 words, the title and one line, because the body renders through JavaScript. The page isn't in llms.txt either. Try it yourself: paste any release URL into an assistant and ask what shipped. This is why no assistant knows anything we shipped in 2026.A day or two for the first step, then about two weeksJK, web eng4 (first step 5)card 9, the full write-up
10Retitle the customer stories with the business type and the number, and add the six missing stories to llms.txt. Seasoned Sales ($15K MRR in two months), SPI ($700K+), Roofing and Solar, Talk Nerdy to Me ($37K+ a month off Kajabi) first.What gets us cited is a named customer with a number. We have that proof for our weakest business types, and the stories are titled "community."A morningArina5card 13
11Send product the nine rulings as one ask, and commission the coach customer story.Several pages can't claim things until product answers. And customer proof takes the longest of anything here, so I'd start it first.Hours to askJK, Arina5rulings
12Tag AI referrers in analytics and HubSpot (chatgpt.com, perplexity.ai, claude.ai, gemini, copilot) and add an AI option to "how did you hear about us."Without this we can't tell whether any of this worked. AI referrals show up as direct traffic today.HoursRevOps5change 4

Bucket three: mistakes on our own pages that are easy to correct

Wrong facts and old positioning on pages that already rank. Each one is a morning to a day, and these are the exact sentences teaching every assistant the positioning we're moving away from.
#What to doWhich wrong answers it fixesEffortOwnerScoreDetail
13Correct the wrong Circle facts on pages that already rank. The Kajabi-alternatives fee range, the Patreon-alternatives math, "read-only" Circle AI, "flat regardless of list size," "join the waitlist," "in 2025," API on all plans, order bumps on the Facebook compare page, five pages that disagree on whether we have certificates, and "co-pilot" on the thirteen pages that still say it.These are the exact stale facts the benchmark heard back from assistants, and they're on our pages, so we're the source. And a wrong price on a page that ranks is what other sites copy next.A morningArina (blog), JK (site); product for two rulings5card 5
14Fix the shared template on the eight /platform business-type pages. The FAQ structured data asks "How does Circle compare to other community platforms?" The closing band is titled "The complete community platform." Two pages answer the migration FAQ with a customer quote about launching a course.Eight pages say "community platform" in the structured data, which is the part of the page an engine reads first, and one template edit fixes all eight.HoursJK, web eng5card 7
15Put our price and a fee table on the Skool, Mighty and Kajabi compare pages. Port the table already on the Bettermode page. Fee math at $5K, $20K and $40K a month. FAQ structured data.Those three pages carry no Circle price at all. It's part of why SchoolMaker owns the search "Circle pricing" and Mighty ranks first for "Skool vs Circle."HoursJK5card 4
16Rewrite our own entry in the listicles that already rank. Best coaching platforms, community, membership, creator, event and membership-site posts, and the Kajabi-alternatives verdict: selling verbs, the 2026 stack, and we win on conversion.Every sentence the assistant used about us in the coaching benchmark came from our own top-ranked post, and it still didn't recommend us, because that post describes us in retention words. We did the hard part, which is ranking, and then handed the recommendation to someone else.Half a day per postArina5card 11
17Take the newsletter promise down until January. /platform/newsletter-creators and the pages that promise newsletter sending.We rank first for "newsletter plus paid community" on a promise a trial can't use until January. That's going to turn into refund requests.An hourJK5change 3
18The blog sweep: 56 posts say "community platform," 23 carry a stale plan or price word. Start with the two that carry most of it: the 2024 guide to choosing a community platform (14 mentions) and the creator tech stack post we published on Sep 8 of this year (12). Then the rest in batches of ten; most of them are a "community platform" replacement and a paragraph. Full counts per post: blog-sweep-2026-09-17.csv.Perplexity and ChatGPT both cite our blog, and fifty-six of our posts are telling them the positioning we're moving away from.An hour or two per post for most of the 56, half a day for the ones that rank; batches of tenArina5card 11

Bucket four: the new pages. Days each, and this is where I think the trials come from

New pages and rebuilt pages. This is the part that turns the Opportunity rows in the grid above into Strong ones.
#What to doWhich wrong answers it fixesEffortOwnerScoreDetail
19Term pages. Rewrite mastermind, group coaching and cohort on their existing URLs as platform pages: the buyer's word in the title, the mechanics named (live rooms, leaderboards, check-ins, applications, paywalls), one customer with a price, FAQ structured data. Then the ones we don't have: accelerator, challenge, signature program, paid club, Discord alternative for creators, membership site for podcasters, a webinar that sells a course, high-ticket program.For six of these the assistant already ranks us first or second from memory, searches, finds nothing to cite, and drops us. Mighty owns mastermind with one page. I think these are the pages most likely to change an answer, and it happens within weeks of publishing.Two days for the first, a day each after thatArina writes, JK reviews5 for six; 3 for group coaching and cohortcard 10
20Put 2026 on every feature page. Events: public RSVP, public webinars, paid tickets, every guest becomes a contact, and "webinar" in the title. Payments: post-checkout upsells, installments, Apple and Google Pay, fee math. Workflows: sale triggers, and Circle AI writes the workflow from a sentence. Email hub, analytics, customization, gamification: fix the FAQs, name the 30-day challenge.Feature pages are what a buyer question retrieves, and not one of ours names a 2026 capability in a buyer's word. The assistant hedged on webinar registration, denied we have upsells and repeated the 20-person live-room cap because the page it read never mentioned any of it.A week or two across ten pagesJK; product on order bumps5 for payments, events, builder; 4 for the restcard 8
21Circle-authored comparisons that rank. Skool vs Circle, Circle vs Mighty, Circle vs Kajabi, written straight, with the fee math and the feature truth. Then /compare/circle-alternatives and a first-party reviews page. And a "what Circle sells that X can't" paragraph plus FAQ structured data on every existing compare page.Compare pages are what assistants cite most. Mighty owns "Circle alternatives," GroupApp owns "Circle review." A buyer researching us reads our competitors describing us.A day or two per pageArina (the posts), JK (the compare block)4card 12
22The coaching page, as the first business-type page. A headline in their words about the time ceiling, one sourced dollar line in the hero, their money-makers in order with what Circle AI does for each, one story with a dollar figure in the title, the product list in the new nouns, the trial.Coaching is named on the homepage and it's a 404 at every URL a coach tries. It's the biggest thing creators sell, 31.6 percent per Kit's 2026 survey, and the coaching money problem, the hours ceiling, is the easiest one to write about. Seasoned Sales is the lead story.DaysJK5grid
23Make selling a product. Three pages that don't exist: sales pages (live today, AI-built), upsells (live today), and funnels as the real sequence we support: public webinar, RSVP, email sequence, checkout, upsell.Digital products is the most common thing buyers say they want to sell, and we have no page. The deciding feature in that world is one-click upsell at checkout, which we've had since early 2026 and nobody can find.Days eachJK; product for order bump timing5card 14
24The website builder page, rebuilt as the template for every product page. Lead with prompt-to-page and recreate-from-URL. Show the output, then the editor.Zero mentions of AI on it, weeks after the AI builder shipped free on every plan. Every other product page would be modeled on this one.DaysJK, Rudy5card 8
25The naming system. One line per thing we sell: the noun, the money outcome, what Circle AI does. The noun becomes the nav label; the line becomes the H1 on its page and is reused in ads and the deck.Everything in the later weeks inherits it. It's a document, so it's the easiest thing on this list to change our minds on. I'd like to review it with you before it touches a page.A few days to draftJK, with me5card 18
26The other four business-type pages (education, creator, community, membership) on the coaching template, then nav, trust line, title tags and the /platform hub from the naming system.Nobody in the market has a clean page for each type. Every vertical title tag still says "Community Platform for." The shared pieces go last so we only write them once.Days eachJK5card 18

Bucket five: engineering and product work that runs alongside

I want this scheduled now, but nothing above waits on it.
#What to doWhich wrong answers it fixesEffortOwnerScoreDetail
27List our MCP server where agents look. The Claude connector directory and the official MCP registry. Two forms, about a day of engineering.The only place agents find us that reports usage back to us, and a credibility artifact for Plus deals today. Zero community platforms come back for "best MCP for a community business."A dayProduct, eng5change 1
28Make the help center readable. Exempt known crawlers from the Cloudflare challenge, server-render article bodies, publish a sitemap. As an interim, mirror the top hundred articles as plain HTML on circle.so.help.circle.so returns a 403 and a "Verifying you are a human" page to every AI crawler that was tried (re-checked Sep 23). So every "can Circle do X" question is answered from what the model remembered from 2024.Weeks; the mirror is daysWeb eng4 (mirror 5)card 15
29Public community spaces serve post bodies in the HTML, plus a robots file on community.circle.so.Our own release notes are unreadable to a crawler, and so is every customer's public content. This is the bigger bet: every business on Circle findable by assistants without the customer doing anything.MonthsProduct3card 16
30Localized pages, then the name. Port the new positioning to the German, French and Portuguese homepages; noindex the 180 URLs that serve English under a foreign language tag. A real Product Hunt listing for circle.so; start the Wikipedia notability case.A third of the sitemap either calls us a community platform or mislabels its language. ChatGPT takes half its citations from Wikipedia and we have no article; the stablecoin company does.Days; the Wikipedia case is slowJK, web eng; PMM, PR3card 17

Two things on ownership. I've put the site pages with JK and the blog and long-form pages with Arina, but that's a first guess, and by all means swap rows between you. I care about the order more than the names. And I've left the homepage and pricing page redesigns off this list on purpose, because Sid, Rudy and I own those end to end. The fixes to the current pricing block and /pricing in item 6 are different. Those are corrections to what's live today, and I'd put them with JK. Everything else I'd put with the two of you, if you're up for it.

An idea I'd love your reaction to: 30 days, one improvement live every day

Here's something I've been playing with, and I want to be clear it's an idea, not a plan I've decided on. What if we took thirty working days, September 28 to November 6, and put something live on circle.so every single one of them? I've laid out what that could look like below, with the item numbers in parentheses so every day traces back to a row above. I haven't assumed anything about what else is on your plates, and I know you both have a lot on. So take this as me showing you what I'm thinking, and tell me what's realistic.

My point isn't that this calendar is right. You'd move things, and you should. My point is that almost everything on it is a day or less, the bigger pages can run in the background and land on their day, and if it worked, by Friday, November 6 we'd have shipped the whole first three buckets, eleven term pages, three fact pages, three selling pages, comparison pages written by us, five business-type pages, and most of the 56 blog posts corrected. That's most of the list in six weeks, and it would land alongside the new homepage and pricing page. Arina's column is the tighter one, and if a term page needed a second day I'd let the blog batch slip, not the page.

A few things I'd suggest, whatever the calendar ends up looking like. The first is facts before pages. The first week is nothing but corrections, so that when the term pages land in week three the assistant reading them doesn't also find a $49 tier two clicks away. The second is that claims wait on the rulings. The nine product questions would go out on day one, and no page would claim certificates, order bumps or a currency count until product answers. The third is one page of each kind before the family. The coaching page and the website builder page would go live first, so the three of us can react to one of each before the other business types and product pages get modeled on them.

DayDateJK shipsArina ships
Week 1: bugs and facts. Everything here is hours, and none of it needs a meeting.
1Mon Sep 28llms.txt rewritten and live. Product rulings sent as one ask. (5, 11)The wrong facts on the ranking blog posts corrected: Kajabi-alternatives fee range, Patreon math, "read-only" Circle AI, "flat regardless of list size," "join the waitlist." (13)
2Tue Sep 29/mcp canonical fixed, page in the sitemap and llms.txt. (1)The Sep 8 creator tech stack post rewritten in the new positioning. (18)
3Wed Sep 30The visible-bug ticket shipped with web eng: /membership errors, branded-app H1, the /ai card, /discover captions, 14 vs 17, the podcasters heading. (2)The 2024 guide to choosing a community platform rewritten, or retired and redirected. (18)
4Thu Oct 1Review-site profiles: four G2 categories added, Capterra and GetApp prices and the HR category fixed. (7)Customer stories retitled with the noun and the number; the three copy-paste errors fixed. (10)
5Fri Oct 2The shared pricing block and /pricing, with web eng. Newsletter promise taken down. (6, 17)Six missing stories added to llms.txt. The coach customer story commissioned. (10, 11)
Week 2: the facts that have no page yet. RevOps tags AI referrers this week (12).
6Mon Oct 5Pricing-and-fees page live, with the fee math. (8)Our entry in "best coaching platforms" rewritten in selling verbs. (16)
7Tue Oct 6"What shipped in 2026" page live, with plan gating. (8)The "entire online business" post rewritten so we're in the recommended set. (16)
8Wed Oct 7The /platform template fixed: FAQ schema, closing band, migration FAQ, the 2026 block. (14)Best membership platforms and best community platforms posts: our entry rewritten. (16)
9Thu Oct 8Retire, redirect and cleanup ticket shipped: /academy, /demo-video, the internal note, the stubs, the event-hub noindex, the eight 301s. (3, 4)Best creator platforms, event and membership-site posts: our entry rewritten. (16)
10Fri Oct 9Price and fee table on the Skool, Mighty and Kajabi compare pages. (15)The Kajabi-alternatives verdict rewritten. Blog sweep batch one: ten posts. (16, 18)
Week 3: term pages start; feature pages get 2026 on them. Claims wait on the rulings from day 1.
11Mon Oct 12Payments page: upsells, installments, Apple and Google Pay, fee math. (20)Mastermind page rewritten on its existing URL, as a platform page. (19)
12Tue Oct 13Events page: public RSVP, webinars, paid tickets, guest-to-contact; "webinar" in the title. (20)Challenge page, new. (19)
13Wed Oct 14Workflows page: sale triggers, Circle AI writes the workflow. (20)Accelerator page, new. (19)
14Thu Oct 15Product updates, first steps: dated markup, all 32 releases rendered, "new in 2026" blocks on seven feature pages, with web eng. (9)Signature program page, new. (19)
15Fri Oct 16Naming system draft to me; I'll turn it around within two working days. "What Circle is" facts page live. (25, 8)Paid club page, new. (19)
Week 4: the rest of the term pages; the two template pages land.
16Mon Oct 19Kajabi compare page rewritten as the digital business page. (21)Discord alternative for creators page, new. (19)
17Tue Oct 20Email hub, analytics, customization and gamification FAQs fixed; the 30-day challenge named. (20)Group coaching page rewritten. (19)
18Wed Oct 21Website builder page live, rebuilt with Rudy. (24)Cohort page rewritten. (19)
19Thu Oct 22Coaching page live. (22)Membership site for podcasters page, new. (19)
20Fri Oct 23Sales pages page live. (23)Blog sweep batch two: ten posts. (18)
Week 5: the comparison layer and the selling pages.
21Mon Oct 26Upsells page live. (23)Skool vs Circle, written by us. (21)
22Tue Oct 27Funnels page live: webinar, RSVP, sequence, checkout, upsell. (23)Circle vs Mighty, written by us. (21)
23Wed Oct 28Product updates, second step: per-release URLs and capability-style entries. (9)Circle vs Kajabi, written by us. (21)
24Thu Oct 29"What Circle sells that X can't" block and FAQ schema on every compare page. (21)/compare/circle-alternatives, new. (21)
25Fri Oct 30First-party reviews page live. (21)Blog sweep batch three: ten posts. (18)
Week 6: the business-type family, and the shared pieces written once.
26Mon Nov 2Creator business-type page. (26)High-ticket program page, new. (19)
27Tue Nov 3Education business-type page. (26)"A webinar that sells a course" page, new. (19)
28Wed Nov 4Membership business-type page. (26)Blog sweep batch four: ten posts. The tail of the 56 carries into November. (18)
29Thu Nov 5Community business-type page. (26)Every term page and story cross-linked; the six new stories linked from their term pages. (10, 19)
30Fri Nov 6Nav, trust line, title tags and the /platform hub from the naming system. (26)Month-one scoreboard: the thirty-prompt check re-run against the Sep 13 baseline, next to the first AI-referral report from RevOps. (12)

What runs alongside, not on the calendar: RevOps tags AI referrers in week one (12). Product lists the MCP server and answers the rulings (27, 11). Web engineering would be on this calendar on days 1, 3, 5, 8, 9 and 14, plus the event-hub noindex, the 301s and the help-center mirror (4, 9, 28), and I haven't asked them yet. If we decide to do this, lining them up is on me, along with Rudy's time on the website builder page. The localized pages and the Wikipedia case (29, 30) can wait until December. Justine's affiliate content program pairs with Arina's listicle work so the third-party pages say the right thing too.

What we need before certain pages can say certain things

Each of these is a fact the site currently states two different ways, or a fact an assistant got wrong because no page states it. JK, if you're up for it, I'd send this to product as one ask early on, because the term pages in week three depend on the answers.

Rulings needed from product before pages can say it

Two decisions are mine, and I'll get to them quickly so they don't hold anything up. Whether to close the rest of the affiliate gap to Skool's flat 40 percent, and either way, how we get the June terms in front of the reviewers who still publish the old ones. And whether we put read-only MCP access on every plan as a 90-day test, the way Kajabi did.

What not to claim yet. As of the September audit these were internal-only or marked "up next" in public: order bumps, workflow branching, forms 3.0, breakout rooms, reactions on posts, and upgrade-at-checkout upsells. Newsletters ship in January 2027. The Thinkific and Substack comparison pages already claim branching, and they shouldn't until it's announced. Please confirm every one of these dates with product before a page says it. I'd rather we under-claim for a month than have a page promise something a trial can't find.

How we'll know it's working

AI referrals, tagged and reported monthly, with homepage landings broken out, since most ChatGPT referrals land on the homepage. Today we can't see any of this, because AI referrals show up as direct traffic. The first month gives us a baseline, and I'd rather not quote anyone else's numbers as ours before we have it.

The monthly check of what assistants recommend. Thirty short prompts in buyers' words, drawn from the 28 terms and 20 situations in the appendix, scored for whether we show up, where we rank, how we're described, and who gets cited. Quinn runs it by hand for free against the September 13 baseline. The grid above is the scoreboard. The Opportunity rows, where the assistant believes in us but has no page to cite, should turn Strong one at a time as the term pages go live, and the wrong prices should stop coming back once the pricing block and the fees page are up.

Organic traffic holding. Part of this is defense. B2B software organic traffic is down about 29 percent year over year as AI answers absorb the clicks, and if we keep what we have while that's happening, that's a win, and I want us to say so.

And I'd give every page a quarter before we decide it didn't work. Live search results move in weeks. What the models remember from training moves over a year or two. We'll see the first kind this year and the second kind next year, and that's fine.

I know this is a lot of pages. But the first bucket is about a day of work, the first three together are a couple of weeks, it's all facts, and it fixes the sentences that are teaching every assistant on the internet the exact positioning we're trying to move away from. I'd love to see those gone before the new homepage goes up.

