Home / Blog / aisearchradar.io Report: False Positives on Our Site
Site Audit

aisearchradar.io Report: False Positives on Our Site

13 October 2026 · AI Web Page SEO
aisearchradar.io scored aiwebpageseo.com 81 out of 100. Checked against the live site, many of its recommendations asked for things already there, including ai.txt, Person schema, sameAs links, knowsAbout and a last-updated date. It reported JSON-LD syntax errors without naming any, said a definition was explained late when it is in the main section, recommended HowTo schema for a rich result Google stopped showing in 2023, projected score gains past 100, and offered rewrites with a wrong price.

What happened

On 13 October 2026 we ran the AI visibility audit from aisearchradar.io on aiwebpageseo.com. It gave the site a technical score of 81 out of 100, split into four pillars: Entity Analysis 90, Content Readiness 85, EEAT Signals 71 and Tech Foundation 77. A second part of the same product, a content analyser, scored our home page and our Real Traffic Audit page 76 each.

The summary listed 12 critical items to "fix these first" and 18 that "need work". Most of the detail behind those counts sat behind "Upgrade to see all audit details" and "View Plans" buttons, so what follows covers every recommendation the report actually showed us.

Scores like these are only worth something if the findings behind them are true. So we checked each recommendation against the live site and the published standards. Many of them describe things the site already has. Several of the suggested rewrites would add claims to our pages that are not true. This post goes through them one by one, so you know what to look for when a report like this lands on your desk.

The short version

Finding 1: "Deploy an ai.txt file at the root"

What the report said. Listed as a high priority under Technical Infrastructure.

What the site has. An ai.txt file at https://aiwebpageseo.com/ai.txt, last updated on 12 October 2026, the day before the report. Some tool pages, such as the Security Audit and the Accessibility Audit, have their own ai.txt as well.

Why it matters. A finding that a file is missing is easy to check: open the address. If a tool reports a missing file that loads in a browser, it either did not request it or did not read the answer.

Finding 2: Person schema and an author bio

What the report said. "Implement Person schema for the primary author" and "Create an Author Bio section for John Carlsson", both high priority, with the reason that language models weight content higher when it can be tied to a Person.

What the site has. The home page's structured data includes Person markup for John Carlsson with his job title and 18 profile links in sameAs, and the home page links to the About page. Our demo report pages carry a byline that links there too. The point about attributing content to a named person is sound. The site already does it.

Finding 3: sameAs, social profiles and knowsAbout

What the report said. Three separate recommendations: link the Organization to social profiles, expand its sameAs links, and add knowsAbout.

What the site has. The Organization node lists 18 sameAs profiles, from LinkedIn, X and YouTube to Trustpilot and business directories, and it has a knowsAbout list. Three recommendations for one thing that already exists inflates the "Needs Work" count without anything to do.

Finding 4: "Lack of a clear Last Modified date"

What the report said. The index page has no clear last-modified date, which makes freshness harder for crawlers to judge.

What the site has. The home page footer shows the current version and "Last updated: 12 October 2026", and the WebPage markup carries dateModified. The report also calls the home page a "Learning Hub". The Learning Hub is a different page.

Finding 5: "Fix JSON-LD syntax errors"

What the report said. A high-priority item with no page, no line and no error message.

What we checked. Every JSON-LD block on the home page parses as valid JSON. A syntax error is a precise, checkable fact. A report that claims one should say where it is.

Finding 6: the praise was wrong too

Getting the good points wrong matters as much as the bad ones, because it shows how the page was read.

Finding 7: HowTo schema

What the report said. Convert "the fix, step by step" sections into HowTo schema to appear in step-by-step AI results.

What Google says. In August 2023 Google announced that HowTo rich results would no longer be shown, first on mobile and then on desktop. The markup is still valid schema.org, and our guide pages already include it, but it no longer earns a rich result in Google. A recommendation that promises otherwise is out of date.

Finding 8: "Verify agents.json follows the MCP standard"

The report does not say which file it means or what is wrong with it. The Model Context Protocol specification does not define a site file called agents.json, so there is no MCP standard for such a file to follow. A recommendation should name the file, the rule and the failure.

Finding 9: rewrites that would add errors

The content analyser offered ready-made text to paste in. Some of it is wrong about our own products:

Language models quote what is on your page. If a tool's suggested rewrite puts a wrong price or an invented policy on it, that is what gets quoted. Check every number and every claim in suggested text before you use it.

