Google AI Overviews

Google AI Overviews (originally Search Generative Experience, or SGE) are AI-generated summaries at the top of Google results, above the ranked list of links. They synthesise an answer from several web sources and show a small set of citations, typically two to five, that a user can click to verify or read more.

AI Overviews launched in the US in May 2024 and expanded to more than 100 countries and 8 languages throughout 2024-2025. In 2025, Google also launched AI Mode, a fully conversational, multi-turn experience that lives on a separate tab but uses the same underlying system.

Coverage is material. BrightEdge reported Overviews on 48 percent of Google searches in February 2026, up 58 percent year on year. Conductor's analysis of 21.9 million queries put the rate at 25.1 percent. The gap reflects query mix: Overviews appear far more on informational and comparison queries than on transactional or navigational ones.

The effect on the links underneath is large. Dataslayer's analysis of Search Console data (January to April 2026) found queries that consistently trigger an Overview produce meaningfully lower click-through even at top positions. Ranking first on a query that triggers an Overview is worth much less than being one of the few sources cited inside it.

AI Overviews run on Google's Gemini models layered on top of the standard Google index. The process is a constrained version of Retrieval-Augmented Generation:

Triggering. Not every query gets an Overview. Google models which queries benefit from a synthesised answer (informational, how-to, comparison) and which are better served by links (navigational, transactional, brand). Rates vary by category, and health and finance topics have historically triggered less often than travel or product comparisons.

Candidate retrieval. When an Overview triggers, Google retrieves candidates from its index with a bias toward clear answer structure, complete schema, recency and authority. BrightEdge reported pages with complete product schema were 2.5 times more likely to be cited (2025).

Synthesis and citation. Gemini composes the answer from the retrieved sources and attaches a few inline citations. There are almost always more candidates than citation slots, and that selection is where optimisation lives.

Seer Interactive found 85 percent of Overview citations came from content published in the last two years, and 44 percent from the last year. Freshness acts as a filter rather than a tiebreaker.

The impact pulls in two directions:

Lower click-through on the links beneath. Queries that trigger an Overview see click-through fall even for first-ranked pages. Semrush and Dataslayer both reported double-digit compression on Overview-triggering queries in 2025 and 2026. When the question is answered inside the Overview, the click is absorbed.

Higher quality from cited clicks. When a retailer's page is cited and the user clicks through, that visit is high intent. Adobe Analytics reported AI-referred traffic to US retailers up 393 percent year on year in Q1 2026, converting 42 percent better than traditional search traffic (Adobe, April 2026).

Volume down, quality up. For retailers whose product and category pages get cited, Overviews are a net gain. For retailers relying on volume from low-intent informational queries, they are a drag.

The practical consequence is to know which of your queries trigger Overviews and which do not. On triggering queries the job is to be a cited source. On the rest, traditional SEO still applies cleanly. This is one of the first things a Paz expert checks when investigating a traffic drop.

The levers are well understood by now:

1. Answer-first copy under each heading. A direct 40 to 60 word answer right under the heading. Citation often happens at section level, so a well-structured section can earn it even when the page is not top ranked.
2. Complete schema on product and article pages. Product, FAQPage, HowTo and Article schema all raise the odds. For a product page, product schema with name, image, description, offers, brand, sku, gtin, aggregateRating and MerchantReturnPolicy is the baseline.
3. Freshness. Visible dateModified, refreshed examples and periodic updates to the pages that matter most.
4. Entity clarity. Consistent brand naming, Organization schema, named authors, and third-party presence on review sites and industry publications.
5. Unblocked crawler. robots.txt must allow Google-Extended alongside Googlebot. A retailer who blocks it cannot be cited whatever the content quality.
6. Original data. Overviews favour content with named studies and proprietary figures that competitors cannot copy.

For product pages, items two and one are where the gaps usually are, and across a catalog they are steady work. Paz handles that: a Paz expert finds the pages missing schema or answer-first content, fixes them from your sources, our experts check the changes against agreed guidelines, and the fix publishes at the source. See the product page SEO service.