Mostly I want to know whether this feels right to the two of you, and what you'd change. Happy to go through any of it live. And if you think I've got the order wrong, or the whole shape of it, tell me. I'd rather you own the sequencing than inherit mine.

Andy

Appendix: the evidence, the full write-ups, and every page's verdict

Everything below is the research as it was written, kept in full so you can go to the row you need. A few words you'll see: cold is what the model answers from memory with no search; live is what it answers after retrieving pages. Strong, fragile, opportunity, absent are the four states in the grid above. Situation 12 means the twelfth of the twenty buying situations. Product truth is our internal record of what ships on which plan, from the September audit. Card numbers refer to the eighteen fix cards in section B.

A. What buyers ask and what assistants answer

By business type, in buyers' words

Strong third parties recommend Circle unpromptedFragile Circle appears only when its own pages rankOpportunity the assistant already believes in Circle and lacks a page to citeAbsent not recommended, and the assistant does not believe either
By business type, in buyers' words
Business typeStatusWhy
Paid online communityStrongFirst cold and live on the category prompts. Loses the "creator with 50,000 followers" persona to Skool.
Paid mastermindOpportunityCold rank 1, live absent. Mighty owns the term with one page. Circle has none.
Live program or challengeOpportunityCold rank 2 to 3, live absent. The buyer word is "challenge"; Circle has every mechanic and no page saying so.
AI sets my business upOpportunityCategory has no name yet. Three small vendors claim "AI builds it from a prompt"; Circle shows up as an answer-bot.
Paid Discord alternativeStrongFirst live in all three cells, Circle's best result anywhere. The prior is weak and nobody frames Circle for gamers.
AcceleratorOpportunityCircle's strongest prior in the research; live results are startup-accelerator software. Nobody has written for a founder coach.
Signature program, paid clubOpportunityDead queries: no platform ranks at all, and the cold prior favors Circle. A page has no competition.
Group coaching programFragileRanks first or second only when Circle's own listicle is in the results. Kajabi owns "all-in-one."
Cohort-based courseFragileSecond on specialist pages, last on generic lists, framed as the add-on beside Teachable.
Newsletter plus paid communityFragileFirst live, but both top results are Circle's own pages. Persona prompt drops Circle.
Whole business in one placeFragile"All-in-one" and "digital business platform" belong to Kajabi. Circle is "community-first," a sub-category.
Paid membershipAbsentAbsent for "best platform to sell a membership" and for podcasters. Third in one generic cell, filed "community-focused."
Online courseAbsentNot a "sell a course" answer in anyone's head, including the assistant's. One page of five lists Circle, fourth.
Paid newsletterAbsentA clean zero in all six cells. Only the January product and a fee story move this.
Selling products: digital products, templates, ebooks, tripwiresAbsentStorefronts and funnel builders own it. The deciding feature is one-click upsell at checkout, which Circle has and nobody can find.
High-ticket, bootcamp, masterclass, certification, academyAbsentEach noun carries a strong non-Circle meaning: funnel, LMS, one-off event, credential, course platform. Disco and Teachfloor are taking these with Circle's own story.
Private podcast, Patreon alternativeAbsentReal category mismatches: private RSS feeds and tipping. "Membership site for podcasters" is the winnable adjacent page.
One-on-one coachingAbsentThe category is "coaching software," won by booking-and-payments tools. Circle enters at group.

How assistants see Circle today

AudienceGradeWhat is trueBinding constraint
RecommendersB−Circle places first or second when the buyer says "community," "membership," or "coaching community." Absent for newsletter plus community in cold answers, for YouTube-audience monetization (Skool wins), and for anything developer or API. Behind Mighty Networks on "best AI features" with a weaker product and a louder story.The pages models read are competitor-owned, and Circle's llms.txt still says "community platform."
OperatorsBFirst-party hosted MCP since April 16, 2026, OAuth, read-only or full access, copy-paste setup for Claude, Cursor, and ChatGPT. Second in the creator category only to beehiiv. Shipped before Kajabi's and Whop's; Mighty and Skool have none.Absent from the two that matter, the Claude connector directory and the official MCP registry. Present on Zapier MCP and Pipedream, which have broad reach and weak intent. Gated to Business and above; Kajabi gives it on every plan.
BuildersD+No community or membership platform is wired into Lovable, Bolt, Replit, or v0. Membership means Supabase auth plus a Stripe table. Circle's official npm SDK is dead at v0.0.4 since July 2024.A $199 a month wall before an agent can touch anything, and no create-then-claim path.

Across the first benchmark, Circle's own domain appeared on two of eight buyer questions. The domains that appeared most were group.app, schoolmaker.com, and mightynetworks.com. All three are competitors. Mighty ranks first for "Skool vs Circle" and "Circle alternatives." SchoolMaker owns "Circle pricing." GroupApp owns "Circle review." A buyer researching Circle reads Circle's competitors describing Circle.

The mechanism is affiliate economics. Skool pays 40 percent recurring. SchoolMaker pays 40 percent recurring. Circle paid $120 once until June 24, 2026, and now pays $100 per referral plus 10 to 20 percent recurring, tiered by volume. The pages models read have not caught up: they still say $120 once. So every reviewer had a reason to rank Skool first and write "Skool alternatives" rather than "Circle alternatives," the gap is now smaller than they think, and nobody has told them.

The facts assistants repeat about Circle that are wrong: a $49 Basic tier and a $419 or $999 Enterprise tier, none of which exist; "hidden" transaction fees, which are on the pricing page at 2, 1, and 0.5 percent; "no real courses," from before quizzes shipped; "no email marketing," which is two years stale. Circle AI, credits, the MCP server, and the newsletter are absent from cold knowledge entirely. Five or more independent pages repeat the same criticisms, fees that never reach zero, Email Hub at $99, no certificates, a 20-person live room cap, automations needing Zapier, and repetition makes an answer engine treat them as settled. Some complaints are not stale: billing after cancellation, unannounced price increases, slow support. Content can't fix those. Support and billing can.

Circle.so has no Wikipedia article. The USDC stablecoin company and Kajabi both do, and ChatGPT takes half its top citations from Wikipedia. On Product Hunt the "circle" slug belongs to the crypto company. GetApp files Circle partly under HR software. Trustpilot sits at 3.9, and much of the one-star volume appears to come from confused end members of customer communities rather than paying customers.

What is already good: the developer surface. An admin API with a full OpenAPI spec, headless member and auth APIs, SDKs in four languages, an llms.txt on both the marketing site and the docs, and a markdown twin of every docs page. Circle built the agent-facing product and then didn't tell anyone where agents look.

Business type by business type, in buyers' words

Twenty-eight terms, three short prompts each, answered from training knowledge with no search and then against live web retrieval. Cold rank is what an assistant believes. Live rank is what the pages it retrieves say. Sorted by status, opportunities first.

Business type or termStatusCold rankLive rankWho wins liveThe word the market usesWhat is missing
Paid mastermindOpportunity1absentMighty Networks, with a dedicated "mastermind group platform" pagenone liveThe sharpest gap. The assistant's own prior already favors Circle; it only lacks a citable page. Circle has no page for the term.
Live program or challengeOpportunity2 to 3absentMighty Networks, Exercise.com, Practice Better, Heightsnone; the buyer word is "challenge," not "community"Not a recognized software category. Mighty gets credited with challenges, leaderboards, daily check-ins. Circle has every mechanic and no page saying so.
AI sets my business up for meOpportunity32 to 5, then absentHeights, Mighty Co-Host, Membership.io"community platform with an AI agent that answers member questions"The pre-Eclipse story. On the prompt closest to the Eclipse pitch, three smaller vendors say "AI builds your membership from a prompt" and Circle is absent. The category has no name yet; the phrase is unclaimed.
AcceleratorOpportunity1 to 2absentTeachfloor, AcceleratorApp, Visiblestartup-accelerator management software, Amazon seller programsCircle's strongest prior in the whole research. Nobody has written for a founder coach running a twelve-week program.
Signature programOpportunity2 to 5absentnobody; e-signature softwarea dead queryNo platform ranks at all, so the cold prior is what buyers get. A page has no competition.
Paid clubOpportunity1absentJoin It, Raklet, Wild Apricotdues-collection software"Run a paid club online" is a cold number one for Circle and no community platform ranks for it.
Paid online communityStrong1 to 21, then absentSkool for the creator persona; Mighty, Podia, Memberful"community platform," grouped with Skool and Mighty; Kajabi called "digital business platform"Wins the category prompts, loses the persona. "Creator with 50,000 followers" belongs to Skool. Third parties quote $18 a month and "0% fees," both wrong.
Paid Discord alternativeStrong3 to 5, or absent1 in all threeMighty on the top slot; GroupApp with "0% fees""Discord alternative," monetization, courses, events, branding in one placeCircle's best live result anywhere. The prior is weak: for a gaming creator the assistant's instinct is "keep Discord, gate it with Whop." Nobody frames Circle for gamers; Guilded takes that slot.
Group coaching programFragile1 to 21 to 2, or absentKajabi, framed "all-in-one""community engagement, group programs, client retention"Every Circle sentence came from Circle's own blog listicle. When that page is not in the results, Circle vanishes. Retention words, not selling words.
Cohort-based courseFragile22 on specialist pages, 7 of 7 on generic listsMaven, Disco, Kajabi"community platform," "pick Circle when community is the product"Framed as the add-on beside Teachable, not the seat. Competitor pages say Circle lacks native live video and has a "foundational" course builder.
Newsletter plus paid communityFragile1 to 31, then absentSubstack, beehiiv, Ghost, KitCircle's pages say "all-in-one"; third parties say "community platform" bolt-onFirst place is Circle quoting itself. The finance-writer persona drops Circle entirely. Circle's pages promise newsletter sending before it ships.
Whole business in one placeFragile22 to 4, or absentKajabi, "replaces five to seven tools""community-first," a sub-category of all-in-oneCircle's blog holds the top-ranked page and Circle is still not in the recommended set, because the post is a neutral comparison. Mighty gets more coaching verbs than Circle does.
Paid group, VIP groupFragile3 to 51, then absentTelegram bots, direct-sales Facebook groups"paid group" buyers have not decided to leave the chat appFirst on the plain prompt via membership listicles and Circle's own VIP-tag help content; gone the moment the phrase becomes "VIP group" or "Telegram alerts."
WorkshopsFragile3 to 5, or absentabsentTalkspresso, ClickMeeting, FreshLearn"run and sell": live video, registration, payment, replayCircle enters cold only when the assistant assumes recurrence. The unclaimed slot is "recurring workshops plus the group between sessions."
One-on-one coachingAbsentabsentabsentPaperbell, Satori, CoachVantage, Practicenone; the category is "coaching software," defined by booking, packages, contractsLow fit. An upstream stage; Circle enters at group.
Online courseAbsentabsent4 on one page of fiveTeachable, Thinkific, Kajabi, Podia"community" or "course-driven membership," for people who "already have an audience"Circle is not a "sell a course" answer in anyone's head, including the assistant's.
Paid membershipAbsent3 to 53 in one cell of threeKartra, Podia, Payhip; for podcasters Supercast, Captivate, Patreon"community-focused," "monetizes a community"Absent for "best platform to sell a membership" and for the podcaster, where private RSS feeds decide. Mighty has claimed "community-first membership."
Paid newsletterAbsentabsentabsentSubstack, Kit, beehiiv, Ghost, MailerLitenoneA clean zero in all six cells. Tables score on transaction fee and free-tier cap. January's launch needs a fee story and a cap story to enter them.
High-ticket offerAbsent4, or absentabsentKajabi, GoHighLevel, ClickFunnels; Teachable and Mighty via their own guides"funnel," "coaching platform"Circle framed as post-sale delivery, never the seller.
BootcampAbsent4 to 5absentTeachfloor, Kajabi with a $220K case study"LMS," "course platform"Teachfloor lists cohorts, community, peer review, and payments in one sentence under the buyer's noun.
MasterclassAbsentabsentabsentTalkspresso, FreshLearn, ClickMeetinga one-off live event or a single courseMighty is absent too; a category-shape problem, not a Circle ranking problem.
Certification programAbsentabsentabsentDisco, "AI-native, purpose-built for training businesses"an LMS purchase: assessment, credentials, SCORMDisco is using Circle's positioning language on this query. The assistant is not confident Circle issues certificates.
Academy, schoolAbsentabsentabsentKajabi, Thinkific, Podia; lesson-booking softwarea course platform; Thinkific claims "academy" on its own pageStrong non-Circle default meanings. A noun-only page will not win these.
Patreon alternativeAbsent3, or absent3, or absentKo-fi, Buy Me a Coffee, Fourthwall, Gumroad, Podiatipping, membership, shopCircle's own blog ranks top four and still earns no recommendation for the illustrator. A real category mismatch, not stale knowledge.
Private podcastAbsentabsentabsentBeamly, Supercast, Castos, Captivate, Transistorprivate RSS feedsWithout a native private feed Circle cannot compete for the phrase. "Membership site for podcasters" is the winnable adjacent page.
Digital products, downloadsAbsentabsentabsentGumroad, Payhip, Sellfy, Lemon Squeezy; Stan and Beacons for TikTokstorefronts and link-in-bio storesSamCart calls itself a "digital business platform," the phrase Circle wants.
Templates, printables, ebooksAbsentabsentabsentEtsy, Creative Market, Gumroad, KDP"marketplace"Not Circle's fight.
Low-ticket offer, tripwireAbsent54 in one cellClickFunnels, Systeme.io, ThriveCart; Kajabi and Podia with checkout upsells"low-ticket funnel," "value ladder"The deciding feature is one-click upsell at checkout, never attributed to Circle. Circle has post-checkout upsells since early 2026 and nobody can find them.

Circle wins only when the buyer already says community. For every other word, Circle is absent, filed under "community-first," or present only because its own blog ranked and got quoted back. Outside community, no third party recommends Circle on its own for any of the twenty-eight.

The word the market attaches to Circle is community platform. The word it attaches to Kajabi is digital business platform and all-in-one. SamCart calls itself a digital business platform too. Disco calls itself AI-native and purpose-built for training businesses. Teachfloor, which never came up in the first pass, ranks on accelerator and bootcamp by telling Circle's cohort-plus-community story under the buyer's nouns.

The cheap flips are the blue rows: the assistant already believes in Circle and finds nothing to cite. Accelerator is Circle's strongest prior in the research. Signature program and paid club are dead queries where no platform ranks at all. Discord alternative is Circle's best live result anywhere, first in all three cells, and nobody frames Circle for gamers. The red rows split in two: real category mismatches like private podcast feeds and tipping, and the selling world of digital products and funnels, where the deciding feature is one-click upsell at checkout, which Circle has had since January and nobody can find.

Twenty real situations, asked two ways

Buyers don't type "best membership platform." They describe a situation with four requirements in it. Twenty situations in that register, each asked unaided and then with Circle named, cold and live. The named answers were scored requirement by requirement against the product-truth file from the September site audit. Situations one and two are the two prompts I asked first.

SituationUnaided, coldUnaided, liveNamed: capabilitiesWhat the assistant got wrong or missed
1. Coach scaling 15 one-on-one clients to weekly group calls, paid monthly14rightCold pricing is the 2024 four-tier ladder with 4 percent fees; live pages add a phantom $419 or $499 Enterprise tier.
2. YouTuber running a webinar that sells a course3absentunderstatedPublic RSVP, public webinars, and AI-built sales pages with checkout appear in neither training nor any page on the web, so registration and checkout were hedged. Cold suggested pairing Circle with Kajabi.
3. Podcaster, 20,000 listeners, paid tier with bonus episodes and a private group2absentrightPrivate podcast feed is a real gap, and the answer said so. Circle AI surfaced only because circle.so/platform/podcasters is citable.
4. Teachable seller adding community and office hours, one login13partlySSO plan gating wrong cold. Circle's own Teachable SSO help pages still rank though that app closed to new creators in July 2024. Teachable results quote Circle at $39 a month.
5. Substack writer adding a paid community, keeps the newsletter1absentpartlyCold misnames Email Hub "Marketing Hub" and assumes it reaches members only, so it hands the newsletter back to Substack.
6. Consultant's $5,000 mastermind with an application13rightNo page anywhere presents Circle Forms as application intake, so both modes sent the buyer to Typeform plus Zapier. Content gap, not product gap.
7. Fitness coach's 30-day challenge, leaderboard, $49 entry, then membership22rightCleanest situation. The help center's thirty-day leaderboard window matches the use case exactly.
8. Six-week live cohort, 40 people, assignments, certificate, $80015partlyProduct truth says certificates exist; every 2026 review says they don't and no help article contradicts them. Needs a check on Circle's side.
9. $3,000 program, application, three installments, onboarding, cohort community23rightSame application-form gap as 6. Stale "Enterprise $360" cold.
10. Instagram creator selling a template pack, a course, and a community, one checkout32rightAll requirements correct both modes. No third-party page mentions the AI website builder as a storefront.
11. Author's paid book club2absentrightLive results route to Substack, MemberSpace, and book-club tools. No Circle page ranks.
12. Spanish teacher, students in ten countries paying in their own currency3absentwrongCold said one currency per community, the requirement that decides the deal. Marketing pages say 25 local currencies; the help center can't be read by a crawler; product truth marks the count uncertain.
13. Yoga studio: live classes, on-demand library, in-person tickets, own brand3absentrightLive results route to Momence and Momoyoga. White-label plan gating inconsistent between the pricing page and product truth.
14. Agency client portal with onboarding course and monthly retainer2absentpartlyCold sent invoicing outside Circle; Stripe invoices exist. Live results are agency-portal tools.
15. Church app with small groups, sermons, and giving5absentrightGiving and pay-what-you-want are a real gap, so both modes went to Tithe.ly. Branded-app pricing is not public.
16. Music producer: one-time sample packs plus a feedback membership34partlyNo retrieved third-party page describes one-time digital downloads on Circle. Stale "Enterprise $360" cold.
17. Career coach: template, cohort upsell at checkout, order bump, sales page33wrongCold denied post-checkout upsells and called landing pages thin; a 2026 article says Circle lacks landing pages, checkout, upsells, order bumps, and affiliate tracking, wrong on four of five. Order bumps are publicly undocumented.
18. Leaving Kajabi at $199, dead community, clunky email12partlyCold said courses must be rebuilt by hand and members re-enter cards; the migration service moves both. Two of three "Kajabi alternatives" lists omit Circle. Email Hub add-on breaks "same money as Kajabi."
19. Beginner, 500 followers, wants a free plan55right"No free plan, 14-day trial, starts at $89," so Circle is "graduate to it later." The pricing page shows no monthly-billing figure, so third parties fill in $99 or $129.
20. 40-person SaaS wanting a customer community with a knowledge base and SSO42wrongCold gated all SSO to Enterprise and denied any AI support agent; Auth0 SSO is on Professional and Business, AI agents and the knowledge engine on Plus. No Circle page names a knowledge base.

