Do AI Overviews show up on local searches?

We published that zero on 7 August and said plainly that we did not trust it yet. A gap in the measuring tool would look exactly the same as an absence in Google.

Twelve days and two further readings later, the local zero is still there. The instrument has changed underneath it, and our two non-city readings do not agree with each other.

What changed since 7 August?

The original finding was one Ahrefs pull on one day. Three things have happened to it since, and only one of them is comfortable.

The city zero has now been produced by two independent instruments. That weakens the tool-blindness explanation we raised against ourselves, without closing it, because three live queries settle nothing.

The non-city figure is the uncomfortable one. Either those five cost-and-comparison queries are unusually prone to triggering an overview, or coverage moved between the two dates. At that sample size we cannot separate the two.

How did we measure the original 160?

• Set A, 80 commercial queries, seeded from seo services, digital marketing agency, google ads management, social media marketing and website design
• Set B, 80 informational queries, question form, from the same seeds
• Both sets filtered to a minimum monthly volume of fifty, both fixed before we looked at a single result
• Inside Set A, 30 queries carried a city name, of the form digital marketing agency in Delhi

Why does adding a city name change the result?

Strip the city queries out and commercial searches still carried an AI Overview 22 times in 50. Add a city name and it went to nothing. For a query like digital marketing agency in Coimbatore, the map pack and the organic results already answer the question being asked.

"A summary helps someone who is learning. It does not help someone who wants a shortlist of firms near them."

That reading is ours, but Google states the underlying rule directly on its AI features documentation. AI Overviews "are only shown when our systems determine that it is additive to classic Search, and as such, often don't trigger."

Google describes the informational side too, naming a query fan-out technique that issues "multiple related searches across subtopics and data sources" to build a response.

Does a local business appear in AI answers at all?

Yes, and this is the part the original version of this post missed. An AI Overview failing to trigger on your city query does not mean your business is absent from Google's generative AI surfaces. Those are two different questions and we ran them together.

• Does an AI Overview appear above a city-qualified search? On our evidence, rarely
• Can a local business appear inside a generative AI response? Google documents that it can
• Where does that information come from? Business Profile and Merchant Center, not your pages

Google's July 2026 guide says that "where appropriate, generative AI responses can include product listings, product information, and information about local businesses". Your Business Profile is now an input to generative AI answers rather than only to the map pack, and accuracy there costs nothing.

How reliable is the instrument behind these numbers?

Three of our four readings came from Ahrefs. Google's optimisation guide has a sentence about that which we are obliged to quote against ourselves.

"Be wary of third-party tools that promise ranking success or claim to use "internal" Google metrics. No third-party tool has access to our internal ranking or AI systems."

We agree, and we have our own evidence for it. In a separate check on 17 August, one domain appeared in the live AI Overview for an Indian commercial query while the same tool's index recorded zero citations for that domain.

That is one domain on one query, so it proves nothing about the rate. It does show the two views can disagree, which is the only claim we make from it. It is also why we would rather trust the 11 August browser check, the one reading here that did not arrive through an intermediary.

Which requirement do Google's two pages disagree on?

• The setting exists, and Google's help page for it says sites are included in generative AI features by default
• The July page treats that inclusion as a condition of eligibility
• The December page, still live and unchanged, says there are no additional technical requirements
• So this is something to confirm rather than something to fix, and it is the only item here somebody could have switched off without telling you

What does Google say you should actually do?

• 1

  On optimisation: "There are no additional requirements to appear in AI Overviews or AI Mode, nor other special optimizations necessary."

• 2

  On markup: "You don't need to create new machine readable files, AI text files, or markup to appear in these features. There's also no special schema.org structured data that you need to add."

• 3

  On chunking, the July guide says there is no requirement to break content into small pieces and no ideal page length.

The generative AI optimisation guide also has a line about the clicks that still arrive, and it is worth reading to the end: "when people click from search results pages with AI Overviews, these clicks are higher quality (meaning, users are more likely to spend more time on the site)."

That parenthetical matters. Google is claiming longer time on site, not more revenue, and anyone turning it into a conversion promise is going further than the source does.

How do you measure this yourself now?

There is a dedicated report, which there was not when we first published this. Search Console now carries a Generative AI performance report covering AI Overviews and AI Mode impressions. Two limits are worth knowing before you go looking for it.

• It is not fully rolled out. Google says it is going to a subset of website owners, allowing for thorough testing before rolling it further
• It reports impressions only, grouped by page, country, date or device. There is no query dimension and no intent dimension
• It is a filtered view rather than a new bucket, drawing on the Web search type, so those impressions already sit inside your existing totals
• Absence of the report in your property is a rollout gap, not a finding about your site

What we would actually do with this

• 1

  Split your keyword list by intent, then split the commercial half again by whether a place name appears

• 2

  Open the Generative AI performance report if your property has it

• 3

  Check the generative AI inclusion setting in Search Console, because the July guidance treats it as a condition of eligibility

• 4

  Keep your Business Profile accurate, because Google documents it as an input to answers about local businesses

• 5

  Expect the informational list to lose clicks, and stop judging it on clicks alone

• 6

  Leave your location pages alone, because on this evidence they are not where the pressure is

• 7

  Re-run the check quarterly, because our own two non-city readings already disagree with each other

On point five, the metric that replaces clicks is brand visibility in AI answers, which is a different measurement rather than a softer one.

On point six, writing a clear answer near the top of a page is ordinary content craft that happens to survive extraction well. None of this contradicts Google's no special optimisation line, because Google is describing eligibility and says none of it is required to appear.

What can this study still not tell you?

• Instrument coverage. A zero is what a gap in Ahrefs' tracking would produce for India-locale local queries. The 11 August live check makes that less likely, because a second instrument produced the same zero, but three live queries do not settle it
• A zero is not proof of never. Nought out of thirty is still consistent with a true presence rate as high as 9.5%, on a one-sided Clopper-Pearson bound at 95% confidence with n=30. It means rare, not absent
• Stability. Our two non-city commercial readings, 22 of 50 and 5 of 5, were taken ten days apart on different query sets and do not agree, so this is not a settled rate
• Selection. Every seed is a marketing-services query, so this measures our own market. The intent pattern may travel and the rates almost certainly will not

The honest position is narrower than the headline. City-qualified commercial searches in our sets rarely carried an AI Overview, measured twice, on small samples both times. Treat that as a lead worth acting on cheaply, and see how we would run the same segmentation as part of a full SEO engagement.