AI Search ยท By Jeff Deutsch

How Google AI Overviews choose citations

Last updated

What the Overview rewards

Google's AI Overview cites sources that give it something it cannot generate itself: first-party data, a named expert's judgment, an experience that reads like a person rather than content. Informational queries are increasingly eaten whole, a question like whether body temperature rises during sleep no longer sends traffic to anyone, so the only durable position is being the source the answer is built from.

The signal I watch is citation velocity around algorithm updates. After the April 19 update I tracked one entity whose AI citations from Google jumped by 142,000, from a base of 10,000, with no new pages shipped that month. The update changed who Google trusted, not what existed. Meanwhile a competitor that started with half a million organic visitors a month and a big Overview footprint dropped roughly 60 percent. The Overview giveth and taketh at update cadence, which is why the traffic-impact numbers are so violent.

Writing pages the Overview quotes

My content method uses the AI Overview itself to determine content order: the sections the Overview generates for a query are the sections buyers want, in that order. But replicating the Overview's content gets you nothing, because Google already has that answer. The client's unique point of view is what earns the citation, which means every page needs a named author saying something checkable that no one else has published. This is the same discipline as GEO and the heart of SEO that still works.

The warning label: Google is actively separating real businesses from spam operations shipping thousands of AI-generated pages. I have been on the wrong side of automated content once, at ContactOut, where launching AI content at scale without human review earned crawl issues and penalties that took longer to fix than doing it properly would have taken. Scale is fine. Unreviewed scale is how programmatic SEO goes wrong. Being cited still builds brand and topical authority even when the click stays with Google, so the effort pays either way, and answer engine optimization covers the prompt-side version of the same play. One more practical habit: keep a monthly screenshot log of the Overviews on your priority queries. The citations rotate quietly, and the log is often the first place a competitor's new content strategy becomes visible, weeks before any rank tracker notices. It also gives you the before-and-after evidence that makes the case for investing in the mention engine internally.

FAQ

How do Google AI Overviews pick which sites to cite?

They favour sources offering what the model cannot generate: first-party data, named expert judgment and genuine experience. Comprehensive-but-anonymous content gets summarised, not cited.

Do AI Overviews steal traffic?

On informational queries, yes, often entirely. The durable position is being the cited source, which still builds brand mentions and topical authority that show up in other channels even when the click stays with Google.

Can algorithm updates change AI Overview citations?

Dramatically. After the April 19 update, one tracked entity gained 142,000 Google AI citations on a base of 10,000 without shipping anything new, while a competitor with half a million monthly visitors dropped about 60 percent. Updates reallocate trust, not just rankings.

Want to know where you stand in AI search?

I ran growth at Superpower (3k to 200k organic visitors a month), ContactOut and VIPKid. The AI Search Visibility Audit runs 25 of your buyer queries through ChatGPT, Gemini, Perplexity and Google AI Overviews, then hands you a citation scorecard and an ordered fix list. Free, delivered in five business days.

See the audit