When the buyer names Circle, the capability answers are right or nearly right in sixteen of twenty. Unaided, Circle is first or second from memory in nine of twenty and absent live in nine of twenty, because the pages that rank for a yoga studio, a church, a language teacher, or an agency are vertical tools, and the pages that rank for a webinar or a podcast tier are category roundups Circle isn't in.

The wrong facts are the same every time. Cold knowledge carries the 2024 plan ladder with 4 percent fees and a retired Enterprise tier. Live pages add phantom $419 and $499 tiers and a $39 Basic. What shipped in 2026 is unfindable rather than unwritten. Public RSVP, public webinars, AI-built sales pages, post-checkout upsells, Forms as application intake, the migration service, SSO by plan, AI agents on Plus all exist as one-paragraph entries on circle.so/product-updates, which is readable and well built. But it is a changelog ordered by date, no buyer question retrieves a changelog, the feature pages that do get retrieved don't carry the facts, the community posts holding the detail render through JavaScript so a crawler or fetch tool gets the title only (a browser-driving agent does see the full post; the crawlers and fetch tools that write answers do not), and the help center serves every crawler the audit tried a Cloudflare challenge page with nothing behind it. Where a readable page exists, the answer corrects itself on the spot. The pricing page fixed every price. The migrations page fixed the whole "rebuild your courses by hand" story. The podcasters page was the only place Circle AI ever surfaced.

Real gaps for product rather than content: giving and pay-what-you-want, private podcast feeds, a knowledge base or a page saying the knowledge engine is one, and a free plan, which is January. Three things to verify on Circle's side: whether certificates exist, which plan white-label sits on, and how many currencies one community can price in.

What buyers call it

Two passes, no customer data. The open web: search completions and questions, first-person posts on Quora, Indie Hackers, Medium, Substack, Skool, and Gumroad, the exact labels on eighteen competitor sites, G2 and Capterra categories, and the creator-economy reports. And our own files: the September business-types study and the site crawl. Reddit blocks the crawler and is a gap.

Kit's 2026 creator survey lists what creators sell: coaching and consulting 31.6 percent, courses and digital products 23.1, newsletters 16.4, e-commerce 7.5, podcasts and video 5.5, communities 5.1. The category Circle is named for is five percent of what creators sell.

"Community platform" appears in Skool's headline, Mighty's headline, and the G2 category name. Buyers say paid community, paid group, or membership site. In our own customer stories nobody calls their thing a community. They name it: Part-Time YouTuber Academy, The Lab, London Writers' Salon, Entreprenista League, PMM Camp, Jay Shetty Certification School, Notion Mastery. The buyer's noun is the container brand. The category word is ours.

My list held. Mastermind, cohort, challenge, accelerator, bootcamp, paid group, inner circle, private podcast, certification, signature program, high-ticket, digital products, Patreon alternative, paid Discord, membership site, academy, school, and club each have two or more independent buyer sources. Book club, peer group, alumni network, and collective are weak or institutional. Four terms are used at medium frequency and claimed by none of the eighteen competitors: signature program, inner circle, high-ticket, masterclass. Same for the parenting, language, and trades niches.

RankTermWho already claims it on their siteBenchmarked
1digital products, digital downloadsStan, Teachable, Podia, beehiiv, Patreon, Kit, Whopwave 2: absent
2mastermindMighty (a dedicated page), Podia; Circle has a blog postwave 1: opportunity
3cohort, cohort-based courseMaven, Heartbeat; Circle says "live cohorts"wave 1: fragile
4paid Discord serverWhop, Memberful, LaunchPasswave 2: strong
5challenge, 30-day or 7-dayMighty, Everfitwave 1: opportunity
6group programPodia, Kajabi, Teachable; Circle says "group program"wave 1: fragile
7high-ticket offer or programnobody in navigation; Skool hosts the nameswave 2: absent
8evergreenKajabimodality
9self-pacedCirclemodality
10private podcast, bonus episodesPatreon, Substack, Memberful, beehiiv, Kajabiwave 2: absent
11templates, presets, printables, plannersPodia, Teachable, Stanwave 2: absent
12paid subscriptionsSubstack, beehiiv, Kit, Memberfulprice shape
13tiers, paywalled tiersPatreon, Memberful; Circle says "paywalled tiers"price shape
14Patreon alternativeFanCircles; Circle has a compare pagewave 2: absent
15workshopMaven, Mighty, Kitwave 2: fragile
161:1 sessionsStan, Podia, Teachable, Circlewave 1: absent
17academyThinkificwave 2: absent
18masterclassnobodywave 2: absent
19bootcampMaven, Trainerizewave 2: absent
20paid group, VIP group, private groupWhop, GoHighLevelwave 2: fragile
21tripwire, low-ticket offerStan, blog onlywave 2: absent
22signature programnobodywave 2: opportunity
23acceleratorMaven, in course titleswave 2: opportunity
24certificationMaven, GroupAppwave 2: absent
25inner circlenobodynot run

The prize

No internal data was available and no study exists on whether coaches or creators consult AI when choosing a platform. This is external evidence applied to Circle's known volumes, with the assumptions visible.

InputValueSource
B2B SaaS share of sessions from AI2.8% to 11.8%, depending on whether AI Overviews and dark traffic countConductor; First Page Sage, Aug 2026
Buyers who start software research with an AI chatbot51%, up from 29% a year earlierG2 buyer survey, 1,076 buyers, Mar 2026
Bought from a vendor they had never heard of, on AI guidance33%same
Conversion multiple of AI-referred visitors vs organic2.5x is the defensible number; 23x (Ahrefs) is the ceiling and its author doubts it scalesFirst Page Sage; Ahrefs, Jun 2025
B2B SaaS organic sessions, year over year−29.4%First Page Sage
Share of ChatGPT referrals landing on the homepage62 to 63%, up from 26 to 29% after the May 2026 changeFirst Page Sage
ScenarioAI share of sessionsConversion multipleAI-sourced trials a month, on 4,000
Bear3%1.5xabout 180
Base4%2.5x360 to 400
Bull8%3xabout 800

The contrarian track puts today's number at 80 to 120 trials a month, which is a test, not a pillar. Both reads agree on three things. Circle can't measure this channel today, because AI referrals hide inside direct traffic. The homepage is now the AI landing page, which ties this to the site refresh. And the retail conversion statistics everyone quotes don't cover software. The number nobody should ignore is different from the traffic number: one in three B2B buyers now purchases from a vendor they had never heard of, on an AI recommendation. That bypasses brand awareness, where Circle is weakest against incumbents. And the downside is bigger than the upside. B2B SaaS organic sessions are down 29 percent year over year as AI answers absorb clicks. A 25 percent organic decline erases the whole base-case gain. Part of this work is defense.

Who does this well

My two hypotheses (Stripe grows because agents use it; Whop's CLI pulls builders) check out, each with a correction.

Six mechanics recur, in order of leverage. Remove the human from the critical path: provision first, claim later. Ship two MCP servers, docs and operations. Make docs machine-addressable, with markdown at every URL. Default-on beats opt-in. Be the default inside someone else's builder. Price so the agent can say yes, with no card and an instant key.

What ports to Circle: the docs-versus-operations split, markdown at every URL extended to the marketing and help surfaces, a create-then-claim flow where an agent stands up a community seeded with spaces, a course skeleton, and paywall tiers before the founder signs up, first-party plugins so installation is one command, and confirmation gates on destructive writes, which matter more for Circle than for Stripe because an agent misfiring on a live member list is unrecoverable. What doesn't port: protocol authorship, which needs transaction volume the labs want, and agent checkout, which Instant Checkout's failure buried. Three gaps nobody in creator platforms has closed: agent-created accounts, discoverability inside assistants by default, and un-gated agent access. Kajabi put MCP on every plan. Circle gates it to Business and above.

Two uses for the MCP server today, independent of any growth bet. Agents already operate Circle through the browser, via Claude in Chrome and Perplexity's Comet, whether Circle lists anywhere or not; listing decides whether Circle controls how well that works and whether it can see any of it. And the server is a sales-credibility artifact for Plus deals right now.

B. The five changes, and the eighteen site fixes, written up in full

The five changes the research ranked highest, as written on Sep 14

Each one is written the same way: what is wrong and how we know, what it is costing, what to do, what you get, what it takes, and a no-brainer score. The score is a judgment call on effort against reward, from 1, don't do it without more evidence, to 5, obvious, do it this week. The product bets and the commercial decisions are not on this list. They are bigger levers with a worse ratio and they are in the plan.

1. List the MCP server where agents look

What's wrong
Circle shipped a hosted MCP server in April 2026 with OAuth and good docs, and it is listed in neither the Claude connector directory nor the official MCP registry. Circle's own setup docs tell users to add it as a "custom connector," which is what you do when you are not listed. It is on Zapier and Pipedream, which have reach but no intent.
What it costs
Directory listing is what makes a connector eligible for in-chat suggestions to any Claude user whose task matches. Circle is opted out by omission. The directory is also the only agent surface that reports usage back, so today Circle has built the capability and can measure none of its use. And when a buyer asks an assistant "best API or MCP for a community business," zero community platforms come back, so Circle is not losing that question, it is not considered.
What to do
Submit to the Claude connector directory: a Claude Team org, a title and read-only or destructive hint on every tool, a privacy policy, test credentials. Publish to the official MCP registry with DNS proof of circle.so. Confirm the server is on the July 2026 protocol version, which was a breaking change.
What you get
Presence at the moment of intent inside Claude, the first usage numbers for the operator lane, and a credibility artifact for Plus deals from day one. The pipeline upside is unknown, since nobody has published signups from a listing. The downside is zero.
Effort, owner
Two forms and about a day of engineering. Product and engineering.
No-brainer
5 of 5. A day of work, no downside, and it is the only way to get usage numbers for the operator lane.

2. Write one page for each term where the assistant already believes in Circle

What's wrong
For accelerator, paid mastermind, 30-day challenge, signature program, paid club, and Discord alternative, the assistant ranks Circle first or second from memory and then finds nothing to cite when it searches, so Circle drops out of the live answer. Mighty owns mastermind with one dedicated page. Signature program and paid club are dead queries where no platform ranks at all.
What it costs
These are buyers who would have been told Circle if a page existed. Mastermind and challenge are among the five terms buyers use most. Today the answer goes to Mighty, Teachfloor, or nobody.
What to do
One page per term, modeled on Mighty's mastermind page: the buyer's word in the title, the mechanics named in the buyer's words (live rooms, leaderboards, daily check-ins, applications, paywalls), plan gating, a real customer with a number. Then the same for group coaching and cohort, where Circle appears only when its own listicle ranks.
What you get
These are the cells most likely to flip within weeks, because the model's prior is already on Circle's side and live retrieval changes on the next query. Expect the blue rows in the glance grid to turn green one by one, and the monthly panel to show it.
Effort, owner
Days per page. JK.
No-brainer
5 of 5 for the six terms where the prior already favors Circle. 3 of 5 for group coaching and cohort, where Kajabi and Maven hold the page and a new page competes rather than fills a vacuum.

3. Fix the facts machines repeat

What's wrong
circle.so/llms.txt, the file written for machines, says "all-in-one community platform" and "community-first, not course-first or marketing-first." Capterra and GetApp show "from $49," and GetApp files Circle under HR software. G2 has Circle in the community category and not in coaching, courses, membership, or newsletter. The pricing page shows no monthly-billing price. Circle's pages promise newsletter sending before the January launch. And the 2026 capabilities sit on the changelog and nowhere else.
What it costs
Every wrong claim in forty named answers traces to one of these. Assistants quote $49, $419, and $999 tiers that don't exist. G2's categories are the taxonomy models use to define a market, so being absent from coaching and courses means being invisible to those questions. And the machine-readable description teaches every model that reads it the exact category Circle is trying to leave.
What to do
Rewrite llms.txt to the digital business frame, one paragraph. Correct the review-site profiles and add the four G2 categories. Publish one crawlable pricing-and-fees page with monthly prices and the total-cost math at several revenue levels. Take the newsletter promise down until January. Put each 2026 capability on the feature page it belongs to with plan gating: upsells and order bumps on payments, webinar registration on events, application forms on forms, the migration service on the compare pages.
What you get
In the benchmark, one readable page fixed the switching story and every price on the spot. This is the same effect applied to every wrong fact at once, and it starts working on the next query.
Effort, owner
Hours each. PMM.
No-brainer
5 of 5. Hours of work against forty wrong answers, and every item is a correction of a fact, not a bet on a message.

4. Measure it

What's wrong
Circle can't see this channel. AI referrals land in analytics as direct traffic, "how did you hear about us" has no AI option, there is no recurring measure of what assistants recommend, and nobody has pulled how many customers have connected the MCP server.
What it costs
Every number on this page is someone else's panel. Without a baseline the flip and kill conditions can't fire, and the program gets defended with proxy metrics, which is how hyped channels outlive their usefulness.
What to do
Tag chatgpt.com, perplexity.ai, claude.ai, gemini, and copilot referrers in analytics and HubSpot. Add an AI option to the attribution question. Run a thirty-prompt panel monthly, short prompts in buyers' words, scored for presence, rank, framing, and who is cited. Pull MCP connections and tool calls by plan from Circle's logs.
What you get
A baseline within a month, then a monthly read on whether changes two, three, and five moved anything. If the first month already shows base-case numbers, the question changes from whether to invest to why we weren't measuring.
Effort, owner
Hours. RevOps, with Quinn running the panel by hand for nothing.
No-brainer
5 of 5. Nothing else on this page can be judged without it, and it costs a few hours.

5. Get Circle into its own answers

What's wrong
Circle's blog holds the top-ranked page for "entire online business" and for "best coaching platforms," and the assistant still does not recommend Circle from them, because the posts are neutral comparisons written in retention words like engagement and community. Meanwhile every public community post, including the release notes, renders through JavaScript, so an agent gets a title and nothing else, and the help center is worse: it serves a Cloudflare challenge page to every crawler the audit tried, GPTBot and Google-Extended included, and returns 403 on repeat requests. It has no sitemap either.
What it costs
Owning the ranking slot without owning the recommendation is the most common failure in the benchmark. It means Circle did the hard part, ranking, and gives the answer away. And the help center being invisible to crawlers means every "can Circle do X" question is answered from training memory, which for Circle facts is stale to 2024.
What to do
Rewrite the handful of Circle pages that already rank so Circle is in the recommended set with selling words, and so "all-in-one" and "digital business platform" are attached to Circle on circle.so. Take the help center out from behind the Cloudflare challenge for known crawlers, add a sitemap, and serve articles as plain HTML with the body in the page. Do the same for the community's public spaces.
What you get
The pages already rank, so the change reaches assistants on the next retrieval. The help center fix corrects hundreds of capability answers at once and makes the help articles indexable for the first time.
Effort, owner
Days per page inside the site refresh; days of web engineering for the help center. JK, Rudy on the homepage, web engineering.
No-brainer
4 of 5. The page rewrites are a 5, since the pages already rank and the copy is ours to change. The help-center rendering is real engineering, which is the one point off.

Ranked lower on ratio: MCP read-only on every plan (a decision with an unsized cohort), the CLI as create-then-claim (a product bet, months), a Wikipedia entry (big if it lands, slow, not in Circle's control), closing the rest of the affiliate gap to Skool (Circle already moved to recurring in June; the rest is a margin decision). The page-by-page audit of the whole site, 215 pages read as a crawler reads them, is at Circle site audit for agents: eighteen fixes in this same format with scores, the product rulings needed, and every page's verdict.

Each card: what's wrong and how we know, what it's costing, what to do, what we get, effort and owner, and a one-to-five score. Effort words are the reviewers': quick is hours, medium is days, long is weeks.

This week: hours each, all no-brainers

Seven fixes. Each is a single component, file, tag or ticket, and each removes a wrong fact from the pages assistants already read. The cards draw on the five review tables below and, where it says so, on the benchmarks on the main page. Effort words are the reviewers': quick is hours, medium is days, long is weeks.

1. Fix the shared pricing block, and the pricing page

What's wrong
One component embedded on fifteen pages, including /pricing and every /platform page, lists "Content co-pilot" as a Business feature, shows plan availability as icons a crawler cannot read, puts "$49/month" beside the profile-fields add-on where it reads as a $49 tier, and shows no monthly prices. The page has no product or offer markup.
What it costs
This one block is most of the stale-price problem. Assistants quote a $49 Basic tier because a crawler sees "$49/month" on the pricing page with no label. Third parties fill in monthly prices of $99 or $129, which the twenty-situation benchmark found, because Circle publishes none. And a model reading the plan table cannot tell which plan carries which feature.
What to do
Render plan availability as text. Add monthly prices next to annual. Rename co-pilot to Circle AI. Label add-ons as add-ons. Add a fee-math example at three revenue levels. Add SoftwareApplication markup with an Offer per plan. Publish a markdown twin of the page.
What you get
In the benchmark the pricing page already corrected every price the moment it was read. This makes it correct on the first read and removes the unlabeled $49 that started the $49 myth, on fifteen pages at once.
Effort, owner
Hours. Web engineering and PMM.
No-brainer
5 of 5. One component, fifteen pages, and the wrong fact the benchmark repeated most.

2. Rewrite llms.txt and regenerate llms-full.txt

What's wrong
The file written for machines opens "Circle is the all-in-one community platform" and says "Circle is community-first, not course-first or marketing-first." It quotes 18,000 communities where the site says 20,000, links a retired AI Copilot page, describes Plus as branded apps, omits fees, MCP, product updates, six customer stories and every /platform page. The full-site file holds 122 blog posts and 19 stories and no pricing, product, compare or platform page.
What it costs
Every model that reads Circle's own machine-readable description learns that Circle is a community platform, in Circle's own words, with wrong numbers. The full-site file gives a model the blog and nothing about the product.
What to do
Rewrite the header as the digital business definition with the business types named, prices and fees. Add sections for 2026 capabilities with plan gating, MCP, Discover, Studios, the platform pages by business type, all compare pages with the Facebook URL corrected, all 25 stories, the twelve current AI posts, and product updates. Drop the two 2023 co-pilot posts. Regenerate the full file from product, pricing, compare and platform pages first, blog last.
What you get
Circle's own machine-readable description says digital business platform, with the right numbers, and links the pricing, product, compare and platform pages that back it up.
Effort, owner
Hours. JK for the words, PMM to check the facts, web engineering for the full file. The core review assigns it to JK and the structure review to PMM.
No-brainer
5 of 5. An hour of writing to replace the file that calls Circle a community platform in Circle's own words.

