Blog

Why Isn't My Site Cited in AI Overviews? 6 Real Causes

← All posts
TL;DR
  • Google says a page needs to be indexed and eligible to show with a snippet to be cited, with no extra requirements. Most misses come from six causes: eligibility, a query that shows no overview, a buried answer, a sub-question another page answers better, no visible proof, or source types that win the slot.
  • Three causes are directly fixable, two only partly, and one is not a problem at all.
  • Ranking in the top 10 does not guarantee a citation, and schema markup has not been shown to help.

Most sites miss an AI Overview for one of six reasons: the page isn't eligible, the query never triggers an overview, the answer is buried, another page covers a sub-question better, the page shows no proof, or other source types win. Three of these you can fix directly.

Key takeaways

  • Google says there are no extra requirements for AI Overviews. A page must be indexed and eligible to show with a snippet, and nothing more (Google Search Central, checked 27 Oct 2026).
  • Ranking in the top 10 does not guarantee a citation. Ahrefs found about 38% of cited pages ranked in the top 10 in March 2026. In its July 2025 study the figure was about 76% (via Search Engine Journal, checked 27 Oct 2026).
  • Google also says AI Overviews often don't trigger at all. Not being cited on a query with no overview is not a failure.
  • Adding schema markup has not been shown to help. Ahrefs found no clear effect on citations (details below).

Quick recap: the six causes at a glance

#CauseCheck it yourselfFixable?
1Page isn't eligibleURL Inspection, robots.txt, meta tagsYes
2Query shows no overviewSearch it in a private windowNot a problem
3Answer is buriedRead your first 100 wordsYes
4Another page answers a sub-question betterList follow-up questions, search eachPartly
5No visible proof on the pageLook for author, source, original dataYes, slowly
6Other source types winOpen the overview and read who is citedPartly

You ranked well, searched your own keyword and found the overview citing someone else. That is a normal and annoying moment, and it rarely means something is broken. Below is each cause, a way to confirm it and what to do.

How does Google pick pages for AI Overviews?

Google doesn't publish a ranking recipe. Its guide describes two things. First, the eligibility gate: the page must be indexed and snippet-eligible. Second, a technique called "query fan-out," where the system runs several related searches across subtopics to build its answer. That is why the cited pages often differ from the ones ranking for your exact keyword (Google Search Central, checked 27 Oct 2026).

Studies disagree on how much ranking matters. Ahrefs reported about 76% of cited pages in the top 10 in July 2025, then about 38% in March 2026 on a much larger sample. BrightEdge reported about 17% in February 2026 using a different method (via Search Engine Journal, checked 27 Oct 2026). Don't average these numbers. The safe reading is that ranking helps but doesn't decide.

Cause 1: Is your page eligible to be cited at all?

Fixable: yes

Google's documentation sets two conditions: the page is indexed and it can show with a snippet. A page that is blocked, set to noindex or set to nosnippet fails before quality is even considered. The same guide lists the basics: allow crawling in robots.txt and in any CDN or host rules, link to the page internally, and keep important content as text.

Check it yourself: Paste the URL into Search Console's URL Inspection tool and confirm it says the page is indexed. View the HTML Googlebot received and look for noindex or nosnippet. Then read your robots.txt and any bot-blocking rules at your host or CDN.

Cause 2: Does the query even show an AI Overview?

Fixable: not a problem

Google says overviews appear only when they add something beyond classic results, so they often don't trigger. One Search Engine Land contributor notes that transactional, navigational, branded and local searches are less likely to show one. If your best keywords are commercial, the missing citation may just mean there was nothing to be cited in.

Check it yourself: Search the exact query in a private window, a few times on different days. Overviews come and go. Note whether one appears at all before diagnosing anything else.

Cause 3: Is your answer buried?

Fixable: yes

One analyst argues that systems pulling answers from pages favour a clean, early, self-contained answer over a long warm-up. Whether or not that matches Google's internals, it describes a common pattern: a page about "project management software" that opens with features won't be pulled for "how do I manage a remote team's workload?"

Check it yourself: Read only your first 100 words and ask whether they fully answer the title question. Then check each H2: does the section under it make sense alone? Rewrite the opening and the weakest sections first.

