← The Brief

Google expands AI Overviews on branded queries, inflating GSC impressions

Lily Ray reports rival sites now citing inside brand-name AIOs register as #1 rankings in Search Console.

Niklas Buschner

Google has started showing AI Overviews on a much wider set of branded queries over the past week, according to SEO practitioner Lily Ray, who credited Chris Long and DemandSphere for flagging the shift. Ray posted the observation on LinkedIn on 2 October 2026 and said she is seeing the pattern across many client accounts.

The mechanical consequence sits inside Google Search Console. Every AIO citation and sidebar link shown before a user expands more results is logged as position #1. Rival sites cited inside an AI Overview for your brand name can now register as #1 rankings in competitors' GSC, with the low click-through that implies.

Ray also pointed to a spike in the Generative AI filter in GSC over the past week, echoing a recent observation from Glenn Gabe. She framed the wider rollout as a way for Google to raise its daily AIO user count, and noted that 25 years of user habit has trained people to search a brand name to reach that brand's site.

Verifying it on your own property

Open GSC and compare the last seven days against the prior period, filtered to branded queries. Then apply the Generative AI filter. A jump in impressions with flat or falling clicks points to AIO expansion as the cause.

Check which queries now show an AIO by running the brand terms manually. Note which third-party sources are being cited. Those are the sites that will show as ranking #1 for your brand name in their own Search Console.

What the finding changes

The reporting here is one practitioner's observation across a client book, corroborated by two other named SEOs. It is strong enough to act on measurement, not strong enough to rebuild reporting around.

Branded impression counts in GSC are now a less reliable proxy for brand demand, because they include AIO surfaces the user may never click. Treat the Generative AI filter as the cleaner read, and separate branded from non-branded before drawing conclusions about either. The response is measurement discipline across several signals, not a single tactical move against the citation itself.