What an Overview is, exactly
An AI Overview is a generated answer at the top of a Google search results page, supplied for queries Google's systems classify as suitable for synthesis. The Overview is produced by a model with retrieval over Google's index; the sources it draws on are surfaced as citations within and beneath the synthesis.
The Overview is not the same as the conventional ten blue links beneath it. It uses a different reranker, a more conservative authority threshold, and — most consequentially — a different content shape preference. Where classic search rewards keyword fit and link density, the Overview rewards what we have come to call passage-level citability: short, attributable, evidence-bearing passages that the model can quote.
Why it differs from classic search
Three differences are operationally material.
- Authority threshold is higher. The Overview suppresses sources that classic search will still surface in the top ten. The system is asymmetrically biased toward avoiding incorrect synthesis.
- Source mix is narrower. A typical Overview cites three to six sources. Classic search exposes a hundred. Inclusion is therefore meaningfully harder.
- Content shape matters. The Overview prefers passages that stand on their own as quotable claims. Long, marketing-led pages without quotable substance are present in the index but absent from the Overview.
The fastest improvement we observe in Overview presence comes from rewriting existing pages — not adding new ones — to expose quotable, attributed passages that the model can lift cleanly. The same content reorganised, not new content commissioned.
The four moves that work
One — entity reconciliation, by Google's lights specifically
Google's entity graph (the Knowledge Graph) is the system the Overview is most heavily anchored to. Reconciling a subject in Google's terms — Knowledge Panel where applicable, Google Business Profile, structured data, sameAs assertions — is the prerequisite to every other move.
Two — passage rewriting
Existing pages are reorganised so that the substantive answers to the prompt live in short, attributed, evidence-bearing passages. The institution typically reduces page length while increasing citability.
Three — authority graph distance
The Overview heavily favours sources within one or two hops of recognised authorities. Earning credible third-party citations — not link-building — is the work that closes this distance.
Four — schema appropriate to question type
FAQPage, HowTo, Article, Product, Service, MedicalBusiness, ProfessionalService schema applied accurately. The Overview reranker uses schema as a confidence signal; pages with appropriate schema are reranked more favourably.
What backfires
- Adding visible “optimised for AI Overview” headings or pseudo-FAQ blocks not anchored to the page's substance.
- Stuffing schema markup that does not correspond to the page content, which the system increasingly detects and penalises.
- Removing depth in the belief that the Overview prefers brevity — it prefers citability, which is not the same.
- Acquiring inclusion in low-editorial directories, which Google's system increasingly weights negatively.
How the institution measures it
The Index measures Overview presence as one constituent of the Citation Source Quality and Recency Stability signals, weighted higher in queries Google classifies as Overview-eligible. Subjects on a Google AI-led brief are reviewed quarterly with a stable prompt set; the institution reports the proportion of Overviews in which the subject appears, and the citation rank where present.
A closing observation
The Overview's source list is the part of Google that has begun to behave like an editorial product. The work of being included in it is editorial work, not technical work, regardless of how the brief is described in the agency contract.
The teams that succeed at Overview presence are the teams that have stopped writing for search engines and started writing for the next reader to be quoted. The system is, in this respect, the reader.