Blog · 24 September 2026 · By HeardOf

574 posts: what the AI-visibility category writes about itself

How many posts have the AI-visibility vendors published?

574 across five vendor blogs on 23 September 2026: Profound 246, Otterly 180, AthenaHQ 63, Peec 44, Geoptie 41. A post is a URL the site's own sitemap lists under its blog or articles path with exactly one path segment after it, so index pages and page-2 URLs are out, and Profound's one markdown twin of an HTML post is counted once.

Four of the five are among the nineteen tools whose prices we read in what nineteen AI-visibility tools charge; Geoptie is not in that set and is here because it publishes a blog we could count the same way. The rest of each sitemap is worth a column, because it says what else the site has decided to index.

How often do AI-visibility tools update their blog posts?

Fourteen of 155, on the two sites whose dates can be read post by post. The event is exact: a post published before 1 June 2026 whose modification date is on or after 1 June 2026 — Otterly 8 of its 122 such posts, Geoptie 6 of its 33, read 23 September 2026 from Otterly's WordPress API and Geoptie's structured data.

Earlier edits exist and are a different count. Otterly's API shows 77 of its 180 posts modified 30 days or more after publishing, 74 of them posts from before June; another 77 were modified within a day of publishing. Geoptie's structured data shows 9 of 41 modified 30 days or more after publishing and 21 within a day. So the sites do touch old posts; what the 14 of 155 measures is how many posts older than June were touched in the sixteen weeks since.

The other three sites are excluded because their modification dates measure the site, not the post. Peec's 44 posts all carry a modification date of 21 September 2026 at 08:33:46 UTC, to the second, in their structured data. AthenaHQ's sitemap gives all 227 of its URLs one timestamp, 22 September 2026 at 21:50 UTC, and on the 20 posts whose structured data carries a modification date it equals the publication date. Profound's posts carry no modification date at all, and the dates in its sitemap cluster: of its 245 dated post URLs, 22 share 4 September 2026, 18 share 9 April and 18 share 13 April.

The shape is stacking. Profound publishes 9 to 25 a month, Otterly 4 to 22, Peec 2 to 7, AthenaHQ 0 to 10, Geoptie 15 in May and 0 to 3 in the other months. None of the five dates a post as re-run; the two that expose edits at all expose them as edits.

Do AI-visibility vendors name their competitors in their posts?

Geoptie does in 22 of its 41 posts, 54%, when the names looked for are the other four vendors in this census; AthenaHQ in 12 of 63, Profound in 15 of 246, Otterly in 4 of 180, Peec in none of 44 — in the title or the body text, read 23 September 2026. Against a wider list, the nineteen tools whose prices we read plus Geoptie, the counts rise to 28 of 41 at Geoptie, 20 of 63 at AthenaHQ, 40 of 246 at Profound, 38 of 180 at Otterly and 8 of 44 at Peec.

The count depends on three rules, so here they are. The text searched is the title plus the post's main or article element with navigation, header, footer, aside, scripts and styles removed, and then any passage of ten or more words that repeats on at least 90% of that site's post pages removed too: Peec's pages carried Profound, Semrush and Ahrefs on all 44 before that step, in comparison links its markup does not mark as navigation. A name counts as a whole word spelled with the vendor's own capitals, and no domain forms; and five names that are also English words — Profound, Goodie, Bluefish, Surfer, Conductor — count only when they do not begin a sentence or a heading; on this corpus that changes no count. The site's own name never counts.

Semrush, Ahrefs, HubSpot and Similarweb are named for many reasons besides their AI-visibility products — Otterly announced an app in the Semrush App Center in January 2025 — so the wider column counts a name, not a rivalry. The best-of lists of AI-visibility tools that the rule finds on these blogs are counted separately: of the 14 that name two or more tools, all 14 put the author first, and 30 of the 179 prices they print for other tracked tools are not on that tool's own pricing page as rendered on 24 September 2026, in who comes first in the vendors' own best-of lists.

What does a templated industry page look like in this category?

AthenaHQ's sitemap lists 83 URLs under /industry, and 50 of its 70 use-case pages open with the same H1 and carry the same three H2s in the same order, read 23 September 2026. The 83 are one index, 12 vertical pages and 70 use-case pages — 2 to 9 under each of 11 verticals, and none under the twelfth, agencies.

The 50 begin "AthenaHQ Is Best for [use case] in [vertical]" — or a variant naming no vertical on the five multibrand pages — and then run "Purpose-built for [use case] in [vertical]", "Instant ROI on the Metrics that Matter", "How [vertical] Teams Use AthenaHQ". The other 20 use-case pages carry a heading of their own, such as "Fill the AI Shopping Cart First" or "Be the University AI Recommends", and not that set; the 12 vertical pages and the index carry neither. Open any two of the 50 and the slots are where the use-case name goes.

Whether those pages rank for anything is not something this post can say. We have no search-results record on file that would support a figure either way, so there is none here.

What shape do the vendors' posts take?

Geoptie's posts are the longest, a median 2,910 words, and carry a table in 27 of 41; Profound's are the shortest, a median 920 words. Otterly has a question-shaped subheading in 103 of its 180 posts, 57%, and Geoptie in 15 of 41, all read 23 September 2026.