3. Fix the MCP page's canonical tag and list it

What's wrong
circle.so/mcp carries a canonical tag pointing to circle.so/chat, which tells search engines the MCP page is a duplicate of the chat feature page. It is not in the sitemap and not in llms.txt. The page itself is good.
What it costs
Search engines and llms.txt both miss the MCP page, the operator proof point the research ranks first, because of one line of markup.
What to do
Set the canonical to /mcp. Add it to the sitemap and llms.txt. Link api.circle.so/mcp. Add "since April 2026" and "read-only or full access" to the page.
What you get
The MCP page can rank and be cited, which is a precondition for change one on the main page.
Effort, owner
Minutes for the tag; an hour with the page edits. Web engineering, product for the launch-date and access lines.
No-brainer
5 of 5. A one-line bug hiding a page Circle needs search engines and assistants to find.

4. Put Circle's price and a fee table on the three most-searched compare pages

What's wrong
vs-skool, vs-mighty and vs-kajabi carry no Circle price. The Skool page's fee math is Skool-only. The Kajabi page never claims all-in-one or digital business platform and leaves Kajabi's zero fee unanswered. The Bettermode page already has the right table.
What it costs
This is part of why SchoolMaker owns the query "Circle pricing" and why Mighty ranks first for "Skool vs Circle." The pages that should answer "what does Circle cost" carry no Circle price for an assistant to quote.
What to do
Port the Bettermode price table to the three pages. Add a three-level fee table ($5K, $20K, $40K monthly sales) against each competitor's fee and plan price. Add FAQPage markup. Rewrite the Kajabi page as the digital business page: $89 plus 2 percent against $143 to $179 plus 0 at three levels, with upsells, AI sales pages, webinars, Forms, affiliates and MCP named, and an honest line that Circle has no funnels feature.
What you get
The three matchups buyers search most resolve on circle.so with a number an assistant can quote.
Effort, owner
Hours for the tables. The Kajabi rewrite is the one medium item in the compare group, two or three days. JK and PMM.
No-brainer
5 of 5. The table exists on another page; this is copying it to where the searches are.

5. Correct the wrong Circle facts on pages that rank

What's wrong
Quoted verbatim in the reviews: the Kajabi alternatives table says "Transaction Fee 0.5–4%" and "AI Agents (Enterprise)"; the Patreon alternatives post computes "$140–$290" at $5K a month where the real figure is $189 to $249; the Facebook Groups post says Circle AI is "read-only" with GA on "July 31, 2026"; the newsletter listicle says plans are "flat regardless of list size" when Email Hub is an add-on; the AI-native post still says "Join the waitlist for Circle AI"; the AI agents page says agents are included "in 2025"; the Mighty alternatives post says API on all plans; the Facebook compare page claims order bumps, which product truth says not to claim; five pages disagree on whether certificates exist; competitor prices differ across Circle's own posts.
What it costs
Assistants repeated these exact stale facts in the benchmark, and they learned them from Circle's own pages. A wrong price on a ranking page is what third parties copy next.
What to do
Fix each quoted line. Get a product ruling on certificates and order bumps and make every page agree. Pick one figure for each competitor price and use it everywhere.
What you get
Circle's own pages stop supplying the wrong facts that assistants and third parties repeat.
Effort, owner
Hours per page, a morning in total. JK and PMM; product for the two rulings.
No-brainer
5 of 5. Factual corrections on pages that already rank, with no messaging decision needed.

6. Retire, redirect and clean up in one ticket

What's wrong
Live embarrassments: lorem ipsum on /academy; /demo-video is a Portuguese sales form under an English "Book a Demo" title; /migration/courses publishes an internal note, "This one might be worth discussing internally"; /v3 is a 2024 launch page telling people to revert by November 30, 2024; /affiliate-faqs is empty; /2021-review and /2022-review are stubs; two 2023 co-pilot-era posts are in llms.txt, one of which says the feature is "available exclusively to customers on our Business and Enterprise plans"; two customer stories carry another founder's quote by copy-paste; one carries another's stat block; the event hub is nineteen 2021 to 2023 transcripts, about 190,000 words, that say community up to 211 times a page; and every URL an agent guesses (/customers, /compare/kajabi, /help, /fees, /about, /changelog) returns 404.
What it costs
Stale pages are what models quote. The event hub is the largest block of text on the site after the blog, and all of it is 2021 to 2023 community-platform talk. Guessed URLs matter because agents try them before they search.
What to do
Redirect /v3, /demo-video, /affiliate-faqs, the two review stubs and the two 2023 posts. Fix the placeholder text and the internal note. Correct the three story errors. Noindex the event-hub transcripts or cut each to a dated summary. Add 301s for the guessed URLs.
What you get
Crawlers stop ingesting the 2023 and 2024 pages, the placeholder text, the internal note and the copy-paste errors.
Effort, owner
A day. Web engineering and JK, PMM for the event hub.
No-brainer
5 of 5. Deletions and redirects, with almost no writing.

7. Fix the shared vertical template on the platform pages

What's wrong
The /platform business-type pages, eight to eleven of them depending on how the review counted, share a template whose FAQ markup asks "How does Circle compare to other community platforms?" and whose closing band is titled "The complete community platform". Two of them, podcasters and YouTubers, answer "Do you offer migration services" with a customer quote about launching a course.
What it costs
The pages written for each business type all teach the category word in structured data, which is the part of the page an engine reads first. The broken FAQ is on the only page where Circle AI surfaced in the benchmark.
What to do
Change the schema question and the band title in the template. Fix the two broken FAQs. Add the 2026 capability list with plan gating as a template block.
What you get
Eight business-type pages stop calling Circle a community platform, in one edit.
Effort, owner
Hours. JK and web engineering.
No-brainer
5 of 5. One template, every business-type page.

The next few weeks: days each

8. Put 2026 on the feature pages

What's wrong
Not one feature page names a 2026 capability in a buyer's word. Events never says webinar, registration, ticket or public, and its FAQ lists only the Pro plan's live limits, which feeds the 20-person cap criticism. Payments never says upsell or order bump. Website builder does not mention AI. Workflows has no sale triggers and its FAQ says Business and above. Customization says custom domains are on "Pro and Enterprise." Analytics says filters are Plus-only. Gamification never says challenge.
What it costs
Feature pages are what a buyer question retrieves. Situation 2 hedged registration and checkout, the tripwire cells never attributed one-click upsell to Circle, and the live-room cap myth persists, because the feature page the assistant retrieved never mentions any of it.
What to do
Events: public RSVP, public webinars, paid tickets, guest to CRM contact, a plan table for stream and room limits, retitle with webinar. Payments: post-checkout upsells, installments, BNPL, Apple and Google Pay, paid tickets, tiers, your own affiliate program, fee math, reconcile the migration gating with customer success. Website builder: the AI builder first, sales page in the title. Workflows: sale triggers, Circle AI writes the workflow, fix the FAQ. Email hub: the price ladder inline, forms as application intake. Analytics and customization: correct the FAQs. Gamification: run a 30-day challenge. Headless: the developer block with OpenAPI, SDKs and MCP. Chat: one line naming a paid Discord alternative.
What you get
The pages that answer "can Circle do X" say yes, with the plan, before the assistant has to guess.
Effort, owner
Hours per page in the reviews; a week or two of PMM and JK time across the ten pages. Product to confirm order bumps.
No-brainer
5 of 5 for payments, events and website builder. 4 of 5 for the rest, which are smaller corrections on lower-traffic pages.

9. Product updates: full notes on circle.so, capability-style entries, feature-page links

What's wrong
The changelog shows five of 32 releases to a crawler, links 30 of the 32 to community posts that render a title only to any fetch without JavaScript (verified Sep 15 across GPTBot, ClaudeBot, PerplexityBot, Googlebot and browser user agents, plus a live LLM fetch tool: 12 to 13 words; the same post in headless Chromium with JavaScript renders 2,685 words, so a browser-driving agent can read it and a retrieval crawler or fetch tool cannot), gives almost no release its own URL on circle.so (the Eclipse post is the exception), links to no feature page, checked directly on the live page, and is not in llms.txt.
What it costs
The only first-party record of 2026 shipping is a changelog, which no buyer question retrieves, with detail a crawler cannot read.
What to do
Render all 32. Host full notes at circle.so/product-updates/slug. Write entries per capability with the buyer's words, the plan and the date. Link feature pages to releases and back. Add dated Article markup and llms.txt entries. Then, as product work, serve post bodies in the HTML on public community spaces.
What you get
The year of releases becomes citable, and the feature pages carry the facts.
Effort, owner
A day or two for markup, rendering and links; two weeks for per-release pages and rewritten entries; months for the community rendering. PMM, web engineering, product.
No-brainer
4 of 5 overall; the first steps are a 5, the community rendering a 2 until product sizes it.

10. Turn the how-to posts into term pages, and write the missing ones

What's wrong
The mastermind post is a seven-step how-to where Circle appears at step five in retention words, sixteen selling words to ninety-three retention words. The group coaching and cohort posts are the same shape. Mighty wins mastermind with one page titled for the term. For challenge, accelerator, signature program, paid club, Discord alternative for creators, membership site for podcasters, a webinar that sells a course, and a high-ticket program, no Circle page titled for the term exists; the Discord post is a roundup and the podcasters page lacks the wording.
What it costs
These are the blue rows in the main page's glance grid: the assistant already believes in Circle and finds nothing to cite. They are the cells most likely to flip, and the only thing missing is a page to cite.
What to do
Rewrite mastermind-groups, group-coaching and cohort-based-courses on their existing URLs as platform pages: the buyer's word in the title and H1, the mechanics named, one real customer with a price, FAQ markup. Merge cohort-program into the cohort page. Write the eight missing pages the same way. Lead stories exist for most: Roofing and Solar Reform Alliance for mastermind, SPI for accelerator and membership, Whoggga for paid club, Seasoned Sales for coaching. Challenge, signature program, Discord alternative, podcasters, webinar and high-ticket need a customer found.
What you get
The blue cells are the ones most likely to move, and they move within weeks of publishing, because live retrieval changes on the next query.
Effort, owner
Days per page, JK. Product to confirm certificates before the certification page.
No-brainer
5 of 5 for the six blue terms. 3 of 5 for group coaching and cohort, where Kajabi, Maven and Teachable already hold the page, per the business-type benchmark.

11. Rewrite Circle's entry in the listicles that already rank

What's wrong
The best-coaching-platforms post is the top-ranked page for the largest buyer segment and describes Circle as "community engagement, group coaching, and member retention." The community, membership, creator, event and membership-site posts do the same. The Kajabi alternatives verdict concedes conversion: "Kajabi optimizes for conversion; Circle optimizes for retention." None of them names the 2026 selling stack.
What it costs
Every Circle sentence the assistant used in the coaching benchmark came from this page, and it still did not recommend Circle. That was the benchmark's most common failure: Circle's page ranks, the assistant reads it, and still recommends someone else.
What to do
Rewrite the Circle entry, bottom line and FAQ of each in selling verbs: sell a package with installments, take applications with Forms, run a public webinar into checkout, upsell the membership. Put Circle in the first-revenue stage of the creator post. Rewrite the Kajabi verdict so Circle wins conversion. Fix the wrong fee on the course post and state certificates under Circle once product confirms.
What you get
The pages Circle already ranks with start recommending Circle.
Effort, owner
A day per post, JK.
No-brainer
5 of 5. These pages already rank. Only the copy is wrong.

12. Add a "what Circle sells that X cannot" block and FAQ markup to every compare page

What's wrong
Every head-to-head closes on "The complete community platform" and "serious community businesses." One names public RSVP, one hints at the AI website builder, and none names Forms as applications or the MCP beyond a footer link. Every one has a long FAQ and no FAQPage markup. llms.txt lists eight of eighteen and links the Facebook page at a wrong URL.
What it costs
Compare pages are what assistants cite, and Circle's still call it a community platform and name nothing that shipped in 2026.
What to do
One paragraph per page listing post-checkout upsells, AI sales pages, public webinars with RSVP capture, Forms as applications, MCP and the migration service with plan gating. Use the beehiiv page's H1 as the group template. Add FAQPage markup and the missing llms.txt entries in one engineering pass. Then write /compare/circle-alternatives, so the query about Circle resolves on circle.so, and a first-party reviews page for "Circle review."
What you get
The compare pages, the pages assistants cite most, say what Circle sells now.
Effort, owner
Days for the block, a day for the markup pass, medium for the two new pages. JK and web engineering.
No-brainer
4 of 5. Broad and controllable; a point off because it is many pages rather than one component.

13. Make the customer stories carry the buyer's noun and a number

What's wrong
Nine of 25 story titles say community; none says coaching, membership, course, mastermind or newsletter. Only five put a revenue figure in the title or stat block. The stories index shows a crawler 209 words and links six of 25. Six stories, including the newest and most revenue-heavy, are missing from llms.txt.
What it costs
Named customers with numbers are what win citations, and the proof for the weakest business types exists and is unlabeled: Seasoned Sales at $15K MRR in two months, SPI at $700K+ membership revenue in 2024, Roofing and Solar at $12K to $15K application-only, Talk Nerdy to Me at $37K+ monthly off Kajabi.
What to do
Retitle the five lead stories with the business type and the revenue. Add the six missing stories to llms.txt. Server-render the stories index with business-type filters and a number per card. Link each story from its term page.
What you get
Every weak business type except newsletter gets a citable customer with a dollar figure; newsletter has no strong story until the January product.
Effort, owner
A morning for the edits, medium for the index. JK, PMM, web engineering, Rudy on the index.
No-brainer
5 of 5 for the edits. 4 of 5 for the index rebuild.

14. Publish the pages that should exist

What's wrong
There is no page for pricing and fees explained, for what shipped in 2026, for what Circle is in plain facts, for the selling stack, for Circle Plus inclusions, for a knowledge base, for webinars, for checkout and upsells, for forms, or for running your own affiliate program.
What it costs
Each missing page produced a wrong answer in the benchmark: no page corrects the hidden-fee myth, the entity question resolves to a fintech, digital products is the top buyer term with no page, situation 20 found no knowledge base, situation 2 found no webinar page.
What to do
Write them in that order. The fees page and the what-Circle-is page are hours; the 2026 page and the selling-stack page are days; the rest are quick.
What you get
The facts an assistant needs exist at a URL with the buyer's word in it.
Effort, owner
Hours to days. PMM and JK.
No-brainer
5 of 5 for fees, 2026, and what Circle is. 4 of 5 for the rest.

Longer: weeks to months

15. Make the help center readable

What's wrong
Every help article serves a 2.9 KB Cloudflare challenge shell, with the title only, to every crawler the review tried: ClaudeBot, GPTBot, PerplexityBot, Google-Extended and CCBot. Googlebot itself was not tested. Repeat requests get 403. The help sitemap redirects to a 404. There is no article markup.
What it costs
Every "can Circle do X" question is answered from training memory, which for current models stops in mid-2026 and for Circle facts is stale to 2024. Situations 12 and 20 in the benchmark were wrong for this reason.
What to do
Exempt known crawlers from the challenge. Server-render article bodies. Publish a sitemap. Add Article markup. As an interim, mirror the top hundred articles as HTML on circle.so.
What you get
Hundreds of capability answers correct at once, and help content indexable by any crawler for the first time.
Effort, owner
Weeks. Web engineering; the interim mirror is a medium item on its own.
No-brainer
4 of 5. High upside, real engineering. The interim mirror is a 5 on its own.

16. Public community spaces serve post bodies in the HTML

What's wrong
Circle's own public community, which runs the same renderer every customer gets, serves post bodies through JavaScript. A crawler gets a title and one sentence. community.circle.so has no robots file.
What it costs
Circle's own product updates are unreadable, and so is every customer's public content, which is the discoverability bet in Phase 4 of the main plan.
What to do
Server-render public posts, add robots.txt and Article markup. Product work.
What you get
Every customer's public community becomes discoverable to assistants by default. The competitor research found nobody in the category has this.
Effort, owner
Months. Product.
No-brainer
3 of 5 today, because it is unsized. It becomes a 5 once product scopes it, and the interim is item 9.

17. Localized pages

What's wrong
The German and French homepages still carry the community-platform headline: "Die komplette Plattform für deine Community" and "La plateforme communautaire complète". 180 localized URLs in the sitemap, mostly blog pages, serve English content under German, French and Portuguese language tags. The Brazilian pricing page has a monthly and annual toggle the English page lacks.
What it costs
A third of the sitemap either calls Circle a community platform or serves English under the wrong language tag, and the one localized page that gets something right, the Brazilian pricing page, is one few buyers open.
What to do
Port the digital business homepage copy to the three localized homepages. Noindex untranslated pages. Copy the monthly and annual toggle to /pricing.
What you get
Removes a third of the sitemap from the problem.
Effort, owner
Medium. JK and web engineering.
No-brainer
3 of 5. Cleanup with modest upside, except the pricing toggle, which is a 5 and belongs in item 1.

18. Structural rebuilds

What's wrong
The /platform hub is titled "The Complete Community Platform" with community 45 times. Three course pages split one query. /plus/saas holds the best B2B FAQ on the site and is noindexed. /plus/branded-app shows a crawler twelve "Loading..." placeholders. The integrations directory, the partner directory and the prompt library render mostly through JavaScript. The sitemap has no lastmod dates.
What it costs
Split authority on the queries Circle loses, invisible proof for the B2B and Plus stories, and no freshness signal for the 2026 pages.
What to do
Retitle and rewrite the platform hub. Merge the course pages to one. Merge /plus/saas into customer-communities or remove noindex. Server-render branded-app, integrations, partners and all 268 prompts. Add lastmod to the sitemap.
What you get
One canonical page per query, and proof pages a crawler can read.
Effort, owner
Medium. JK and web engineering.
No-brainer
4 of 5. Each is small; together they are a couple of weeks.

C. Every page's verdict

Page by page

The five review tables, one row per page: whether a crawler can read it, the verdict, what was found with stale text quoted, the recommendation, why, effort and owner. Verdicts and effort are colored: green keep and quick, amber fix and medium, blue rewrite and merge, red retire and long. Each group ends with the pages that should exist and that reviewer's top five.

Core commercial pages: home, pricing, platform, AI, MCP, Plus, migration

Every page was read as a crawler sees it, without JavaScript. Quotes are verbatim. Three defects repeat across the group and are referenced below as T1, T2, T3.

