Does llms.txt improve AI visibility?

No. Ahrefs examined the server logs of 137,210 domains and found that 97% of the llms.txt files published received no request at all in a month. Of the requests that did arrive, most came from bots that are not AI search engines, and Slackbot fetched the file more often than PerplexityBot did. So far, AI engines are not fetching it. GetMOW does not sell it, and this site does not publish one.

Does schema.org markup help AI visibility?

Not measurably. Google’s own documentation says no special schema.org structured data is needed to appear in AI Overviews or AI Mode, and Ahrefs’ controlled test of 1,885 pages that added schema found no meaningful rise in AI citations on any platform. Schema still earns rich results in ordinary search, which is what it is for, so GetMOW keeps it as search hygiene and never bills it as AI visibility work.

Is AI visibility just SEO with new vocabulary?

Partly, and the overlap is the floor rather than the ceiling. Ranking in ordinary search still supplies the largest single share of AI Overview citations, so on-page SEO is not obsolete. What is new is that the pages deciding an AI answer are mostly third-party pages, that engines fan one question out into many, and that being mentioned on independent pages correlates with AI visibility three times as strongly as backlinks do. None of that work sits inside an SEO retainer.

Do brand mentions matter more than backlinks for AI visibility?

Yes, by the largest margin in the published data. Across 75,000 brands, Ahrefs found branded web mentions correlate with AI Overview visibility at 0.664 and backlinks at 0.218. Brands in the top quarter for web mentions drew 169 AI Overview mentions; the next quarter drew 14. A mention on a page the engines already read is worth more than a link from one they do not.

The evidence

Most of it is being present
where the engines read.
Very little is being liked.

Nearly everything the evidence says moves AI visibility is about being present in the third-party pages engines draw on. A smaller part is being readable by a crawler at all. The smallest part is what an answer says about you once you are in it. GetMOW’s work is weighted the same way, and the tactics below that the evidence says do nothing, or worse, are not sold at any price.

Reviewed 10 September 2026

Three layers, in order

Readable comes first. A crawler has to reach a page, read it without running scripts, and find passages that make sense on their own. None of the major AI crawlers execute JavaScript, so text that only exists after a script runs does not exist for them.

Present is where the weight is: the roundups, comparison pages, review platforms, directories and trade press an engine actually draws on when a buyer asks who to consider. A company that is not in those pages is not in the answer, however good its own site is.

Preferred is what the answer says once the company is in it: the tone, whether it reads as a recommendation, and whether the facts are right. It is the layer everyone wants to buy first and the one the evidence says to work on last.

What has evidence behind it

Each of these is a kind of work with a published study behind it, linked at the foot of the page. The list is shorter than the list of things sold under the name AI visibility, which is the point of publishing it.

Ranking in ordinary search: Pages in Google’s top ten still supply the largest single share of AI Overview citations, 38% in Ahrefs’ latest read, down from 76% a year earlier. Ordinary on-page work is not obsolete. It is the floor.
Being in third-party listicles and comparisons: Across 1,056,727 citations studied by Wix, listicles took 40.9% of citations on commercial questions, and third-party listicles out-cited brand-owned ones 80.9% to 19.1%. Getting into the page that already ranks beats writing another one.
Covering the sub-questions: Google states that AI Overviews and AI Mode may issue multiple related searches across subtopics to build one answer. A page that answers only the head question misses most of what the engine goes looking for.
Being retrievable at all: Vercel’s crawler data shows none of the major AI crawlers render JavaScript, and OpenAI states that a site opted out of OAI-SearchBot will not be shown in ChatGPT search answers. A blocked or script-only site is invisible before any other work starts.
Being mentioned on independent pages: Across 75,000 brands, Ahrefs found branded web mentions correlate with AI Overview visibility at 0.664, three times the 0.218 for backlinks. Brands in the top quarter for mentions drew 169 AI Overview mentions to the next quarter’s 14.
Substantive, dated freshness: Across 16.975 million cited URLs, AI citations ran 25.7% fresher than organic results, and ChatGPT cited pages 458 days newer. A changed date over unchanged text is not freshness, and this site does not do it.
Quotations, statistics and cited sources in the text: In the GEO benchmark of 10,000 queries, adding quotations lifted position-adjusted visibility by about 43%, statistics by about 33% and cited sources by about 28%. Concrete claims with named sources are what an engine can lift.

What GetMOW refuses to sell

These four are sold widely. The evidence says they do nothing for AI visibility, or worse, so GetMOW does not sell them, and this site carries none of the first.

llms.txt: Ahrefs read the server logs of 137,210 domains: 97% of the llms.txt files published received no request at all in a month, and Slackbot fetched the file more often than PerplexityBot did. This site does not publish one.
schema.org markup as an AI lever: Google states that no special schema.org structured data is needed to appear in AI features. Ahrefs tracked 1,885 pages that added it against 4,000 controls and found no meaningful rise in AI citations on any platform. Schema is ordinary search hygiene, and it is kept and labelled as that here.
Keyword stuffing: In the GEO benchmark, keyword stuffing cut position-adjusted visibility by about 9%. It is the one tactic on this page the evidence says actively hurts.
Core Web Vitals beyond a baseline pass: Across 107,352 pages visible in AI answers, correlations between Core Web Vitals and AI visibility ran between -0.05 and -0.18: a constraint, not a lever. A severe failure hurts; a faster pass does not help.

Questions about the evidence

Does llms.txt improve AI visibility?: No. Ahrefs examined the server logs of 137,210 domains and found that 97% of the llms.txt files published received no request at all in a month. Of the requests that did arrive, most came from bots that are not AI search engines, and Slackbot fetched the file more often than PerplexityBot did. So far, AI engines are not fetching it. GetMOW does not sell it, and this site does not publish one.
Does schema.org markup help AI visibility?: Not measurably. Google’s own documentation says no special schema.org structured data is needed to appear in AI Overviews or AI Mode, and Ahrefs’ controlled test of 1,885 pages that added schema found no meaningful rise in AI citations on any platform. Schema still earns rich results in ordinary search, which is what it is for, so GetMOW keeps it as search hygiene and never bills it as AI visibility work.
Is AI visibility just SEO with new vocabulary?: Partly, and the overlap is the floor rather than the ceiling. Ranking in ordinary search still supplies the largest single share of AI Overview citations, so on-page SEO is not obsolete. What is new is that the pages deciding an AI answer are mostly third-party pages, that engines fan one question out into many, and that being mentioned on independent pages correlates with AI visibility three times as strongly as backlinks do. None of that work sits inside an SEO retainer.
Do brand mentions matter more than backlinks for AI visibility?: Yes, by the largest margin in the published data. Across 75,000 brands, Ahrefs found branded web mentions correlate with AI Overview visibility at 0.664 and backlinks at 0.218. Brands in the top quarter for web mentions drew 169 AI Overview mentions; the next quarter drew 14. A mention on a page the engines already read is worth more than a link from one they do not.

Sources

Every statistic on this page was read from these pages in September 2026.

Start with what the answers say about you.

Before you weigh any method, including this one, see the answer ChatGPT, Gemini and Perplexity give today when a buyer asks who to consider in your category. Free, written by hand, and yours whatever you do next.

Get your category report

Most of it is being presentwhere the engines read.Very little is being liked.