We ran 13 technical checks on 20 AI visibility tools' own websites.
Every AI visibility tool sells AI readiness. We ran the Scout technical check, the same 13 weighted technical checks any site gets, on the homepages of 20 of them, including our own. Four pass everything. Three of the four lowest scores belong to the tools AI engines name most often. Here is the whole table, with the raw check detail, so you can re-run it.
Scout technical check score, what is not passing, and how often AI engines named the tool.
Score: Scout technical check, 13 weighted technical checks on the homepage, 5 Oct 2026, HTTP only, no model involved. Named: how many of 20 AI answers to five "best AI visibility tools" questions named the tool, AI answer check of 29 Sep 2026 (ChatGPT, Gemini, Claude, Perplexity, five questions each). Two instruments, two columns, never one number.
| Tool | Technical check (13 checks) | Not passing (of 13) | Named in AI answers |
|---|---|---|---|
| Frase frase.io | 100 | all 13 pass | 0 of 20 |
| Instly Scout instlyscout.com | 100 | all 13 pass | 0 of 20 |
| Rankability rankability.com | 100 | all 13 pass | 0 of 20 |
| Semrush semrush.com | 100 | all 13 pass | 4 of 20 |
| HubSpot hubspot.com | 97 | description (warn) | 3 of 20 |
| Writesonic writesonic.com | 97 | description (warn) | 0 of 20 |
| Conductor conductor.com | 95 | alt text (fail) | 0 of 20 |
| Evertune evertune.ai | 93 | description (fail) | 1 of 20 |
| Geoptie geoptie.com | 90 | llms.txt (fail) | 1 of 20 |
| MarketMuse marketmuse.com | 90 | llms.txt (fail) | 0 of 20 |
| Yext (Scout) yext.com | 90 | llms.txt (fail) | 0 of 20 |
| AirOps airops.com | 89 | llms.txt (warn), description (fail) | 0 of 20 |
| Ahrefs ahrefs.com | 88 | llms.txt (fail), alt text (warn) | 3 of 20 |
| AthenaHQ athenahq.ai | 88 | schema (fail) | 2 of 20 |
| BrightEdge brightedge.com | 80 | schema (fail), description (warn), alt text (fail) | 0 of 20 |
| Surfer surferseo.com | 75 | llms.txt (fail), title (warn), description (fail), alt text (fail) | 0 of 20 |
| Profound tryprofound.com | 74 | schema (fail), llms.txt (fail), description (warn) | 4 of 20 |
| Scrunch AI scrunch.com | 73 | schema (fail), Canonical URL (fail), title (warn), H1 (warn), alt text (warn) | 4 of 20 |
| Peec AI peec.ai | 70 | schema (fail), llms.txt (fail), description (warn), alt text (fail) | 4 of 20 |
| Otterly AI otterly.ai | not measured | its site refused our checker (HTTP 403); not scored, not a failure | 4 of 20 |
Weights: indexing allowed 20, JSON-LD schema 18, llms.txt 15, AI-crawler access 15, non-browser client access 12, sitemap 10, canonical 10, title 10, description 10, H1 8, bare domain 8, image alt text 7, HTTPS 5. A warn scores half. Findings that carry no points (brand missing from the opening text, orphan pages in the sitemap) are listed below the table, not in the score. Instly Scout's own row carries one such finding; we fixed it the same day and left the reading as measured.
The tools AI names most are among the least AI-ready, by their own category's checks.
Named 4 of 20, scored 74, 73 and 70
Profound (74): no JSON-LD schema, no llms.txt, meta description 177 characters, brand not in the opening text. Scrunch AI (73): no schema, no canonical tag, two H1s, title 83 characters. Peec AI (70): no schema, no llms.txt, alt text on 0 of 210 images, brand not in the opening text.
All three were named in 4 of 20 answers on 29 Sep. Semrush, also 4 of 20, scores 100: the exception that shows the checks and the naming are different things.
What that does and does not mean
It does not mean these are bad tools. These are homepage checks, one reading, one day. Profound is cited by AI engines more than anyone in this category and its product is not what we measured. It means the technical layer is the floor: failing it did not stop them being named, and passing it (Frase, Rankability, Instly Scout, all 100) did not get any of us named more than once.
What gets a tool named is being in the pages the engines read for these questions, which are round-ups and comparison articles. We measured that too: who AI names for AI visibility tools, and where it reads.
Two things the score does not count, but engines notice.
Brand missing from the opening text
Profound, Peec AI, AthenaHQ, MarketMuse, Rankability and, on the day of the check, Instly Scout: the brand name appears in the browser tab but not in the headline or first paragraph. Engines quote the page body, not the tab. On ipmicompare.com, five days after fixing exactly this, Gemini named the site in words that paraphrase its new first sentence, and ChatGPT listed it as a source, which on its own does not count as naming (one reading pair; case study).
Orphan pages in the sitemap
Pages listed in the sitemap that no other page links to, so crawlers may never fetch them. Found on 15 of the 19 measured sites, from 1 of 60 sampled (Frase) to 57 of 60 (HubSpot). Only the first 60 sitemap URLs are sampled, so the counts are a floor, not a total.
Full per-check detail for every site, including status codes and URLs tried, is in the dated record behind this page. Otterly AI's site refused our checker (HTTP 403), so it is listed as not measured: a refusal is not a failed check, and we do not score what we could not read.
Free: the homepage x-ray shows the surface checks at once, and the emailed copy and the Chrome extension carry an AI readiness score built on the same 13 checks as this table. Then the AI answer check tells you whether any of it matters yet.
Method: Scout technical check (diagnose engine, 13 weighted checks, deterministic, HTTP only, crawler and browser Accept headers both tried, a 403 or timeout recorded as not measured, never as a failure). Run 5 Oct 2026 from Bangkok. Named-in-answers counts from the Instly AI answer check of 29 Sep 2026 on the five buyer questions for "AI visibility tools". Instly Scout is one of the 20 and is measured with the same code as the rest.