The short answer
Google AI Overviews pull from pages that state a direct answer near the top, carry clear structural signals (headings, schema markup), and already rank well organically for the underlying query — if your page buries the answer under a long introduction, has no structured data, or doesn't rank on page one already, it's very likely being skipped in favor of a competitor who does. Fixing the structure usually matters more than fixing the ranking.
What AI Overviews actually reward
Google's AI Overviews synthesize an answer from a handful of sources it considers both relevant and reliable, then show 3-5 of them as citations. Three things consistently separate cited pages from skipped ones:
1. The answer is stated directly, early. AI Overviews extract short, specific claims — not general topic coverage. A page that answers "what's the best CRM for a 5-person team" in the first two sentences beats a page that spends three paragraphs on company background before getting there.
2. Structured data is present. Schema markup (Organization, Product, FAQPage, Article — whichever fits the content) gives Google an explicit, machine-readable version of your claims, rather than making it infer them from prose. Sites with zero structured data are working with a real, avoidable disadvantage here.
3. The page already ranks reasonably well organically. AI Overviews aren't pulling from nowhere — they lean heavily on pages Google's traditional ranking already considers relevant and trustworthy for the query. A page buried on page 3 of organic results is unlikely to suddenly appear in an AI Overview citation for the same query, no matter how well-structured it is.
The most common reason sites get skipped
Across the audits we've run, the single most common pattern isn't bad content — it's generic category pages competing against specific answer pages. A page titled "Our Products" or "Features" loses to a competitor's page titled "Best Time-Tracking Software for Remote Teams in 2026," even when the generic page's underlying product is objectively better. AI Overviews reward specificity because specificity is what makes a claim extractable and citable — a generic page doesn't have one clean, quotable answer to lift.
Five fixes, ranked by effort vs. impact
- Rewrite your highest-intent page's opening two sentences to directly answer the core question, before any company background or scene-setting. Lowest effort, often the highest impact single change.
- Add basic schema markup (Organization at minimum, FAQPage if you have any FAQ-style content) — a one-time technical fix, not ongoing content work.
- Publish or restructure one page per key query you want to win, rather than relying on a single broad page to cover everything. Specific beats comprehensive for AI citation purposes.
- Add a comparison table where genuinely relevant (vs. competitors, vs. alternatives) — tables are unusually citation-friendly since they're already structured, extractable data.
- Publish
llms.txtat your site root — a plain-language summary of what your site is and does, aimed at AI crawlers specifically. Newer and less universally adopted than schema, but cheap to add and directly addresses how AI systems parse a site's purpose.
Checking whether any of this actually worked
The only real way to know if these fixes changed anything is to test the specific query again after making them — not to assume structural best practices translate into citations automatically. You can test one query free right now against your own site and see whether Google's AI Overview currently cites you, or run the full $49 audit to check 10-15 realistic queries across AI Overviews plus ChatGPT, Perplexity, and Claude at once, with a ranked fix-it plan specific to what's actually missing on your site.