  • T1, shared pricing block. Fifteen pages embed the same plan cards. The Business card lists "Content co-pilot" (the pre-2026 name). Per-plan checkmarks are icons, so a crawler sees feature names with no plan attached. "$49/month" sits beside the profile-fields add-on and reads like the mythical $49 tier. No monthly-billing prices anywhere.
  • T2, shared vertical template. Eight /platform pages carry the same FAQPage schema question, "How does Circle compare to other community platforms?", and a band titled "The complete community platform".
  • T3, inconsistent numbers. "18,000+ communities" (llms.txt) vs "20,000"; "48K courses" vs "63K"; customer community "10,000+" vs "27K" vs "30K+".
URLCrawler-readable? (words)VerdictFindingsRecommendationWhyEffortOwner
/Yes (1,272)fixTitle says "A new era for digital businesses" but there is no h1, and the Organization schema description still reads "Build a home for your community, events, and courses". "community" 28 times. No price, fee, or 2026 capability in text. "Smarter funnels" though no funnel object exists.Add an h1. Rewrite meta and schema descriptions to the digital-business line. Add one crawler-visible paragraph: what Circle is, from $89, fees 2/1/0.5 percent, what shipped in 2026.62 percent of ChatGPT referrals land on the homepage. Findings 1, 4.quickRudy design, JK
/pricingYes (1,300)fixT1 in full. Plans and fees are right. SSO providers listed with no plan mapping in text. Enterprise revert will change columns.Render plan availability as text. Add monthly prices and a fee-math example. Rename "Content co-pilot". Label add-ons. Publish a markdown twin.Corrected every price on the spot in the benchmark. Situations 1, 9, 16, 19, 20.quickweb eng, PMM
/pricing-calculatorYes (388)keepClean FAQ schema, "digital business" in h1, "14,000+ real digital businesses". Not in llms.txt.Add to llms.txt. Add one line with Circle's own plan prices so it is not mistaken for Circle's pricing.Selling-word page with buyer nouns. Finding 6.quickJK
/platformYes (1,621)rewriteTitle "The Complete Community Platform"; "community" 45 times. Good facts buried: BNPL, installments, "100+ triggers". FAQ: "the most complete, easy-to-use, and AI-native community platform". Not in llms.txt.Retitle to the digital-business frame. Keep the FAQ, change the category noun. Add the 2026 list with plan gating. Add to llms.txt.The page a model reads to define Circle. Findings 1, 2.mediumJK
/platform/coachingYes (1,782)fixDescription says "all-in-one coaching platform" (good). Good 1-on-1 FAQ. T1, T2. No installments, BNPL, one-time packages, or application forms named. Not in llms.txt.Name the coaching selling stack: packages, installments, BNPL, paid workshops, upsell to membership, Forms for applications. Schema question "for coaches". Add to llms.txt.Group coaching is Fragile; Kajabi wins "all-in-one". Situations 1, 6, 9.quickJK
/coachingRedirects to /platform/coachingkeep301 to the canonical page; content identical.Nothing.Avoids a duplicate.quickweb eng
/platform/course-creatorsYes (1,518)mergeh1 "The best courses are powered by community" frames courses as an add-on, the frame that loses to Teachable. Overlaps /platform/courses. T1, T2.Fold the retention argument into /platform/courses as one section; redirect this URL.Two course pages split authority and one teaches the losing frame. Cohort cell.quickJK, web eng
/platform/coursesYes (1,714)fixTitle has "Sell Courses" and names upsells. Missing quizzes, certificates, drip, the Eclipse builder. "63K courses" vs "48K" elsewhere. T1, T2. Not in llms.txt.Make this the canonical sell-a-course page: pricing shapes, upsells, quizzes, certificates, drip, AI-built curriculum. Fix numbers. Add to llms.txt."Online course" is Absent; models doubt certificates exist. Situations 8, 18.mediumJK
/platform/creatorsYes (1,638)fixGeneric "Community Platform for Creators & Solopreneurs". "AI content co-pilot" in the feature strip. T1, T2.Retitle to an outcome ("Turn your audience into a digital business"). Replace co-pilot with Circle AI.Persona prompts drop Circle; the page never says what a creator sells.quickJK
/platform/newsletter-creatorsYes (1,559)fixWins the newsletter-plus-community cell today. But "Offer flexible pricing models including pay-per-article and special editions" is not in product truth, and "Templates for landing pages, newsletters" reads as newsletter sending before January. T1, T2.Remove pay-per-article until verified. State plainly: Email Hub sends broadcasts (add-on); a native newsletter product is coming. Keep the community-for-readers pitch.Buyers arriving on the promise cannot use it until January. Situation 5.quickJK, product
/platform/podcastersYes (1,546)fixThe only page where Circle AI surfaced in the benchmark. Broken FAQ: "Do you offer migration services" is answered with "I just launched my first course about 6 months ago". Steps header says "Launch your course on Circle". T1, T2.Fix the FAQ and header. Add "membership site for podcasters" language and an honest line on gated audio.Podcaster membership is Absent; this page is already citable. Situation 3.quickJK
/platform/youtubersYes (1,590)fixSame broken migration FAQ. Good outcome h1. Nothing on public webinars, RSVP, or AI-built sales pages. T1, T2.Fix FAQ. Add the webinar-to-checkout sequence with plan gating. Add price and fees.The 50K-YouTuber persona goes to Skool; situation 2 hedged checkout.quickJK
/platform/instagram-creatorsYes (1,559)fix"Send newsletters and updates directly from Circle, without extra tools" without saying Email Hub is an add-on. T1, T2.Qualify the email line. Add a one-checkout example (template, course, community).Situation 10 was right; add the storefront proof.quickJK
/platform/community-managersYes (1,648)keepRetention vocabulary fits this persona. "AI content co-pilot" in feature strip. T1, T2.Swap co-pilot for Circle AI.This buyer does say "community".quickJK
/platform/customer-communitiesYes (1,487)merge"Embed community into your product with SSO" and "built-in AI agents" with no plan gating; no "knowledge base" wording. Overlaps /plus/saas, which is noindex. T1, T2.Merge /plus/saas into this page. State SSO and agents by plan. Call the knowledge engine a knowledge base.Situation 20 gated SSO wrong and denied an AI support agent.mediumJK, PMM
/platform/professional-communitiesYes (1,507)fixOnly page mentioning "Screen members with applications and vetting tools" and "dues, and renewal reminders". Title "Professional Development Communities" is nobody's search term. T1, T2.Name Forms as the application tool. Retitle toward associations, paid clubs and peer groups."Paid club" is a cold number one with no page. Situations 6, 9.quickJK
/membershipYes (1,089)fixFAQ says "over 10,000+ members" (T3). Business card omits Circle MCP; Plus card adds "Advanced analytics" (drift from /pricing). No installments, trials, tier upgrades, upsells, or fees named.Add the selling stack in text: billing intervals, trials, installments, tier upgrades, upsells, fees by plan. Sync cards to /pricing."Paid membership" is Absent; the page never says what a membership can charge.quickJK
/coursesYes (348)mergeEclipse launch page. No pricing, gating, quizzes, certificates, or selling verbs. llms.txt still describes it as "drag-and-drop builder".Redirect to /platform/courses, or make this the canonical course page. Not three course pages.Three URLs split the one query Circle loses.quickJK, web eng
/aiYes (1,827)fixh1 "Build your digital business" but title "Build your community business" and description "your community co-pilot". "Analytics Manager" skill carries the Notification Manager text. No credits number or plan line. Never states "builds a community from a prompt". Not in llms.txt.Fix title and description. Port the "mindfulness community" example from /eclipse-release. Add "included on every plan, 2,500 credits a month". Fix the skill card. Add to llms.txt."AI sets my business up" is an unnamed category three small vendors are claiming.quickJK
/ai-agentsYes (863)fixFAQ: "AI Agents are included for Circle Plus customers at no additional cost in 2025." Title "Introducing". No analytics or paywall-scoped agents.Update the plan sentence. Add "scope an agent to a paid tier" and agent analytics.Finding 2: agents on Plus unknown to models.quickJK
/ai-workflowsYes (758)fix"5,000 AI credits/month" on Plus collides with "Circle AI credits 2,500/mo" on /pricing; different meters.Rename to "AI action credits" and add one sentence separating the two.A model reading both pages will merge them.quickJK, PMM
/mcpYes (676)fixCanonical tag points to https://circle.so/chat, telling search engines this page is a duplicate. Not in llms.txt. Good setup steps and Business gating. No docs link, launch date, or read-only mode named.Fix the canonical. Add to llms.txt. Link api.circle.so/mcp. Say "since April 2026" and "read-only or full access".A good server nobody can find, and the page tells crawlers to ignore it.quickweb eng, product
/eclipseYes (1,188)mergeFive FAQ questions have no answers in the crawler view, including "Will these new features be available on my plan?" and "Are there any additional token costs?" Body is duplicated testimonials.Fold into /eclipse-release or server-render the answers.The unanswered questions are exactly what models need.quickweb eng
/eclipse-releaseYes (1,421)keepBest "digital business" prose on the site, the prompt-to-community example, Discover and Studios. Same empty FAQs. Dated "Rollout begins today". Not in llms.txt.Render the FAQ answers. Add a date line. Add to llms.txt.Citable proof for the AI-builds-it category.quickJK, web eng
/plusYes (1,137)fixStrong proof; inclusions are adjectives, not a list. llms.txt describes Plus as "White-label branded mobile apps", which is wrong.Add a plain inclusions list (up to 10 agents, AI workflows, SSO, 0.5 percent fee, 10K Email Hub contacts, sandbox, optional branded app). Fix the llms.txt line.Plus deals need the facts in text.quickJK
/plus/saasYes (1,445) but noindexmergeBest B2B FAQ on the site (SSO provisioning, API to warehouse, ROI case). Invisible to search.Merge into /platform/customer-communities, or remove noindex.Situation 20 had no page to cite.quickJK, web eng
/plus/branded-appPartial (562, no h1, twelve "Loading..." placeholders)rewriteCrawler sees no feature list, in-app purchases, app builder detail, or pricing model.Server-render: inclusions, IAP with store commission, push, custom screens, "add-on to Plus, quoted by sales".Situation 15: branded-app pricing is not public.mediumweb eng, JK
/plus/customer-storiesPartial (218)fixOnly three stat tiles render; story cards are JS.Server-render the card list with one-line outcomes.Named customer proof wins citations.quickweb eng
/plus/supportYes (762)keepClear migration timelines and CSM tiers.T3 consistency only.Fine.quicknone
/plus-salesJS shell (94)keepForm page, two h1s.One server-rendered paragraph on what Plus is. Low priority.Not a citation target.quickweb eng
/migrationYes (1,032)keepStrong: source platforms, timelines, eligibility. Bug: the email section's button reads "Request course migration". "~5 business days per course" conflicts with /migration/courses.Fix the button and align timelines.This family corrected the Kajabi-switch story. Situation 18.quickweb eng
/migration/coursesYes (736)fixInternal note published in the FAQ: "This one might be worth discussing internally. We get this request often." Timeline "2-4 weeks".Replace with the real quiz answer (not migrated today). Align timeline.Credibility on a citable page.quickJK
/migration/emailYes (596)keepHonest evaluation-gated program.Add the source-platform list from the hub.Fine.quickJK
/migration/paymentsYes (1,106)keepClear $250K priority and "$50K and $249K" waitlist. Typo "we can connection the subscriptions".Fix typo.Fine.quickJK
/llms.txtYes (5,432)rewriteOpens "Circle is the all-in-one community platform" and "community-first, not course-first or marketing-first". "18,000+ communities". Links "AI Copilot" at /ai-copilot. Describes Plus as branded apps. No fees, MCP, Discover 2.0, Studios, Eclipse, /platform pages, or 2026 capabilities. "community" 257 times.Rewrite the header as the digital-business definition with prices and fees. Add sections: 2026 capabilities with plan gating, MCP, Discover, Studios, vertical and compare pages. Fix counts.Finding 4. An hour of work.quickJK

Pages that should exist and don't

  • Pricing and fees explained. Total cost at $5K, $20K, $40K monthly sales, monthly vs annual, Email Hub add-on, an honest line against Skool's $899 threshold. No page corrects the "hidden fees" myth. Quick. PMM.
  • What shipped in 2026, with plan gating. Public RSVP and webinars, AI website builder, upsells, Forms as intake, SSO by plan, migration service, agents on Plus, MCP. Every wrong capability answer in the benchmark comes back to this missing page. Medium. PMM.
  • What is Circle, plain facts. Plans, what it is and is not, distinct from Circle Internet Group. So a model reading "Circle" knows which company is meant. Quick. PMM.
  • Sell on Circle. The selling stack on one page: billing intervals, installments, BNPL, trials, upsells, coupons, affiliates, paid events, in-app purchases. "Digital products" is the top buyer term and Circle has no page. Medium. JK.
  • Circle Plus inclusions and pricing model. A text list, separate from the proof page. Quick. JK.

Top five fixes by upside per effort

  • Fix the shared pricing block and /pricing (T1): rename "Content co-pilot", text-render plan availability, add monthly prices and fee math. One component fixes fifteen pages and most wrong facts. Quick, web eng and PMM.
  • Rewrite llms.txt. Circle's own machine-readable description still says "community platform". Quick, JK.
  • Fix the /mcp canonical and list it. The page tells crawlers it is a copy of /chat. Quick, web eng.
  • Fix the vertical template (T2): schema question, the "complete community platform" band, the broken podcasters and YouTubers FAQ. Eight pages at once. Quick, JK and web eng.
  • Fix /ai title and description and add the prompt-to-business example. An unnamed category is being claimed by smaller vendors while Circle's page calls itself a co-pilot. Quick, JK.

Feature pages and standalone pages

Word counts are crawler-visible words from metrics.csv. Every page is crawlable HTML. "Content co-pilot" survives only on /brand, /2023-review and the November 2025 entry of /product-updates. Pattern on the feature pages: 2025-current copy, 2023-era FAQs, and not one names a 2026 capability in buyer words (webinar, registration, upsell, order bump, application form, sales page).

Feature pages

URLReadable (words)VerdictFindingsRecommendationWhyEffortOwner
/eventsYes (921)fixTeaches "virtual events for members." Never says public, webinar, registration, ticket or lead. FAQ: "Pro Plan: up to 100 total participants" and "up to 15 participants", no Business or Plus numbers, which feeds the "20-person live room cap" myth. llms.txt promises "paid ticketing"; the page never mentions paid events.Add "Run public webinars and paid workshops": public RSVP (all plans), public live chat, recurring public events, paid tickets, every guest becomes a CRM contact. Add a plan table for stream and room limits (100/200/flexible, 15/20/flexible). Retitle "Webinars, workshops and paid events."Finding 2: situation 2 (webinar that sells a course) was hedged because no page says registration. Kills the cap criticism.quickJK, PMM for limits
/paymentsYes (713)fixFees correct (2/1/0.5). No upsell, order bump, paid ticket, tier or built-in affiliate program. "over 25 local currencies" is unverified. Migration FAQ says "FREE comprehensive migration service" with no plan gate; /customer-success gates it to annual Business+.Add "Sell more at checkout": post-checkout upsells (live, all plans), multi-year installments, BNPL and Apple/Google Pay by name, paid tickets, subscription groups for tiers, your own affiliate program (Pro+). Verify order bumps before naming them. Add the fee math. Reconcile migration gating.Finding 2: one-click upsell decides the tripwire cells and is never attributed to Circle. Gives "hidden fees" a citable correction.quickJK, product to confirm order bumps
/website-builderYes (621)fixNo mention of AI. Sells "drag-and-drop", "100+ pre-built blocks." Says "sales pages" once. FAQ good.Lead with the Aug 20 AI builder: describe a sales page and get it in your brand; rebuild from a URL or screenshot; SEO audit. Retitle "AI website and sales page builder."Finding 2 and the "AI sets my business up" cell where three small vendors claim prompt-to-page and Circle is absent.quickJK
/email-hubYes (808)fixClosest to current. Missing public RSVP lead capture and forms writing to profile fields. Pricing FAQ says "you can go here" with no numbers, so an agent cannot correct the $99 figure third parties repeat.Add the price ladder inline (free to 100 contacts, 10K included on Plus). Add a "forms as application intake" bullet. Say "newsletter" nowhere until January.Finding 2 ("no email marketing" is two years stale; Email Hub at $99 is a load-bearing criticism with no first-party correction).quickJK
/workflowsYes (608)fixFAQ: "Workflows are available on our Business and above plans." Pricing TSV gives Professional 20; verify. No revenue triggers, no AI. Engagement words only ("award badges").Add "Automate the sale": 21 paywall triggers, upsell accepted/declined/abandoned, change subscription, coupons. Add "Circle AI writes the workflow from a sentence." Say "branching" only once product confirms. Fix the FAQ.Finding 6 (retention words lose to selling words); "automations need Zapier" is a repeated criticism.quickJK, product
/analyticsYes (529)fixFAQ: "Filters and data grouping are available only on Circle Plus plans," likely stale since the 2024 rebuild; verify. Missing Copilot insights (ask a question, get a CSV), upsell analytics, AI agent analytics. Title "Community Analytics."Add the three 2026 items and correct the FAQ. Retitle "Revenue and member analytics."Finding 2.quickJK
/headlessYes (905)fixEnterprise sales page ("Talk to sales" five times). Typo "Circles provides." Nothing on Admin API v2, OpenAPI, SDKs, MCP or plan gating.Add a developer block: API on Business+, OpenAPI link, SDK languages, MCP, docs llms.txt. Fix typo.Builders audience graded D+; the research found the developer surface strong and untold.quickJK, web eng
/integrationsYes (234)rewriteThe directory is JavaScript; crawlers see a filter bar, one Exit Five quote and two stats. llms.txt promises "Native connections to Stripe, ConvertKit, Mailchimp, Zapier" that the page never states.Render the integration list as HTML grouped native / Zapier / embed, with a sentence each. Add MCP as an integration.Finding 3 (JS pages are invisible).mediumweb eng
/mobileYes (176)merge51 words of body. A download stub.Fold into a section of /customization or /plus/branded-app, or add 200 words on what members do in the app (chat, live rooms, courses, recording).Thin pages earn nothing.quickJK
/desktop-appYes (636)keepCurrent and clear. "hope to announce later this year" for branded desktop is undated.Date or remove the promise.Minor.quickJK
/chatYes (565)keepSolid. FAQ links point to community.circle.so posts (JS). Compares itself to "Slack and Discord."Add one line: "a paid Discord alternative: chat spaces plus paywalls, courses and events in one place."Discord alternative is Circle's best live result; reinforce it.quickJK
/discussionsYes (688)keepCurrent; names AI summaries and AI workflows. "AI workflows" is Plus-only, unflagged.Add plan tag.Accuracy.quickJK
/gamificationYes (460)keepCorrect gating stated. No "challenge" language.Add "run a 30-day challenge: daily check-ins, leaderboard, rewards via workflows.""Challenge" cell: Circle has every mechanic and no page saying so.quickJK
/customizationYes (770)fixFAQ: "on Pro and Enterprise plans, you can set up a custom domain." Custom domain is on all plans and the plan name is 2023. "Train AI agents" (Plus) unflagged.Rewrite the FAQ against the pricing table.Stale plan names are what models quote.quickJK
/discoverYes (433)fixEvery featured creator is labeled "Productivity expert", including Jay Shetty and Dana Malstaff. No FAQ, no plan note, no purchase mechanic.Fix labels. Add: free to list on any paid plan, reviewed in 48 hours, "Join or purchase" on the listing, 100K+ visitors a month. Add FAQ schema.Discover is an acquisition channel no competitor has; agents cannot cite what the page does not say.quickJK, Rudy
/inboxYes (212)fixEclipse-era, thin, not in llms.txt. The only feature page that says "digital business" (three times).Add 200 words and a FAQ (all plans; AI Inbox needs Plus agents). Add to llms.txt.Thin page; good framing worth extending.quickJK

