That is the shape of the problem in the newest search data, and it is not a ranking problem.

What the 43% figure actually says

Similarweb’s generative AI report put AI Overviews on 43% of US searches in May 2026, up from roughly 15% in January 2025. Reporting on the same dataset describes Google’s AI search becoming the default rather than a layer sitting on top of it.

Treat the number itself with some care. Similarweb has not published its query sample, device mix or vertical weighting, and it does not define precisely what counts as an Overview appearance. Other measurements land a lot lower. Adthena recorded AI Overviews on 18% of US searches in June 2026, and an academic study put the activation rate under 14%. Every source agrees on the direction. None of them agree on the size.

So the useful takeaway is not the number. It is that the share is climbing fast enough that a plan built on ten blue links is depreciating faster than most content calendars assume.

Ranking and being cited are now two separate jobs

This is the part that catches teams out. You can hold position one and lose the citation. The AI layer is not picking the best page. It is assembling an answer from sources it has some reason to trust, and brand familiarity is one of those reasons.

For an established name that is free. For everyone else the work moves. Better content does not fix an absence of recognition, because recognition is built somewhere other than your own site.

What that changes in practice

• Entity consistency. The same company name, the same description, the same category language everywhere you appear. Directory listings, your LinkedIn page, trade association records, supplier databases. Models form an idea of what your business is from the agreement between those sources, not from your homepage.
• Mentions where the model already looks. One paragraph about you in a trade publication moves citation odds more than another post on your own blog. That is uncomfortable if your entire plan was publishing.
• Unbranded content specific enough to quote. Our kitchen equipment supplier will get cited for descaling intervals by machine type long before it gets cited for “best commercial dishwasher.”

What can go wrong

The obvious failure is treating this as a volume problem. If citation follows recognition, twenty more posts on a site nobody links to changes nothing, and it burns the hours that a single trade placement would have used better.

The second failure is over-correcting. Some businesses read the AI Overviews coverage, decide organic search is finished, and shift budget to paid before measuring anything. Look at your own numbers first. If your traffic sits in transactional queries, or in a vertical where Overviews rarely fire, the headline percentage may barely touch you. Local service businesses in particular tend to see far less of this than national averages suggest.

This also may not be worth prioritising at all if your site is six months old with no organic base. Recognition compounds, which is a polite way of saying it starts slowly. A brand new site chasing AI citations is optimising the last mile of a journey it has not begun.

The measurement gap nobody has closed

There is no clean way to see how often you are cited. Search Console reports clicks, not appearances inside an answer. The workable substitute is manual and dull: pick fifteen queries that matter commercially, check them monthly in a clean browser session, and log whether an Overview fires and who gets named. It is a small sample and it will never be statistically sound. It still beats guessing.

Worth knowing that Google has also started rolling out controls here. The AI search opt-out limits some of this exposure, though what you give up in return is worth understanding before you touch it.

Start with the fifteen queries

Before you rewrite a single page, find out whether the answers Google is now giving on your behalf mention you at all. That one check tells you whether your problem is content or recognition, and those two problems need completely different budgets. Most teams assume the first and are quietly losing to the second.

Editor’s note: This area changes quickly, so check the latest platform policy before making compliance decisions.

The post AI Overviews Now Cover 43% of Searches. Being Right Is Not Enough. appeared first on Mark8ng.com.