Cause 4: Does another page answer a sub-question better?

Fixable: partly

Because of fan-out, the overview is assembled from several narrower searches. Your page can rank for the main keyword and still lose every sub-question to a tighter page, yours or someone else's.

Check it yourself: For your target query, write six to eight follow-up questions a reader would ask. Search each and see who ranks and who gets cited. Where your site has no focused answer, add a section under a question heading, or a separate page linked from the main one. You can't control who Google cites, which is why this is only partly fixable.

Cause 5: Does the page show any proof?

Fixable: yes, but slowly

The same Search Engine Land piece says trust signals need to sit on the page itself: who wrote it, where numbers came from, and something that couldn't have been written by someone who never worked in the field. Google's own helpful-content guidance asks similar questions. This one is the slowest to fix, because proof means real work such as original data, named authors and primary sources.

Check it yourself: Open the page as a stranger. Is there an author, a date, a link to a source for every figure, and at least one detail only you could add?

Cause 6: Do other source types win the slot?

Fixable: partly

Pages from forums and video often fill citations. In an Ahrefs dataset, YouTube accounted for 5.6% of all AI Overview citations (via Search Engine Journal, checked 27 Oct 2026). In an Ahrefs Brand Radar listing dated 16 July 2025, reddit.com led by citation count. If the people asking your question are answered on those sites, a thin article loses.

Check it yourself: Expand the overview and write down the domain types it cites: forum, video, brand, publisher. If your page type is absent, a better article alone may not be enough. Consider answering in the places that do get cited.

What does a missing citation not mean?

It is not proof of a penalty. Real manual actions appear in Search Console's Manual Actions report. A missing citation alone is not one.

It doesn't mean you need schema. Ahrefs tracked 1,885 pages that added JSON-LD against matched controls. Citation changes were +2.4% in AI Mode and +2.2% in ChatGPT, both too small to separate from chance. In AI Overviews the change was −4.6%. Every page already had 100+ AI Overview citations, so the test says nothing about uncited pages (via Search Engine Journal, checked 27 Oct 2026). Google adds that no special schema is needed.

It doesn't mean you need new files. Google's guide says you don't need to create machine-readable files, AI text files or special markup to appear.

It doesn't mean your ranking is wrong. A top-10 position is neither necessary nor sufficient for a citation, based on the studies above.

What is a 10-minute check you can repeat?

  1. Pick one query you rank for. Search it in a private window.
  2. Note whether an AI Overview shows, and the date.
  3. If it does, list every cited domain and URL.
  4. Write down your own ranking position for the same query.
  5. Run the six checks above in order, starting with eligibility.
  6. Fix one thing, request a recrawl, and recheck in a few weeks.

If you want a sheet for this, the free AEO audit template has the tables.

FAQ

Why am I not cited in AI Overviews if I rank on page one?

Ranking gets a page into consideration, but selection depends on more. The citation may go to a page that answers a sub-question more directly, to a different source type, or the query may not show an overview at all. Causes 2, 4 and 6 above cover most cases.

Do I need special markup or an llms.txt file?

Google's AI features guide says no extra requirements, no special schema and no new machine-readable files are needed to appear (checked 27 Oct 2026). That guide covers Google, so it doesn't speak for other AI tools.

Will adding schema get me cited?

Not on current evidence. In Ahrefs' matched test of 1,885 pages, no platform showed a clear gain. Schema is still fine for rich results, as long as it matches visible text.

How can I see AI Overview traffic?

Google says sites appearing in AI features are counted in overall Search Console traffic, reported in the Performance report under the Web search type (checked 27 Oct 2026).

How long do fixes take to show?

Google says crawling can take from several days to several months, depending on how often a page is refreshed, and you can request a recrawl after changes.

The bottom line

Missing from an AI Overview is a symptom with six possible sources. One is an eligibility error you can fix in an afternoon. One means nothing at all. The rest are about answer structure, proof and what sort of source gets used. Confirm eligibility first, check that the query shows an overview, then work through the content causes one at a time.

Sources (checked 27 Oct 2026)