Two things the table does not say. Profound's question-shaped subheadings are mostly H3s — on H2s alone it would show 20 questions, not 223 — so counting H2 and H3 together is what makes the column comparable. And a slug is not a reading: the six Reddit posts below include five with the vendor's own data, and only one of them carries a study-shaped slug. The 61 study-shaped slugs were opened and rendered on 24 September 2026: 33 report a measurement of their own, 31 of the 61 state a sample size, 21 name a month, none of the 33 prints or links its full prompt set, and none of the 61 links raw data, in which AI-visibility studies you could check.

What does every vendor write about, and what does almost nobody?

Fourteen titles across all five sites contain complete, ultimate, definitive or comprehensive followed later by guide, and seven titles on four of the five put vs, vs. or versus directly between two of GEO, SEO and AEO, with no word in between, counted from the titles as served on 23 September 2026. The guide count is Otterly 6, AthenaHQ 4, Peec 2, Geoptie 1, Profound 1; the versus count is AthenaHQ 2, Profound 2, Geoptie 2, Peec 1, and none at Otterly, whose "Is GEO Replacing SEO? What 12 Months of Experiments Actually Say" has no vs in it.

Reddit is where the category's own posts disagree with each other. Six posts carry Reddit in the title or slug, and five report the vendor's own data:

Our own count sits beside them, not above them: in our run of 4 September 2026, 19 of 564 citations pointed at reddit.com, 3.4%, every one of them from Perplexity and none from ChatGPT, in who the AI actually cites. Four posts from three vendors carry llms.txt in their slug — Otterly two, Profound one, Peec one — and what those three vendors' own sites serve, one at its root and all three on a documentation host as of 23 September 2026, is in does llms.txt actually work.

Almost nobody, by title. Five titles contain stable, stability, variance, variability, volatility, consistent, consistency, reliable or fluctuate, or the phrases the same question, different answers or keep changing: Otterly's "We Measured How Stable AI Citations and Brand Mentions Really Are (and Why 10 Prompts Isn't Enough)" of 11 September 2026 and its "Why does ChatGPT provide different answers to the same question?" of May 2025, Profound's "AI Search Volatility: Why AI search results keep changing" of July 2025 and its "Variability of Google models: Gemini vs AIO vs AI Mode" of July 2026, and one AthenaHQ title that matches on the word "reliable" and is about traffic data. Seven titles contain revenue, attribution, ROI or pipeline, two of them Profound integration or partnership announcements. No title on any of the five sites contains founder, CEO, CFO, lawyer or product manager. A title is not a reading, and these three counts are titles.

What have the vendors done to their own sites?

Peec's sitemap lists each of its 44 posts twice, AthenaHQ's robots.txt carries a marketing sentence, and Profound's sitemap lists a markdown file, all read 23 September 2026. Each reproduces with one fetch, and we have not asked any vendor about them.

- Peec's sitemap lists its 44 posts under /blog/ and again under /customer-io-test-site/, the same 44 slugs. The pair we opened both answer 200, and the mirror copy carries a canonical tag pointing at its /blog/ address, so a search engine is told which copy counts; the sitemap lists both, 205 URLs in all.

- AthenaHQ's robots.txt, 270 bytes, allows everything, allows /llms.txt, names its sitemap, and then carries an ASCII-art robot and the sentence "AthenaHQ is the leading Generative Engine Optimization platform...". Whether any crawler reads past the directives, we cannot say.

- AthenaHQ has four articles whose URL ends in november-2026, each carrying a published date of 15 February 2026 in the page's structured data. A reader dating the post from its URL and one dating it from its structured data get different answers; both are AthenaHQ's.

- Profound's sitemap lists a markdown file, /blog/ai-shopping-journey-2025.md, beside the HTML post of the same name; it answers 200 as text/markdown at 8,667 bytes, and is the only .md entry among the 635.

- Peec serves a page at /ai-instructions, 200, that asks an AI system answering a question about Peec to "treat this page as a canonical source". We read it as data. Which other vendors serve one, and what each site serves for llms.txt, is in the llms.txt post linked above.

How was this measured?

Everything above comes from one pass on 23 September 2026 with curl and a browser user-agent, one fetch per URL, no JavaScript executed: the five sitemaps and their children, the 574 post pages those sitemaps list, the 83 AthenaHQ industry pages, Otterly's WordPress posts API, AthenaHQ's robots.txt, one Peec mirror page with its original, and Profound's markdown file. The post rule, the text region, the strips, the name rules, the heading rules and the slug patterns are printed in the section that uses them, so a vendor or a journalist can re-run any count without asking us how.

Dates come from each page: structured data where a post has it, Profound's time element, and the on-page date text on 43 AthenaHQ posts that carry nothing else. Otterly's structured data agrees with its API on all 180 posts to the day. RSS would have been simpler. On 23 September 2026 we tried rss.xml, feed, blog/rss.xml and blog/feed on each of the five sites: Profound, AthenaHQ and Peec answered 404 on all four, and Geoptie and Otterly answered 200 on blog/rss.xml and blog/feed with an HTML page rather than a feed, so no feed was used.

A first count on 19 September 2026 gave 568 posts and different figures for several rows. None of it is printed, because the rules behind it were not written down well enough to re-run; the figures here replace it rather than update it. Everything this post says about our own system comes from us and carries its date in the sentence that says it; you cannot check it. Everything else carries a site, a day and a rule. The audit this site sells is described in the post on which engines we check, as of 23 September 2026, and that description is the source of the one sentence about it in the closing.

The vendors' own "best AI visibility tools" lists: 14 of 14 put the author first