Standalone pages

URLReadable (words)VerdictFindingsRecommendationWhyEffortOwner
/studiosYes (765)keepClear done-for-you rev-share offer; "hundreds of thousands of followers" qualifier stated. Not in llms.txt. No FAQ schema.Add to llms.txt and a FAQ.Agents answering "who runs it for me" have no citable Circle page listed.quickJK
/academyYes (818)fixLorem ipsum live: "Sorem ipsum dolor sit amet, consectetur adipiscing elit" on the Circle Foundations card. Title "Academy: Get Early Access" for a live product. "Community Manager Certification later this year" undated.Replace placeholder, retitle "Circle Academy", date the certification.Placeholder text on a page in llms.txt.quickJK
/build-a-communityYes (370)keepLead magnet; names "course, membership, or customer community."Add "coaching program, mastermind, cohort" to the audience list.Vocabulary finding.quickJK
/community-roi-calculatorYes (763)keepMoney page with real quotes (Clouse $450K ARR). "15,000+ others" stale. Calculator is JS; the prose is fine.Update count. Add a "course or coaching add-on" note; retitle to "Membership revenue calculator."Membership cell is Absent; this is a citable revenue page.quickJK
/demo-videoYes (144)retireEnglish title "Book a Demo" over a Portuguese page: "Conheça nossa plataforma", a Brazil sales form, "Trusted by 15,000+."Move to /br/ and redirect /demo-video to the demo page.Confuses the entity for any reader.quickweb eng
/v3Yes (707)retire2024 launch page: "revert back to 2.0 until November 30, 2024."301 to /product-updates.Stale pages get quoted.quickweb eng
/the-making-ofYes (587)mergePodcast plus newsletter "every other week," while /members-only is the current twice-monthly newsletter. Two newsletter pages.Keep as podcast archive, point newsletter sign-up to /members-only.Duplication.quickJK
/members-onlyYes (446)keepCurrent (next issue Sep 15). Not in llms.txt.Add to llms.txt.Minor.quickJK
/ai-prompt-libraryYes (780)fixBest 2026 page in the group: names MCP and Circle AI, "build my whole community based on my website/sales page." Only 12 of 268 prompts render; the rest sit behind "Show more." Not in llms.txt.Render all prompts or paginate with crawlable URLs. Add to llms.txt.Finding 2: the page that proves "AI sets my business up" is mostly invisible.mediumweb eng
/product-updatesYes (947)fixFive 2026 summaries are readable and correct (AI website builder, public webinars, MCP, Eclipse). ItemList schema lists 32 releases; 30 link to community.circle.so, which renders a title and nothing else; "Load more" hides the rest. November 2025 entry says "AI copilot."Render all 32 summaries. Host full notes at circle.so/product-updates/slug as HTML (the Eclipse post already is). Add "now Circle AI" to the copilot entry. Add to llms.txt.Findings 2 and 3: the only first-party record of 2026 shipping, with invisible deep links.mediumweb eng, PMM
/partner-directoryYes (176)fixDirectory is JS; 66 body words.Render the partner list as HTML.Finding 3.mediumweb eng
/customer-successYes (677)fix"13,000+ customers" stale. Migration gating contradicts /payments.Update count, align gating with pricing, link the migration pages.Migration service is a 2026 fact agents lack.quickJK
/affiliate-programYes (1143)keepCurrent: "$100 + 20% recurring" since June 24, 2026, so the research line "Circle pays $120 once" is stale. Says "AI partner that builds, runs, and grows your digital business."Add a "how to describe Circle" block for affiliates with the 2026 capabilities and selling verbs.Finding 5: affiliates write the listicles; give them the words.quickJK
/affiliate-faqsYes (150)mergeEmpty shell: h1 and "Still can't find the answer?"301 to /affiliate-program#faq.Dead page.quickweb eng
/brandYes (979)rewrite or retirenoindex but fetchable. "monetize your community in 6 weeks or less," "all-in-one platform for building and scaling your community," Business plan lists "Content co-pilot."If it is a paid landing page, refresh the plan lists from the pricing table and drop co-pilot; otherwise delete.Finding 4 (co-pilot naming).quickJK
/jay-clouse-community-masterclassYes (419)keepMoney words in title ("Turn Your Audience Into Revenue"), $400K proof.None needed.Already sells.noneJK

Reports, events, year reviews

URLReadable (words)VerdictFindingsRecommendationWhyEffortOwner
/2024-creator-community-reportYes (1652)keepRich data: 93% of top communities charge; half use three tools or fewer.Add a line linking findings to Circle's checkout and consolidation.Citable evidence for "paid" and "all-in-one."quickJK
/2025-community-trends-reportYes (1436)keep"one community member is worth 234 social media followers"; consolidation trend.Keep; cross-link to /pricing and /payments.Same.quickJK
/2026-community-trends-reportYes (446)fixMostly gated; 400 words of teaser.Publish three trends in full on the page.Thin pages are not cited.quickJK
/2025-community-trends-event, /2026-community-strategy-eventYes (762, 620)keepPast events, on-demand. 2026 page still says "Save your seat" for Feb 3.Change CTA to "Watch replay."Accuracy.quickJK
/2021-review, /2022-reviewYes (207, 229)retireUnder 130 body words each; no product facts.301 to /product-updates.Dead weight.quickweb eng
/2023-review, /2024Yes (1076, 976)keepComplete monthly timelines; useful "when did Circle ship X" record. 2024 calls Email Hub "Marketing Hub"; 2023 says "content co-pilot."Add "(now Email Hub)" and "(now Circle AI)" notes.Naming continuity.quickJK
/2025Yes (523)fix"At a glance" renders January only; Feb to Dec are empty for crawlers. Quotes are good ("My AI agent is amazing").Render every month's items as HTML like /2024.Finding 3.quickweb eng

Careers and legal

URLReadable (words)VerdictFindingsRecommendationWhyEffortOwner
/careersYes (879)fixDescription: "the all-in-one community platform for creators and brands." Body: "world's leading branded community platform." Says both "200+ team members" and "160+ team members."Use the current company description and one headcount.Finding 4: framing lives on every page an agent reads.quickJK
/careers/ashby-* (10 postings: Lead Product Designer x2, Growth Lead Discover, Sr Brand Designer, Creative Director, Sr Engineers AI Core / Discover / CMS, Social Media Lead, Sr Quality Engineer)Yes (789 to 1692 each)keepAll crawlable. The Discover roles describe it well: "curated marketplace for finding and buying digital experiences."None.none
Legal (terms, privacy, EU and California privacy, DPA, DMCA, cookie, anti-spam, creator terms, affiliate terms)Yes (all long)keepTerms and privacy updated Sep 1, 2026; privacy opens "Circle is an all-in-one community platform for creators and brands." Swap that sentence when the company line changes; nothing else.quicklegal

Pages that should exist and don't

  • /webinars: public registration, live chat for guests, replay, guest to CRM contact, in the buyer's word. Could be a section of /events but a URL with the word ranks. Effort: quick.
  • /checkout (or /upsells): upsells, installments, BNPL, coupons, tiers, paid tickets, fee math. The tripwire and low-ticket cells turn on this. Effort: quick.
  • /forms: application intake for masterminds, accelerators and cohorts, writing to profiles and segments. Effort: quick.
  • /affiliates (customer feature): "run your own affiliate program" for paywalls, distinct from Circle's affiliate program. The research flagged the two-page confusion. Effort: quick.
  • /live: stream and room limits by plan in one table, plus recording, transcripts, breakout rooms once verified. Effort: quick.
  • circle.so/product-updates/slug: full release notes as HTML on the marketing domain. Effort: medium, web eng.
  • /whats-new-2026 capability page with plan gating (from the research's five changes). If site-core owns it, cross-link every feature page to it. Effort: quick, PMM.

Top five fixes by upside per effort

  • /product-updates: render all 32 releases and host the notes on circle.so. It is the only first-party record of 2026 and its links are invisible today.
  • /payments: add upsells, installments, tickets, own affiliates, fee math, retitle. Fixes the "hidden fees" myth and the missing upsell in the selling cells with one page.
  • /events: add public RSVP, webinars, tickets, plan limits, retitle with "webinar." Fixes situation 2 and the live room cap criticism.
  • /website-builder: put the AI builder first and "sales page" in the title. The prompt-to-page cell has three small vendors and no Circle.
  • Retire batch in one ticket: /v3, /demo-video, /affiliate-faqs, /2021-review, /2022-review, and fix /brand and /academy placeholder text. Stale pages are what models quote.

The comparison layer: compare hub, head-to-heads, alternatives posts

Reviewed Sep 14, 2026 against the research page, product truth (Sep 3) and the crawler fetches. All 32 pages are crawler-readable HTML; the problems are framing, missing 2026 facts, schema and llms.txt coverage.

Findings that apply to the whole group (letters referenced in the table)

  • A. Retention framing. Every head-to-head closes with "The complete community platform" and "serious community businesses"; only vs-beehiiv and vs-heartbeat say "digital business". Blog posts crown Circle "best for community-focused creators", the wording that loses to "sell, checkout, launch, revenue" (finding 6).
  • B. 2026 capabilities absent. No head-to-head names public webinars or RSVP lead capture (vs-hivebrite alone), the AI website builder (vs-beehiiv hints at it), Forms as application intake, or the MCP (footer link only). Post-checkout upsells appear on five pages and vs-thinkific concedes checkout commerce to Thinkific. vs-facebook-groups claims "Order bumps and upsells"; product truth says order bumps are not publicly documented and must not be claimed yet.
  • C. Schema. Every head-to-head has a long FAQ but only Organization and WebSite schema. Blog posts carry BlogPosting plus FAQPage, except discord-alternatives.
  • D. llms.txt. Lists the hub and eight head-to-heads, with Facebook at the wrong URL /compare/vs-facebook (the page is /compare/vs-facebook-groups). Missing: teachable, thinkific, discord, podia, substack, beehiiv, heartbeat, passion-io, higher-logic; blog posts hivebrite, podia, teachable, thinkific and both vs posts. The preamble still says "community-first, not course-first or marketing-first".
  • E. Price facts. The metrics.csv flags for "$39", "$49", "Basic plan" and "content co-pilot" here are false positives (competitor prices; "Copilot" is the current name). Real stale Circle facts are quoted in the rows. Circle's own price is missing from vs-skool, vs-mighty and vs-kajabi, the three most-searched matchups.
  • F. Competitor prices disagree across Circle's pages. Mighty entry is "$41/mo" on two posts and "$79/mo" on five; Skool Hobby is "$7.50" and "$9"; Kajabi Basic is "$143", "$149", "$143–$179" and "$179". An assistant reading two pages trusts neither.
  • G. Certificates contradict. vs-thinkific and teachable-alternatives say "no completion certificates"; vs-discord, vs-bettermode and vs-higher-logic say "certificates of completion". Product truth lists them as live; product should confirm and every page agree.

Who should rank for the queries competitors own

  • "Skool vs Circle" (Mighty ranks first): /compare/vs-skool, once it carries Circle's price, a fee table at three revenue levels against Skool's $99 plus 2.9 percent, the "creator with 50,000 followers" persona, and FAQPage schema.
  • "Circle alternatives" (Mighty): no Circle page exists. Publish /compare/circle-alternatives, written straight, so the query resolves on circle.so.
  • "Circle pricing" (SchoolMaker): /pricing, outside this group; every head-to-head should carry the same price and fee block and link to it.
  • "Circle review" (GroupApp): a first-party reviews page; nothing in this group covers it.
  • "Circle vs Mighty" and "Circle vs Kajabi": the vs- pages, once they carry prices, fee math and the 2026 list.

Page by page

URLReadable (words)VerdictFindingsRecommendationWhyEffortOwner
/compare2,643fixH1 "20,000+ communities chose Circle"; llms.txt says "18,000+". Feature grid is unlabeled cells ("Limited / Add-on / Basic") to a crawler. Title says 14 competitors, body 17. FAQ concedes "a simple subscription newsletter (Substack does that more elegantly)". Good: prices, fees, trial, guarantee. A, C, D.H1 to "digital businesses". Labeled HTML table. A 2026 line on every card. Fix 14 vs 17. Soften the newsletter concession before January. FAQPage.Finding 5: the hub is what assistants cite.quickJK, web eng
/compare/vs-skool2,383fixNo Circle price; fee math is Skool-only ("$100 gone" at $1,000). "Multiple currencies (15+)" here, "25+" on vs-thinkific. Good buyer words (mastermind, high-ticket, cohorts). A, B, C.Add the price block and a three-level fee table. Add webinars, AI sales pages, Forms, MCP. Reconcile currencies. FAQPage.The page that should own "Skool vs Circle" cannot be cited for cost.quickJK
/compare/vs-mighty2,111fixNo Circle price. "Unlimited ... API request volume" contradicts API and MCP usage counting against credits since May 2026. Attack tone ("Mighty is stagnant"). A, B, C.Add price and fees. Replace "unlimited API" with real limits. Name MCP (Mighty has none) and the AI website builder. Trim insults. FAQPage.Mighty wins "best AI features" with a louder story; MCP and AI pages are the proof.quickJK, PMM
/compare/vs-kajabi1,446rewriteNever claims "all-in-one" or "digital business platform". No Circle price; Kajabi's 0 percent fee unaddressed. "$143/month ... forced onto their $199/month plan" disagrees with $249 elsewhere. Claims Circle sells "1:1 offers", not in the selling stack. Thin. A, B, C, F.Rewrite as the digital-business page: $89 plus 2 percent vs $143 to $179 plus 0 at three revenue levels; name upsells, AI sales pages, webinars, Forms, affiliates, MCP; concede the missing funnel object honestly. Drop "1:1 offers". FAQPage.Finding 1: Kajabi holds "all-in-one"; this is where Circle contests it.mediumJK, PMM
/compare/vs-teachable2,029keep, small fixBest fee page: 7.5 percent Starter, Circle 2/1/0.5 with prices, plan caps, stack cost. Concedes certificates (G). No upsells (Teachable has them), AI pages, webinars. C, D.Resolve G. Add upsells, AI sales pages, webinars. FAQPage. llms.txt.Already citable on cost.quickJK
/compare/vs-thinkific2,870fix"Does not currently offer native course completion certificates" (G). "Email Hub ... starting at $99/month" omits free to 100 contacts. "Upsell paths through subscription groups" ignores post-checkout upsells live since January; concedes "order bumps, abandoned cart". B, C, D.Rewrite the commerce FAQ around real upsells, coupons, installments; verify order bumps before adding. Fix the Email Hub line. Resolve G. FAQPage. llms.txt.Finding 2: checkout upsell is the feature "Circle has and nobody can find".quickJK, product
/compare/vs-discord2,103keep, small fixCircle's best cell in the research. Fee math correct ($5,000: Discord ~$500, Circle $175 to $245). Claims "certificates of completion" (G). No webinars, MCP. C, D.Resolve G. Add public webinars and the creator-with-an-audience persona. FAQPage. llms.txt.Circle's strongest query.quickJK
/compare/vs-facebook-groups1,809fix"Order bumps and upsells" (do not claim order bumps yet). "Facebook's average engagement rate is 0.063%" unsourced. No Circle price. RSVP named but not public RSVP. B, C, D (wrong URL).Remove "order bumps" until documented. Add price block, public RSVP and webinar lead capture, AI pages. Fix the llms.txt URL. FAQPage.A wrong claim on a cited page becomes a repeated fact.quickJK, web eng
/compare/vs-patreon1,701keep, small fixPrice block present; "$1,500 per month" breakeven checks. Podcaster FAQ ignores private RSS and "membership site for podcasters". C.Add podcaster-membership wording, upsells, webinars. FAQPage."Membership site for podcasters" is the winnable adjacent page.quickJK
/compare/vs-podia1,268fixNo Circle price or fees (Podia Mover is 5 percent; easy win). No upsells, AI pages, webinars. Thin. A, B, C, D.Add price and fee table, upsells (Podia has them), AI sales pages, webinars. FAQPage. llms.txt.Cheapest page to make sell for product-seller queries.quickJK
/compare/vs-substack1,224fixFee block present. Promises "broadcasts, newsletters" via Email Hub; the newsletter product ships January 2027. "Early 2026" Substack facts will date. Thin. C, D.Stay honest until January, then relaunch with the product. Add upsells, public RSVP. FAQPage. llms.txt."Paid newsletter" is a clean zero; only the January product moves it.quick, medium in JanuaryJK
/compare/vs-beehiiv1,667keep, small fixOnly page with "digital businesses" in H1 and body, and "Build courses, paywalls, and workflows from a prompt". Feature table is bare rows to a crawler. Same newsletter caveat. C, D.Use its H1 as the group template. Label the table. Add price, MCP. FAQPage. llms.txt.The frame already exists here.quickJK, web eng
/compare/vs-slack1,139fixOne-word FAQ answers ("No."). "Platform pricing (not per member)" without the number. A, B, C.Two-sentence answers with the price; add webinars, Forms, MCP. FAQPage.A citable answer needs a fact in it.quickJK
/compare/vs-bettermode2,242keep, small fixOnly real side-by-side price table ($199 vs $399). Claims "certificates" (G). "Eliminated its free plan in early 2026" unsourced. C.Resolve G. Add SSO gating, MCP. FAQPage.The model price table.quickJK
/compare/vs-hivebrite976fixOnly page naming public RSVP and checkout upsells, yet thinnest; one-line FAQ answers; no price. A, C.Double the length: association use cases, SSO by plan, price. FAQPage.Thin pages lose citations.quickJK
/compare/vs-higher-logic1,244fixClaims "Completion certificates" (G). AMS answer is Zapier and API only. No price. A, C, D.Resolve G. Add SSO (Plus), API, MCP, price. FAQPage. llms.txt.Buyers ask about SSO and AMS.quickJK
/compare/vs-heartbeat1,149keep, small fixFee comparison present (5/2.5/1.25 vs 2/1/0.5). Names Analyst, Growth Marketer, Strategist, "50+ business skills": best Circle AI copy in the group. "Contact management system (CMS)" typo. No price. C, D.Reuse its AI paragraph everywhere. Fix CMS. Add price. FAQPage. llms.txt.Models know nothing about Circle AI; the story lives here.quickJK
/compare/vs-passion-io1,172keep, small fixPrice block present. "147% membership growth" and "Seven times higher mobile adoption" unsourced. C, D.Source or cut the stats. Add IAP and app-builder facts. FAQPage. llms.txt.Unsourced numbers get repeated as Circle's claim.quickJK, PMM
/compare/kajabi, /compare/mighty-networks, /compare/skool1 each (404)retireJS-shell 404s; unlinked, but assistants guess these slugs.301 each to the vs- page.Guard the guessed URLs.quickweb eng
/blog/skool-alternatives4,314keep, small fixCircle first; "scaling to $5K+/mo, Circle" is selling language. "CirclePlus" label. Skool "$9" here, "$7.50" elsewhere (F). A.Add a "what Circle sells that Skool cannot" paragraph. Fix "CirclePlus". Align prices.Second page in the "Skool vs Circle" cluster.quickJK
/blog/mighty-networks-alternatives4,238keep, small fix"Circle offers API access across its plans" (API is Business and above). Mighty "$79" vs "$41" elsewhere (F). A.Fix the API gating. Align prices. Add MCP and AI pages.Wrong plan gating on a cited page.quickJK, PMM
/blog/kajabi-alternatives5,739fixTable: Circle "Transaction Fee 0.5–4%" (the 2024 fee). "AI Agents (Enterprise)" (plan is Plus). Kajabi Basic "$143–$179" and "$149" in one post. Cedes selling: "Kajabi optimizes for conversion; Circle optimizes for retention". A, F.Fix 4 percent and Enterprise. Rewrite the verdict so Circle wins conversion: upsells, AI sales pages, webinars, affiliates, Forms. One Kajabi price.Repeats the stale fee models quote; hands conversion to Kajabi.quickJK, PMM
/blog/patreon-alternatives6,214fix"$89–$199/month plus a 1–2% transaction fee, which at $5K/month works out to roughly $140–$290" (actual $189 to $249). "Best overall Patreon alternative for community-focused creators". F.Correct the math. Retitle "for creators who sell memberships, courses and coaching".Wrong fee math on the page most cited for Patreon fees.quickJK
/blog/facebook-groups-alternatives5,090fixFAQ: Circle AI is "read-only against your community data. Full general availability lands July 31, 2026" (GA passed; it executes tasks). Good: "own your entire funnel ... digital products". F.Update the AI answer to post-GA truth with plan gating. Add public RSVP.The July date has passed.quickJK, PMM
/blog/discord-alternatives6,211keep, small fixDirect-answer block "Circle is the best Discord alternative for community builders who need monetization ..." is the right shape. No FAQPage schema. A.Add FAQPage, the creator persona, upsells.Schema is the only gap.quickweb eng, JK
/blog/slack-alternatives4,434keepFull price and plan block, webinars named, "AI Agents on Plus", Business "adds workflows, AI tools, and branding removal". Mighty "$41" (F).Align Mighty price. Add Forms, MCP.Best-written Circle entry; the template.quickJK
/blog/teachable-alternatives4,472keep, small fixFee math correct with live-stream caps by plan. "No SCORM or completion certificates" (G). Not in llms.txt. A.Resolve G. Add upsells, AI pages. llms.txt.Already ranks for a query where Circle is "absent".quickJK, web eng
/blog/thinkific-alternatives4,308keep, small fix"The 0.5%–2% transaction fee is the honest trade-off" is the citable line. Gating correct. Mighty fees "2%, 1%, 0.5%" here, "3%, 2%, 1%" in the Teachable post (F). Not in llms.txt.Align fees. Add upsells, AI pages, webinars. llms.txt.Consistency.quickJK, web eng
/blog/podia-alternatives4,387fixH1 shows "who've". Heading "10 Best" for nine entries. "On Circle it is a paid add-on" omits the free email tier. Not in llms.txt. A.Fix entity and count. State Email Hub free to 100 contacts. Add upsells, AI pages, webinars. llms.txt.Small errors read as carelessness.quickJK, web eng
/blog/hivebrite-alternatives3,862keep, small fixSSO gating (Plus) and SOC 2 correct; migration "costs you nothing" (verify by plan). Not in llms.txt. A.Add public RSVP and Forms for applications. Verify the migration claim. llms.txt.Associations ask about applications.quickJK
/blog/free-vs-paid-newsletter2,344keepNot a competitor post. Describes Email Hub newsletters accurately; does not claim the January product. Not in llms.txt.llms.txt; revisit in January.Safe.quickweb eng
/blog/recurring-revenue-vs-one-time-sales3,100keepNot a competitor post. Selling words, named customers, dollar outcomes. Footer "Circle The complete community platform".Change the footer line.Already in buyers' words.quickJK