Finding 10: "Ambiguous pronoun usage"

aisearchradar.io marked both pages down for using "it" instead of repeating the tool's name. That is a style preference. Clear writing helps readers, but repeating a product name in every sentence is not a published ranking factor for search engines or AI answer engines.

Finding 11: "Delayed explanation of confirmed vs likely metrics"

What the report said. The difference between "likely" and "confirmed" visitors on the Real Traffic Audit page "is crucial but explained late in the FAQ section".

What the page does. Both are defined in the page's main section, "How the audit decides who is real", which comes well before the questions and answers: likely people are those whose browser loaded a page and its images, styles or scripts without hammering the site, and confirmed people also submitted a form or used a feature. The FAQ repeats the explanation as question 4 for people who skip straight to the answers. The report found the second mention and missed the first.

Finding 12: "Missing introductory paragraph" on the home page

What the report said. The home page "lacks a clear, high-level introductory paragraph", and as evidence it quoted: "SEO, Security, Accessibility & Schema Audits | aiwebpageseo Skip to main content Independent security checks for aiwebpageseo.com".

What that quote is. It is the page title from the browser tab, the "Skip to main content" link that keyboard and screen reader users need, and the heading of the "Independent security checks" badge strip near the top of the page. None of that is the introduction. The opening paragraph sits directly under the H1 and says what the site does: it checks a site for security, trust, Google spam policies, HTTP errors and accessibility, debugs and generates schema, and shows how people, Google and AI systems read your business.

A tool that reads navigation and badge text as the opening of the content will judge any well-built page the same way, because a skip link at the top is exactly what accessibility guidelines ask for.

Finding 13: "Repetitive content blocks" on the home page

What the report said. The introduction says the audits were "tested on hundreds of live websites" twice, which it called semantic noise.

What the page does. Every section on the home page has a switch between a Plain English and a Technical explanation. A visitor sees one or the other, never both. The two versions say the same thing for two audiences, one in everyday words and one naming the standards, so some facts appear in each. That is the purpose of the switch, not an accident of editing.

Finding 14: score gains that add up to more than 100

For the Real Traffic Audit page, the report said: "Fixing the top issues could improve this page by +24–42 pts". The page scored 76, so the top of that range is 118 out of 100. For the home page it promised +29 to 45 points on a score of 76: up to 121 out of 100. A projected gain that cannot fit on the scale tells you the projection is not calculated from the score.

Finding 15: a suggested definition that adds a legal claim

For the Real Traffic Audit, the report offered a ready-made opening definition. It ends by saying the audit gives an accurate count of human visitors "without compromising GDPR compliance".

What the page actually says is narrower and more careful: access logs contain visitors' IP addresses, which are personal data under UK and EU GDPR, and the audit runs in your browser so the log is never uploaded. Whether any business is compliant with GDPR depends on far more than one tool. A sentence that turns a factual privacy design into a compliance promise is the kind of claim a site owner should never paste in without legal advice.

How to check a report like this yourself

Sources

Frequently asked questions

How can I tell if an AI visibility audit report is accurate?

Check each finding against the live page. If it says a file is missing, open the address in a browser. If it says schema is missing, view the page source and search for the type, such as Person or sameAs, or use a structured data validator. A real syntax error has a page and a line. A report that cannot point to where a problem is has not shown that the problem exists.

Is HowTo schema still worth adding for Google?

Not for rich results. In August 2023 Google announced it would stop showing HowTo rich results, first on mobile and then on desktop. The markup is still valid schema.org and does no harm, and other systems may read it, but a recommendation that promises HowTo rich results in Google Search is out of date.

Should I paste the rewrites an AI audit tool suggests?

Only after checking every fact in them. On our site, the suggested comparison table listed a paid audit as free, and a suggested FAQ made privacy promises on our behalf. AI answer engines quote what is on your page, so a wrong price or an invented policy statement in a pasted rewrite becomes what gets quoted about you.

Does Person schema help AI search?

Tying content to a named person with Person markup, a job title and sameAs links to real profiles is sound practice, because it gives search engines and AI systems a clear author entity to connect your content to. Check whether your site already has it before acting on a report that says it is missing: ours did.

💬 Got a problem?

aiwebpageseo.com is a data-driven SEO and AEO platform providing a free suite of technical website tools.