A '68% of Local Searches' AI Stat Is Making the Rounds Again. For Dealers, It Points Somewhere Nobody's Looking.

If you didn't get this number in a vendor email this week, you will get it next week.

On July 22, Search Engine Journal ran a piece by Adam Heitzman, managing partner at the SEO firm HigherVisibility, under the headline "AI Overviews Now Answer Most Local Searches – How To Get Your Business Cited." The deck carries the number that will end up on the slide: AI Overviews now appear in 68% of local searches versus 39% for local packs. The article calls the difference "The 29% Nobody Is Talking About."

The number is real. Heitzman sources it honestly — he names Whitespark and links the study. But almost nobody who forwards that stat is going to click the link, and the study underneath it says something more specific, and frankly more useful to a car dealer, than the headline does.

I went and read it. Two things jump out. The first is a date. The second is a table that quietly inverts the whole framing for a business like yours.

What the study actually is

The source is a Whitespark case study, "The Prevalence of AI Overviews in Local Search," by Miriam Ellis. It was published May 12, 2025 — about fourteen months before the article quoting it.

Here's the methodology, in Whitespark's own words: "Whitespark manually collected data in three large US cities (Houston, Phoenix, and Denver), across 6 industries (plumbers, personal injury lawyers, dentists, optometrists, medical clinics, and real estate agents). We ran a total of 540 queries across these combinations of city + industry."

So: 540 manual queries, three metros, six verticals. It's an honest, clearly-labeled study, and Whitespark itself calls it "even a modest-sized study like ours." What it is not is a census of local search, and it does not contain a single automotive query. The word "automotive" appears zero times. The only reference to a car in the whole thing is an example personal-injury query — "should I get a lawyer after a car accident in Phoenix?"

That doesn't make it useless to us. It makes it something you have to translate rather than quote. And the translation is where it gets interesting.

The table that flips the headline

Whitespark split its 540 queries into three intent categories: local, informational, and hybrid. A local query is personal injury lawyers in Phoenix. An informational query is how much do personal injury lawyers charge in Phoenix? A hybrid query is should I get a lawyer after a car accident in Phoenix?

Here's how the two features split across those three buckets:

AI Overviews appeared on:

• 15% of local-intent queries
• 92% of informational-intent queries
• 97% of hybrid-intent queries

Local packs appeared on:

• 93% of local-intent queries
• 6% of informational-intent queries
• 17% of hybrid-intent queries

Read that twice, because it's the opposite of the story the 68% tells on its own. The 68% average is not AI eating the map pack. It's AI dominating a different set of searches than the map pack ever served. Whitespark's own finding puts it plainly: the two features "have an inverse relationship, based on query intent." On the traditional local searches — the "near me," the "in Denver," the ones with a business name shape to them — the map pack still shows up more than nine times out of ten, and AI Overviews have made a 15% inroad.

The average of 68% is high because two of the three buckets are almost saturated. It's not because your "Ford dealer near me" result has been replaced.

What this means for your dealership

Every dealer I talk to reads a stat like that and immediately thinks about the same two things: the map pack and the VDP. Those are the assets they can see, so those are the assets they worry about.

Based on this data, that's the wrong worry — or at least the wrong first worry. Let me be specific about how a dealership's search footprint maps onto those three buckets.

Local intent is Chevy dealership Houston, Ford dealer near me, Toyota service center open Saturday. This is map pack territory. It is still, per this data, overwhelmingly map pack territory. Your Google Business Profile is the asset, and it's the asset that pays — which is why we wrote earlier this week about Google's AI increasingly citing its own Business Profile cards instead of dealer websites. That story and this one point at the same page.

Informational intent is how much does a transmission flush cost, how long does an alignment take, what does a 60,000-mile service include, do I need a cabin air filter. Ninety-two percent AI Overview rate. This is your service drive, and it is the single largest pile of unanswered questions on the average dealer website. Most stores have one thin "Service Specials" page and a scheduler. There is no content there for an AI to cite, so it cites a national parts blog or a forum thread instead.

Hybrid intent is the one that should genuinely get your attention: average down payment on a used truck in Phoenix, is a certified pre-owned warranty worth it, best time of year to buy a Silverado, what credit score do I need to finance a car in Denver. Ninety-seven percent. These are the questions a shopper asks three weeks before they ever type a dealership name — the questions that decide which brand, which body style, which store's lot they walk onto. And a shopper who gets that answer from an AI without you in it has quietly narrowed their list before you knew they existed.

That's the exposure. Not your near-me traffic. The research questions in front of it, and the fixed-ops questions behind it.

The honest caveats

I'm not going to overstate a 540-query study from last spring, so here are the three limits you should hold onto:

It's fourteen months old. Google has shipped AI Mode, expanded AI Overviews, and changed the SERP repeatedly since May 2025. The direction of travel since then has been more AI surface, not less. A rerun today would very likely show a higher local-intent number than 15% — but nobody has published that rerun, so 15% is what the evidence supports, and I'm not going to invent a newer figure.

Automotive wasn't in it. Six verticals, none of them retail auto. Whitespark's own strongest finding is that AI prominence tracks industry rather than geography — which means the responsible read is that we do not know auto's exact number, not that we can borrow the dentist's.

But look at the outlier. The one vertical Whitespark flagged as breaking the pattern was real estate: AI Overviews appearing for as much as 50% of local-intent queries, more than triple the average. Think about what makes real estate different from a plumber. Big-ticket. Long consideration window. Heavy comparison shopping. Financing. Inventory that's individually listed and individually priced.

That is the closest thing in this study to a car dealership. And it's the vertical where AI had already pushed furthest into the local-intent bucket fourteen months ago. Draw your own conclusion about where we sit on that curve today.

What to do about it

1. Split your search report by intent before you panic — or relax. Take your top 200 queries and sort them into local, informational, and hybrid. Most dealers have never done this. You cannot act on a 68% average; you can act on knowing that 140 of your queries are informational and you have content for eleven of them.
2. Write the fixed-ops answers. Real pages, real prices or ranges, real time estimates, per store. "How much is a brake job on a 2021 Explorer at our shop" is a page almost no dealership has and every AI is looking for. This is the highest-return content on your site right now and it costs you a service advisor's afternoon.
3. Build the hybrid pages with the city in them. Whitespark's hybrid examples all have the shape [question] + [metro]. Financing, trade-in timing, CPO value, total cost of ownership — answered specifically for your market, not a syndicated OEM blog post that's identical on 400 other dealer sites.
4. Don't strip the Business Profile budget to fund it. Local-intent is still 93% map pack. Hours, photos, review responses, service attributes — that work is not obsolete, and it's free.
5. Track presence per query, not a single score. Check whether an AI Overview fires on your actual queries and whether you're in it. One blended "AI visibility" number will tell you very little, for reasons we've covered before.
6. Date every stat somebody sends you. When a vendor emails you a percentage, ask three questions: when was it collected, in which markets, and in which industries? If the deck can't answer, the number isn't a strategy. That single habit will save you more money this year than any tool you buy.

The bottom line

The 68% figure isn't wrong and it isn't hype. It's just averaged across three very different kinds of searches, and dealers keep applying it to the one bucket where it's least true.

The map pack is not gone. What's happening is quieter: the questions that come before a shopper knows which dealership they want, and the questions that come after they own the car, are now being answered by a machine that reads whatever content exists — and on most dealer websites, that content does not exist. Your competitor down the street hasn't written it either. That's the whole opportunity.

If you want to see which of these three buckets your store actually shows up in — and what an AI says about you when a shopper asks it a question you've never answered on your site — run your dealership through our AI compatibility test and we'll go through the real queries with you.