Pages that should exist and don't

  • /compare/circle-alternatives. Mighty owns the query about Circle; a straight page keeps the buyer on circle.so. Medium.
  • A pricing-and-fees explainer every compare page links to: total cost at three revenue levels against Skool, Kajabi, Mighty, Patreon, Substack, so the "hidden fees" myth has a page to be corrected by. Quick.
  • /compare/vs-whop and /compare/vs-stan. They take "sell digital products" and "creator storefront" queries where Circle is absent. Medium.
  • /compare/vs-disco and /compare/vs-teachfloor. Taking cohort, bootcamp and academy queries "with Circle's own story". Medium.
  • A first-party reviews page to contest "Circle review", owned by GroupApp. Medium.

Top five fixes by upside per effort

  • Price and fee block plus a three-level fee table on vs-skool, vs-mighty and vs-kajabi (port the vs-bettermode table). The most-searched matchups carry no Circle price, which is why competitors own "Circle pricing". Hours.
  • Fix the four wrong Circle facts on blog posts: "0.5–4%" and "AI Agents (Enterprise)" in kajabi-alternatives, "$140–$290" in patreon-alternatives, "read-only ... July 31, 2026" in facebook-groups-alternatives. These are the stale facts models repeat. Hours.
  • One "what Circle sells that X cannot" paragraph on every compare page: post-checkout upsells, AI sales pages, public webinars with RSVP capture, Forms as applications, MCP, migration service, with plan gating. Nothing shipped in 2026 appears on any compare page. Days.
  • FAQPage schema on all compare pages and discord-alternatives; add the missing URLs to llms.txt and fix the Facebook link. One engineering pass. Day.
  • Retitle Circle's entry in every alternatives post from "community-focused" to selling words and rewrite the kajabi-alternatives verdict so Circle stops conceding conversion. Hours per post.

Blog listicles, business-type how-tos, AI posts

Scope: 34 blog posts, each read as a crawler sees it. All render as full HTML, so visibility is not the problem. The 429 rows in metrics.csv for eight listicles are stale; the later fetch in the .meta files returned 200 with full text.

Two patterns. The listicles that rank put Circle at number one but describe it in retention words and never name the 2026 selling stack (public webinars, upsells, installments, Forms as application intake, AI website builder, Circle AI building from a prompt). The how-tos for the words buyers use (mastermind, group coaching, cohort) are advice pieces where Circle is an aside at step five, which is why Mighty's dedicated mastermind page wins a query the assistant already leans Circle on.

Every post carries BlogPosting and BreadcrumbList schema; "FAQ" means FAQPage is also present. "llms" means listed in circle.so/llms.txt.

The table

URL (circle.so/blog/…)Readable?VerdictFindingsRecommendationWhyEffortOwner
best-coaching-platformsyes (5,682)rewrite Circle entryCircle #1, "best overall coaching, course, and community platform". Verdict and FAQ are retention words: "Choose Circle if community engagement, group coaching, and member retention drive your business." Prices right. Never names upsells, installments, Forms as application intake, public webinars, AI website builder, or Circle AI building the program. FAQ, llms. Mar 2026.Rewrite the Circle entry, bottom line and FAQ in selling verbs: sell a package with installments, take applications with Forms, run a public webinar into checkout, upsell the membership after.Finding 6: every Circle sentence the assistant used came from this page and it still did not recommend Circle.mediumJK
best-community-platformsyes (8,233)fixCircle #1. "Best for: creators… who value engagement, retention, and monetization." Stale "Circle serves 18,000+ communities" beside the 20K+ banner. Stale name "a co-pilot to autonomously manage community tasks". No 2026 stack. Teaches "community platform" as the label. FAQ, llms. Updated Feb 2026.Fix count and name. Add one 2026 paragraph. Frame Circle as the digital business platform whose center is community so "all-in-one" sits on Circle's own domain.Finding 4: Circle's own top page is where the model learns the category word.quickJK
best-membership-platformsyes (4,743)fixCircle #1, "all-in-one membership and community platform". Prices and fees right. "AI Copilot and AI Workflows to automate onboarding and moderation" is the old name and hides Plus-only gating. No subscription groups, trials, installments, upsells. Schema datePublished (2026-07-31) is later than dateModified (2026-02-25). FAQ, llms.Add sell-a-membership mechanics (tiers with proration, trials, founding coupons, upsells). Fix AI names, gating, schema dates. H1 toward "platforms to sell a membership".Paid membership is absent for "best platform to sell a membership".quickJK
best-online-course-platformsyes (7,801)fixCircle #1 with the strongest selling copy in the group; names Eclipse, Circle AI, migration. Wrong fee: "Transaction fees run 2% on Professional, 1% on Business, and 0% on higher plans" (Plus is 0.5%). Certificates are attributed to Teachable, Coursera and LearnDash, not clearly to Circle. FAQ, llms. Jul 2026.Correct the fee. State certificates, quizzes and drip under Circle. Add installment pricing and the post-checkout upsell.Situation 8: every 2026 review says Circle lacks certificates and no Circle page contradicts them.quickJK
best-content-creator-platformsyes (4,560)fixCircle only under "Stage C: scale", as "the best community platform for building, engaging, and monetizing". The selling words sit in the Patreon, Gumroad and Podia sections. No Discover, no AI website builder. FAQ, llms. Jan 2026.Put Circle in Stage B (first revenue) with checkout verbs; add Discover as an audience channel.Digital products and the 50,000-follower creator belong to Skool, Stan and Gumroad.quickJK
best-event-platformsyes (5,726)fixCircle #1 "for community businesses". Circle's entry never says public RSVP, public webinar, paid tickets, or guest-to-CRM; only a FAQ line says "Circle lets you run public-facing events". The buyer checklist lists "Breakout rooms" as a must-have; product truth marks them uncertain. FAQ, llms. Apr 2026.Rewrite Circle's key features around the 2026 event stack. Soften breakout rooms until documented. Add a "webinar that sells a course" section.Situation 2 was absent live; no page describes public webinars with checkout.quickJK, product on breakout rooms
best-white-label-community-platformyes (7,185)keep, minorCircle #1 with selling copy (branded checkout, in-app purchases, launch team). "Copilot" once. The $49 flag is Uscreen's price. FAQ, llms. Mar 2026.Rename Copilot to Circle AI.Already the model of a listicle that earns the recommendation.quickJK
best-membership-site-examplesyes (4,438)keep, add paragraphWebsite Builder showcase; no AI website builder or recreate-from-URL (Aug 2026). FAQ, llms. Sep 2025.Add a "built from a prompt or URL, all plans" paragraph.Situation 10: no page mentions the AI website builder as a storefront.quickJK
best-newsletter-platformsyes (4,745)fix nowCircle #1 as a newsletter platform before the product ships (Jan 2027). Wrong twice: "Plans from $89/month (billed annually), flat regardless of list size" and "Circle's plan is flat, so a bigger list doesn't raise your bill." Email Hub is an add-on, free to 100 contacts, $99/mo at 10K. Not in llms. FAQ. Aug 2026.Correct pricing to the add-on ladder. Reframe as "newsletter plus paid community" on Email Hub; hold "newsletter platform" until January.Paid newsletter is a clean zero, and Circle's pages promise sending before launch. A wrong fee on a ranking page gets copied.quickPMM
best-ai-tools-for-community-engagementyes (3,559)keepCircle #1, Eclipse dated. Retention words fit this query. Not in llms. FAQ. Jul 2026.Add to llms.txt.Rebuttal to Mighty's louder AI story.quickweb eng
best-ai-tools-for-creating-online-coursesyes (4,575)keepCircle #1 with selling verbs, Plus gating stated, Discover named. The "co-pilot" flags belong to Heights. Not in llms. FAQ. Jul 2026.Add to llms.txt.Only page that says "AI builds the course, prices the cohort" for Circle.quickweb eng
best-ai-tools-for-managing-online-communitiesyes (2,702)fixCircle #1. Pre-Eclipse: "AI internally with Copilot, Inbox, Agents, and Workflows". Not in llms. FAQ. May 2026.Rename to Circle AI; add to llms.txt.Stale name.quickJK
ai-tools-for-coachesyes (2,751)keepCircle is the closing pick; $89/$199 and Plus gating correct. Not in llms. FAQ. Aug 2026.Add to llms.txt.Correct and citable.quickweb eng
ai-native-community-platformyes (2,790)fixSid's Eclipse post. Stale since GA Jul 29: "You can use them once the waitlist opens. Join the waitlist for Circle AI." Omits credits (2,500/mo) and "all plans". Title anchors "community platform". Not in llms. FAQ. Jun 2026.Replace the waitlist line with availability and credits. Retitle toward "AI-native platform for digital businesses". Add to llms.txt."AI sets my business up for me" is unclaimed; three smaller vendors rank and Circle is absent.quickPMM
ai-lesson-summaryyes (1,505)retire, 301Sep 2023: "Co-pilot is a feature that is part of our Community AI suite, which is available exclusively to customers on our Business and Enterprise plans." No FAQ. In llms.301 to /courses or /ai; remove from llms.txt.llms.txt points agents at a 2023 product and an Enterprise plan.quickweb eng
ai-qa-ideasyes (1,558)retire, 301Sep 2023, same co-pilot and "AI content writer" era. No FAQ. In llms.301 to /ai; remove from llms.txt.Same.quickweb eng
ai-agents-for-communitiesyes (5,897)fixApr 2025: "We just launched our own AI agents feature." No Plus gating, 10-agent cap, knowledge engine or inbox. No FAQ. In llms.Add a dated 2026 block or merge into /ai-agents.Situation 20: the assistant denied any AI support agent.quickJK
ai-workflows-for-community-managementyes (3,359)fixSep 2025. No plan gating; no note that AI actions bill on a separate meter. FAQ. In llms.Add gating and the credit note.Keeps AI credits and workflow runs distinct.quickJK
ai-best-practices-community-content-creatorsyes (3,072)keepJul 2026, current, MCP named. FAQ. Not in llms.Add to llms.txt.Current.quickweb eng
ai-community-management-buildersyes (2,449)fix, minorMay 2026: "an AI Copilot to help you take action". FAQ. Not in llms.Rename to Circle AI.Stale name.quickJK
ai-customer-journey-mapyes (2,774)fix, minorJun 2026: "AI Copilot is Circle's conversational assistant" plus two more. FAQ. Not in llms.Rename to Circle AI.Stale name.quickJK
ai-for-coachesyes (2,369)keepJun 2026. Selling words beat retention words, rare here. One "AI Copilot". FAQ. Not in llms.Rename; add to llms.txt.Coaching is the largest buyer segment.quickJK
ai-for-creator-entrepreneurs-leaner-businessyes (2,421)keepAug 2026, current. FAQ. Not in llms.Add to llms.txt.Current.quickweb eng
ai-for-small-businessyes (3,356)fix, minorJul 2026. Three "AI Copilot". H1 says "small community business". FAQ. Not in llms.Rename; "small digital business" in the H1.Stale name, category word.quickJK
ai-in-community-engagementyes (2,646)keepAug 2026, current. FAQ. Not in llms.Add to llms.txt.Current.quickweb eng
ai-in-email-marketingyes (1,743)keepJun 2026. FAQ. Not in llms.Add to llms.txt.Current.quickweb eng
ai-lead-nurturing-toolsyes (3,096)keep, minorJun 2026. One "AI Copilot"; ten "webinar" mentions, none saying Circle's are public with guest capture. FAQ. Not in llms.Rename; one sentence on public RSVP as lead capture.Puts the 2026 releases where a lead-gen query retrieves them.quickJK
mastermind-groupsyes (4,740)rewrite into term pageMay 2025. Title is a how-to ("Start a powerful mastermind group in 7 steps"). Circle appears at step 5 and the close, in retention words (16 selling to 93 retention). No application intake with Forms, installments, price, or upsell. No FAQ. In llms.Keep the URL; rewrite as "mastermind group platform": buyer word in title and H1, mechanics named (application form, installments, private cohort space, live rooms, tier upsell), one real customer with a price, FAQ schema. Model on Mighty's page.Sharpest gap in the research: cold rank 1, live absent. Situation 6 went to Typeform because no page presents Forms as intake.mediumJK
group-coachingyes (4,209)rewrite into term pageJun 2025, "Group coaching 101". Circle enters a third of the way in and closes on "a strong, engaged, and long-lasting coaching business and community". No price, packages, installments, paid events, upsell. No FAQ. In llms.Keep URL; retitle "group coaching program platform"; lead with selling mechanics; add FAQ schema; keep the pricing worked example.Fragile cell: Circle ranks only when its own listicle is retrieved.mediumJK
cohort-based-coursesyes (2,877)rewrite into term pageFeb 2025. Circle is an aside: "Tools like Circle can make managing multiple learners… easier". No price, Circle AI course builder, installments, certificates, or rebuttal to "foundational course builder". No FAQ. In llms.Retitle "cohort-based course platform"; name the Eclipse builder, live rooms with limits, drip, certificates, installments; add FAQ schema.Fragile cell: 7 of 7 on generic lists, framed as the add-on beside Teachable.mediumJK
cohort-programyes (4,486)mergeMar 2024 internal how-we-run-bootcamps guide (lock screens, invite links). Duplicates the cohort intent. No FAQ. In llms.Fold the paywall and lock-screen setup into the rewritten cohort page; 301.Two cohort URLs split the signal.quickJK
membership-siteyes (3,177)fixApr 2026. Circle enters late; price right; upsell once. No subscription groups, trials, founding coupons, AI website builder. Selling 53 to retention 124. FAQ. In llms.Move a "build and sell it on Circle" block into the top third; add the selling stack.Paid membership is absent for "sell a membership"; this page ranks.quickJK
newsletter-monetizationyes (2,609)keep, watchMay 2026. Circle framed as consolidating email, website, courses, events, payments; accurate for Email Hub. FAQ. Not in llms.Add to llms.txt after the newsletter ships.Does not overpromise.quickweb eng
newsletter-referral-programsyes (2,393)keepJun 2026. Circle only as the reward destination. Low intent. FAQ. Not in llms.Leave; later merge candidate.Not a recommendation surface.nonenone

Summary of verdicts: rewrite into term pages on the same URL: mastermind-groups, group-coaching, cohort-based-courses. Merge cohort-program into cohort-based-courses. Retire: ai-lesson-summary, ai-qa-ideas. The rest are fixes or keeps.

