Which searches actually show an AI answer?
Not the ones most local SEO is aimed at. A 540-query study found AI answers on 15% of "near me" searches and 92–97% of the questions people ask before they get that far.
There is a widespread assumption that AI has taken over search generally, and a matching one that whatever you were doing for Google still covers it. Both are wrong in the same specific way, and the shape of the mistake is measurable.
The split
Whitespark ran 540 local queries and recorded when an AI Overview appeared. The results divide cleanly by what the searcher was trying to do:
| Query type | Example | AI answer appears |
|---|---|---|
| Classic local intent | "med spa near me", "dentist Buckhead" | about 15% |
| Informational and hybrid | "how much does Botox cost in Atlanta", "what should I ask before booking a med spa" | 92–97% |
Nearly six times the rate, in the second group.
Why this is the expensive detail
Local SEO as an industry was built around the first row. Google Business Profile, categories, citations, review counts, proximity — all of it optimises the "near me" query where the map pack appears and, per this data, an AI answer usually does not.
Meanwhile the second row is where AI answers are close to guaranteed. And those questions come earlier. Somebody asking "how much does Botox cost in Atlanta" has not chosen a provider yet. Somebody typing "med spa near me" often has. The AI answer is landing at the point in the decision where the shortlist gets made.
So the common situation is a business doing everything right for the 15%, invisible in the 92%, and unable to see the gap because the tools everyone uses report on the first group.
The second half of the finding
The same study looked at where the citations inside those AI answers point. About 60% of them go to third-party sites — directories, "best of" roundups, local press, forum threads — rather than to the business's own website.
Put the two together and the standard package looks quite different:
- optimise your own site → addresses the minority of citations
- optimise for "near me" → addresses the minority of AI answers
Neither is useless. Both are aimed at the smaller half.
What to do instead
Find the questions your buyers actually ask before they shortlist. Price, comparison, process, what-to-expect, is-it-worth-it, how-do-I-choose. Those are the hybrid and informational queries. For most local categories you can list twenty in an afternoon by remembering what people ask on the phone.
Answer them publicly, in plain language. This page is an example of the format. It exists because the question gets asked.
Get accurately represented where the citations live. If 60% of the citations point at third-party sources, being correctly listed, described and reviewed on the ones your category actually uses is a larger lever than another schema audit.
Then measure the right queries. If you are tracking "med spa near me" you are tracking the 15%. The number that matters is whether you get named in the questions where an answer appears nearly always — which, since those answers change week to week, means tracking over time rather than checking once.
Source: Whitespark, "AI Overviews in Local Search" — 540 queries, split by intent, with citation destinations recorded.