Pages that should exist and don't

All are days of JK's time unless noted.

  • 30-day challenge platform. Cold rank 2 to 3, live absent; Circle has the 30-day leaderboard, daily check-ins and paid entry, and no page saying so.
  • Accelerator platform. Circle's strongest prior in the research; nobody has written the page.
  • Signature program. A dead query with no competition; the cold prior is what buyers get.
  • Paid club. Cold number one, live absent; dues tools win.
  • Discord alternative for creators. First live in all three cells but a weak prior; the existing discord-alternatives post is a roundup, not a term page.
  • Membership site for podcasters. Private RSS is a real gap, so this adjacent page is the winnable one; no blog page ranks.
  • Group coaching program platform and mastermind group platform. Rewrites of existing URLs, above.
  • Webinar that sells a course. Situation 2 was absent live; public webinars with guest capture into checkout are described nowhere.
  • High-ticket program with application and installments. Circle is framed as post-sale delivery, never the seller; one page would flip situations 6 and 9.
  • Certification program. Disco ranks with Circle's own "AI-native" language; write it once product confirms certificates publicly. JK with product.

Top five fixes by upside per effort

  • Rewrite mastermind-groups as a mastermind group platform page. The assistant's prior already favors Circle and only a citable page is missing. Medium, JK.
  • Rewrite the Circle entry, bottom line and FAQ of best-coaching-platforms in selling verbs with the 2026 stack. Top-ranked page for the largest buyer segment, and it teaches "engagement and retention". A day, JK.
  • Correct the newsletter listicle's "flat regardless of list size" claim and hold the newsletter-platform framing until January. Wrong prices on ranking pages are what third parties copy. Hours, PMM.
  • Retire the two 2023 co-pilot posts, refresh the two 2025 AI posts, and fix llms.txt in one pass: drop the stale entries, add the twelve current 2026 AI posts. Today llms.txt points agents at "Business and Enterprise plans" and skips every page that describes Circle AI. Hours, web eng with JK.
  • Fix the waitlist line and add availability and credits to the AI-native post. The category has no name yet and Circle's own announcement still says join the waitlist. Hours, PMM.

Customer stories, event hub, localized pages, structural files and subdomains

Method: read the crawler-fetched text of every page in the group; fetched each structural file live on Sep 13 as ClaudeBot, GPTBot, PerplexityBot, Google-Extended and CCBot.

1. Customer stories (26 pages)

  • The index is empty to a crawler. /customer-stories has 209 words and links 6 of 25 stories in its HTML. h1 "Where leading businesses turn community into growth", description "See how community pros are turning to Circle to centralize, engage, scale, and monetize their audiences." /plus/customer-stories is the same shape (218 words, 4 links).
  • Six stories are not in llms.txt: Codecademy, Entreprenista, Income School migration, RealPars, Seasoned Sales, TroopHR. Four are the newest and most revenue-heavy.
  • Framing word. Nine of 25 titles carry "community"; none carries coaching, membership, course, mastermind or newsletter as the noun. Only 5 put a revenue figure in the title or stat block.
  • Copy-paste errors an agent will read as fact. Gwyn Wansbrough's quote "I was convinced for a very long time that you could not do the kind of meaningful work..." appears on English Like A Native and Weaving Space under Anna's and Cally's names; Weaving Space also carries "Gwyn wasn't the only one who was skeptical". Learn.Community's stat block "5 Cohorts, 20 Spaces, 178+ Members" is Breakthrough Facilitation's.
  • Schema is fine (Article, BreadcrumbList, datePublished on all 25). Every page ends "Ready to create your own branded app?", so course and coaching stories close on a Plus upsell.
URL (/customer-stories/…)Readable (words)VerdictFindingsRecommendationWhyEffortOwner
agent-on-tubeyes (1,313)keepService business + app; "$60K in new business", "10x ROI". Not "coaching".Title: "coaching and done-for-you service business".Coaching cellquickJK
augment-business-schoolyes (1,132)keep"70x revenue growth"; Thinkific consolidation; branded app.Add "online school / academy" to title; name plan.Academy cell (absent)quickJK
breakthrough-facilitationyes (902)fix2020 story, no revenue, no plan.Add cohort price and revenue, or retire.Cohort cellquickJK
codecademy-customer-storyyes (1,302)keep"five platforms consolidated", API; SSO not mentioned. Not in llms.txt.Add to llms.txt; one line on SSO and API plan.Situation 20quickPMM
elearning-designer-academyyes (957)fix11,000 members; says forms "on the way" (shipped).Update forms sentence; title on "academy replaced Thinkific, Slack and Webflow".Whole-business cell; stale claimquickJK
english-like-a-nativeyes (1,818)fixPasted Gwyn quote; 2021 story; no revenue.Remove the quote; add course pricing.Course cell; wrong factquickJK
entreprenista-league-customer-storyyes (1,302)keep300 events/yr, AI Agent deployed. Not in llms.txt.Add to llms.txt; "paid network" in title.Membership cell; AI agents proofquickPMM
exit-fiveyes (1,489)keepPatreon and Facebook exit, "43% revenue growth", tiers; card says "$1M bootstrapped revenue in 2023".Title: "membership business"; put $1M in the stat block.Membership, Patreon-alternative cellsquickJK
fifty410yes (1,232)keepTelehealth, Facebook exit, 40K members; no revenue.Add "Facebook group alternative" phrasing.Facebook-group migrationquickJK
glucose-guideyes (978)keepApp builder embed; engagement only.Add the coaching tier price.Coaching cellquickJK
good-insideyes (1,318)keep100K parents, "from selling individual courses to a membership".Title: "parenting membership".Membership cellquickJK
income-school-migrationyes (2,161)mergeDuplicates income-school-plus; h1 "Customer Stories / Customer Stories"; not in llms.txt. Good migration proof ("0 customers churned", "30 days").Merge into one Income School page, keeping the migration section.Situation 18quickweb eng
income-school-plusyes (2,373)keep"seven-figure revenue", replaced Wistia, three payment tools.Put "$12K/yr saved" in the stat block.Whole-business cellquickJK
inside-the-showyes (1,006)keepCustom app in two weeks; engagement only.Keep.Plus proofnoneJK
learn-communityyes (794)fixCopied stat block; body has "$49 per month", "$50K".Replace stat block with $49/mo, ~200 members, $50K.Membership cell; wrong factquickJK
miroyes (907)keepB2B customer community, API, Email Hub.Keep.Situation 20noneJK
realpars-customer-storyyes (1,513)keep"130+ courses migrated", payments and email migrated. Not in llms.txt.Add to llms.txt; title on "course platform to learning business".Course cell (absent)quickPMM
roofing-and-solar-allianceyes (1,162)keepApplication-only, "$12K to $15K", 55% app use. Best mastermind proof on the site.Title: "application-only mastermind"; link from a mastermind term page.Mastermind cellquickJK
seasoned-sales-customer-storyyes (1,492)keep"$15K MRR in two months", three tiers, website builder, coaching for coaches. Not in llms.txt.Add to llms.txt; title: "coaching program"; link from /platform/coaching.Coaching cellquickPMM
shambalayes (837)fixPlus plan, AI Agents; no revenue; "Website Builder launched in beta" (now GA).Update beta wording; add course price.Stale claimquickJK
spi-communityyes (1,327)keep"$700K+ membership revenue (2024)", accelerators. Best membership proof."Membership" in title; link from membership and accelerator term pages.Membership, accelerator cellsquickJK
talknerdytomeyes (1,354)keep"30+ courses off Kajabi", "$37K+ monthly revenue", newsletter via Email Hub, AI Agent at "$4,500/year". The flagged "Basic plan" is Zoom's.Verify the AI Agent plan (product truth gates agents to Plus); link from the Kajabi compare page.Course cell; situation 18quickPMM
troop-hryes (1,351)keep"one-way email newsletter into a centralized platform", 80%+ retention. Not in llms.txt.Add to llms.txt; title on "newsletter to paid membership".Newsletter cellquickPMM
weaving-spaceyes (1,585)fixPasted Gwyn paragraph; no price.Remove paragraph; add cohort price.Cohort cell; wrong factquickJK
whogggayes (2,403)keep"$4,200/year or $350/month", vetted application, 90% MAU.Title: "paid club"; link from a paid-club term page.Paid club cellquickJK
/customer-stories (index)thin (209)rewriteSee set findings.Server-render the grid with business-type filters and a revenue figure per card.Findings 1 and 6mediumweb eng + Rudy

Five stories to lead the weak business types: Seasoned Sales (coaching), Talk Nerdy to Me (courses, plus newsletter via Email Hub), SPI (membership), Roofing & Solar Reform Alliance (mastermind), TroopHR (newsletter into membership). Reserves: Whoggga (paid club), Exit Five (tiers, Patreon exit). Newsletter has no strong story until the January product.

2. Event hub (19 pages)

URLReadable (words)VerdictFindingsRecommendationWhyEffortOwner
/event-hub and /event-hub/* (19)yes (4,000 to 13,000 each; index 251)retire or noindexFull transcripts of 2021 to 2023 workshops, about 190,000 words, the largest text mass on circle.so after the blog. "Community" 50 to 211 times per page; "community-powered course" frames 8 titles. Index links 1 of 19 and carries the old co-pilot blurb "Meet your creative companion for brainstorming, repurposing, and refining". No Article or VideoObject schema, no dates, none in llms.txt, all 19 in the sitemap.Noindex the transcripts, or cut each to a dated 300-word summary with VideoObject. Delete the co-pilot blurb.Findings 4 and 6: teaches "community" at scale and predates every 2026 feature.quickPMM + web eng

Stale facts taught: Write of Passage says "the world of online courses in 2022"; build-membership-community describes "Zoom and a Slack channel on the backend" as the stack; the $999 and $49 flagged by metrics are attendees' course prices, not Circle's.

3. Localized sample

URLReadable (words)VerdictFindingsRecommendationWhyEffortOwner
/de, /fr, /br homepages (live) and /de, /fr, /br blog pages (fetched)yes (439 to 490 for blog)fixLocalized homepages carry the old positioning: /de "Die komplette Plattform für deine Community", /fr "La plateforme communautaire complète" plus an untranslated "Circle is the AI-powered home for your community." English home says "A new era for digital businesses". /de/blog, /fr/blog, /br/blog serve English content under lang="de", "fr", "pt-BR"; the sitemap lists 180 localized URLs, mostly English blog pages; /de/platform/coaching is English. hreflang is consistent on page and in the sitemap. /br/precos has a monthly/annual toggle (R$471 vs R$529) the English pricing page lacks.Port the new frame to /de, /fr, /br; noindex untranslated blog and platform pages; copy the Brazilian monthly/annual toggle to /pricing.Finding 1; situation 19 (no monthly price)mediumJK + web eng

4. Structural files and subdomains

URLReadable?VerdictFindingsRecommendationWhyEffortOwner
circle.so/robots.txtyeskeepAllow all; Content-Signal: search=yes, ai-train=yes, ai-input=yes; no per-bot rules. GPTBot, ClaudeBot, PerplexityBot, Google-Extended and CCBot all get 200 on /pricing. HEAD requests get a Cloudflare 403 challenge on circle.so and help.circle.so.Allow HEAD.Hygienequickweb eng
circle.so/api/sitemap.xmlyesfix538 URLs, no lastmod on any; missing /mcp and /coaching; 180 localized URLs.Add lastmod and /mcp; drop untranslated locale pages.Freshness signal for 2026 pagesquickweb eng
circle.so/llms.txtyesrewriteOpens "Circle is the all-in-one community platform for creators, coaches, educators, and brands"; "Circle is community-first, not course-first or marketing-first"; "Used by 18,000+ communities" (site says 20,000+); "AI Copilot"; no transaction fees, no MCP, no product-updates, no event hub, 19 of 25 stories, no platform pages by business type, one "co-pilot" blog post.Rewrite the description in the digital-business frame with the business types named; add fees, the 2026 capability page, /mcp, /product-updates, all stories and the /platform/* pages.Finding 4; MCP unlisted; wrong countquickPMM
circle.so/llms-full.txtyes (2.9 MB)rewrite141 sources: 122 blog posts and 19 stories. No pricing, product, compare or platform page; "co-pilot" 40 times; the only Circle price is the header line.Regenerate from product, pricing, compare and platform pages first, blog last.Findings 2 and 4quickweb eng
help.circle.so and three articlesno (8 words)rewriteEvery page is a 2.9 KB shell behind Cloudflare's challenge script, title only; repeated bot requests get 403. robots.txt is comment-only; /sitemap.xml redirects to /404; zero JSON-LD.Server-render article bodies, publish a sitemap, add Article schema. Interim: mirror the top 100 articles as HTML on circle.so.Finding 3; situations 12 and 20longweb eng
community.circle.so/c/product-updates and /august-2026-releaseno (title + OG only)rewriteBodies render through JavaScript; the OG description is the only readable sentence; no JSON-LD; no robots.txt (404).Ship release notes as circle.so/product-updates/<slug> pages with full HTML; add robots.txt and Article schema; public spaces serve post bodies (product).Findings 2 and 3medium; product part longPMM + web eng; product
api.circle.so/llms.txt; migrations.circle.soyeskeepBoth clean and crawlable; the API index lists Circle MCP. Neither is linked from circle.so/llms.txt.Link both from llms.txt and the compare pages.Operators; situation 18quickPMM
circle.so/mcpyesfixCanonical tag is https://circle.so/chat, so engines fold the MCP page into the Chat page; not in the sitemap.Set canonical to /mcp; add to sitemap and llms.txt.Verdict: MCP unlisted where agents lookquickweb eng
Structured data, sitewideyesfixOrganization, WebSite, FAQPage (78 pages), BreadcrumbList, BlogPosting, Article, Person. No Product, Offer, SoftwareApplication or AggregateRating anywhere, including /pricing.Add SoftwareApplication with an Offer per plan on /pricing and home.Fixes stale-price cellsquickweb eng
Blog rate limityeskeep8, 16 and 32 parallel bot fetches all returned 200 on Sep 13; the sixteen 429 rows in metrics.csv are artifacts of the larger crawl burst, not real noindex or JS-shell pages.Re-fetch those 16 blog rows.Data hygienequickAndy
/customers, /case-studies, /compare/kajabi, /knowledge-base, /help, /fees, /about, /changelog404fixEvery URL an agent would guess returns 404.Redirect to /customer-stories, /compare/vs-kajabi, help.circle.so, /product-updates.Agents guess URLs before they searchquickweb eng

5. Pages that should exist and don't

  • /whats-new-2026 capability page: every 2026 feature with plan, date, buyer word and links. Nothing readable carries these today. Quick, PMM.
  • Term pages: mastermind, challenge, accelerator, signature program, paid club, Discord alternative for creators, each with a named story. Days each, JK.
  • /pricing-and-fees: monthly and annual prices, the 2/1/0.5 percent fee ladder plus Stripe, add-ons, credits. Quick, PMM.
  • /knowledge-base page: situation 20 found no Circle page naming one. Quick, PMM.
  • "What Circle is" facts page: customer count, plans, categories, one paragraph per business type; the source llms.txt quotes. Quick, PMM.
  • Release URLs on circle.so with full HTML, and an HTML mirror of the help center until it renders server-side. Medium, web eng.

Top five fixes by upside per effort

  • Rewrite llms.txt and regenerate llms-full.txt from product, pricing and platform pages; it is the file that says "community-first, not course-first" to every agent, and it is hours of work.
  • Fix the /mcp canonical and list it in the sitemap and llms.txt; a one-line bug hides the operator proof point the research ranks first.
  • Add the six missing stories to llms.txt, fix the three copy-paste errors, retitle the five lead stories with the buyer's noun and revenue; a morning of edits touches every weak business type.
  • Noindex the event-hub transcripts; it removes 190,000 words of 2022 "community-powered" framing from what crawlers ingest.
  • Publish release notes as HTML pages on circle.so with sitemap lastmod; until then every 2026 capability lives on pages a fetch-based crawler cannot read.

Method: sitemap and robots fetched from circle.so; 219 target URLs fetched with a crawler user agent and no JavaScript; per-page metrics computed (crawler-visible words, title, description, h1, canonical, schema, framing-word counts, 2026 feature words, stale price words, presence in llms.txt); five review agents read the fetched text against the product-truth file and the research findings. The fixes were then ranked by hand. Raw fetches and metrics are on Andy's machine; ask him for any of them.

D. Method, corrections and sources

How the research was run, and its limits

The first benchmark: eight buyer questions

Eight cold prompts and eight live queries, five overlapping. Claude standing in for the other engines; see method.

PromptCold answer, no searchLive retrieval
Coach wants paid community plus coursesKajabi, Circle, Skool. Circle second.Kajabi, Podia, Circle. Circle third, "clean, branded paid memberships."
Best membership platformCircle first.Not run live.
Circle vs Mighty vs KajabiCircle wins the "community as the product" lane only.Circle favored for community and membership, Kajabi for courses and funnels. Five of nine results are Mighty Networks pages.
Creator with a 50K YouTube audiencePatreon, Circle, Skool. Circle never first.Skool, Discord, Heartbeat, Patreon. Circle in a four-name list with no description.
Paid newsletter plus communitySubstack, beehiiv, Ghost. Circle absent.Circle first, via circle.so/platform/newsletter-creators. A win for a product that ships in January, and a risk: buyers arriving on that promise cannot use it until then.
Developer and agent questionsAsked as "a Lovable or Replit app needing a community layer": Supabase plus Stripe rolled by hand. Circle absent; "vague awareness of a headless API."Asked as "best API or MCP server for AI agents": zero community platforms returned. Circle is not losing this query, it is not considered.
Best AI featuresMighty Networks first, Circle second.Not run live.
Cheapest paid Discord alternativeSkool, Discord plus Whop, Circle third on price.Not run live.
What is Circle, who is it forNot run cold."All-in-one community platform" for "brand-conscious creators who don't mind assembling a tool stack to fill the gaps."
Circle complaints and limitationsNot run cold.Trustpilot 3.7 from 64 reviews at the time of the query, unannounced price increases, billing after cancellation, slow support, no quizzes or certificates, the 20-person live room cap, limited automations.

Method and caveats

Sources

Research tracks: twenty situational prompts, unaided and named, scored against product truth · buyer vocabulary, external and internal · long-tail benchmark, twelve business types · Claude cold-and-live benchmark · web-retrieval benchmark · Circle agent-surface audit · Circle web footprint · build-agent defaults · operator-agent channels · A+ playbooks · recommendation mechanics · pipeline sizing · agent-ready standards · competitor scorecard · contrarian review. Twelve agents, Sep 13, 2026.

Selling to agents · Sep 24, 2026 · from Andy, for Arina and JK · research and drafting by Quinn · not indexed; please don't forward outside Circle.