We Ran 1,100 Car Shopping Searches Through Google. Here's Who AI Overview Actually Cites.

Last week, Google announced its biggest Search redesign in 25 years. The new "Intelligent Search Box" adds multi-modal uploads, conversational AI Mode, and agentic search for paid tiers. Within 48 hours, LinkedIn filled with takes telling dealers their websites were dead.

Most of the dealers we talk to every day were not buying it. They wanted data, not headlines.

So we did something nobody else seems to have published yet: we mapped which US dealer websites Google's AI actually cites for car shoppers right now, and which dealers are dominating those citations from a single page each.

We ran 1,100 real car shopping searches across 20 US metros and tracked every domain Google's AI Overview pulled from. We found that 381 individual US dealer websites are showing up in Google's AI answers today, and a handful of them are pulling national-scale visibility from a single piece of content. The patterns are not what either the panic crowd or the dealer pessimists predicted.

This week we sent 1,100 real car shopping searches through Google across 20 US metros spanning 17 states, captured every AI Overview that returned, and tracked exactly which sources Google's AI cited. Here is what the data actually says.

This study is the quantitative follow-up to our reactive piece on the Intelligent Search Box rollout. That post made the contrarian argument that the panic is misplaced. This one is the data behind that argument.

And because this is a moving target, we are committing to re-running this entire study every 30 days for the next 3 months so dealers can watch the pattern shift in real time.

Want the raw dataset? We are happy to share the full 1,100-query list, the cited-domains CSV, and the aggregate summary with serious researchers, journalists, dealer-group analysts, and partners who want to reproduce or extend our methodology. Request the dataset and we will send it over. If you find an error in our methodology, tell us and we will publish a correction.

How the Study Works

We broke the 1,100 queries into three groups that mirror how a real shopper moves through the funnel.

Research queries (500 total). Top-of-funnel things like "best midsize SUV 2026", "Toyota RAV4 vs Honda CR-V", "is leasing better than buying a car". These are the questions people ask before they have any specific dealer in mind.

Buying queries (500 total). Bottom-of-funnel local intent: "Honda dealer near me", "2026 F-150 for sale Tampa", "Toyota lease deals Atlanta". This is the inventory and "where do I go" stage.

Branded queries (100 total). New for this study. We tested whether Google's AI even attempts to summarize a specific dealer by name: "AutoNation Tampa reviews", "best Honda dealership in Charlotte", "Carvana vs dealership Phoenix". This is the test most dealers have not seen run before.

Markets ranged from DMA #1 (New York) to DMA #143 (Lubbock TX), spanning 17 states with a deliberate mix of large metros, mid-tier markets, and sub-100 DMA small markets. Every search ran through SerpAPI against live Google results with the correct geo location set, so what we captured is what an actual shopper in that city would have seen.

Finding 1: The Funnel Has Already Split in Half

This is the headline finding and we want you to sit with it for a second.

• Research phase: 98.4% of queries returned an AI Overview
• Branded phase: 62.0% of queries returned an AI Overview
• Buying phase: 17.2% of queries returned an AI Overview

Read that again. For top-of-funnel research, AI Overview is essentially the answer Google now serves. For local inventory and "near me" buying intent, the classic SERP is still doing the work in more than 82% of cases.

This is the most important strategic fact in the entire study. AI is not killing dealer websites. AI is replacing the research phase of the funnel. Your local SEO and Google Business Profile are still the cash machine. What is gone is the unbranded, top-of-funnel content traffic, unless Google is citing your page in the AI Overview itself.

The calm news, for dealers ready to hear it: more than 8 out of 10 car-buying searches still return zero AI Overview. Local pack, organic blue links, Google Maps. The same SERP shape your local SEO has been built around for the last ten years is still running on the queries that actually fill your showroom floor. That can change. We are publishing a re-run every 30 days specifically to track whether it does.

Finding 2: One Dealer Page Got Cited 39 Times Across 20 Cities

This is the finding that made us re-check our data three times.

A single page on a Virginia Toyota dealer's website, a long-form blog post titled "Leasing vs. Purchasing a Vehicle: Which Option Is Right for You?", was cited 39 times by Google's AI Overview in our 1,100-query corpus, across all 20 markets in the study. Same URL, same content. Google's AI is using one Virginia dealer's blog post as the answer for car shoppers nationwide who ask "should I lease or buy."

Four other dealers performed almost as well:

• A California Ford dealer got cited 38 times
• A Florida auto mall got cited 38 times
• An Ohio Ford dealer got cited 38 times
• A Maryland Hyundai dealer got cited 37 times for its "Best Time of Year to Buy a Car" blog post

This is the play. Each of these dealers wrote one long, specific answer to a question that hundreds of thousands of people ask. They did not write an inventory page. They did not write a fluff "why we care about our community" blog. They wrote a real answer to a real question. And now Google's AI is rebroadcasting their content as the official answer in cities they cannot physically deliver cars to.

What they have in common is more important than what makes them different:

1. One specific evergreen question per page. Not a content hub. Not a 10-part series. One URL with one complete answer.
2. Question-format title. "Leasing vs. Purchasing: Which Is Right for You?" IS the query a shopper types. AI Overview prefers exact title-to-question matches.
3. Clean, descriptive URL slug. No IDs, no dates, no campaign tracking.
4. Geo-neutral content. They did NOT write "Why Virginia drivers should lease." They wrote about leasing universally. This is the counter-intuitive part: localized top-of-funnel content actually HURTS your AI citation reach. AI Overview rewards answers that travel.
5. No inventory carousel above the fold. AI Overview extracts the page's actual content. If the top 70% of your page is a vehicle widget and the bottom 30% is the answer, the AI ignores you.

Cost to produce: roughly one blog post each. Distribution: free, perpetual, nationwide.

Finding 3: You Are Not Competing With Other Dealers in AI Overview. You Are Competing With YouTube and Reddit.

Here are the top 6 most-cited sources across all 640 AI Overviews in our study:

1. YouTube with 385 citations
2. Reddit with 292
3. Edmunds with 234
4. Facebook with 152
5. US News Cars with 137
6. Cars.com with 125

All US dealer websites combined: 1,535 total citations spread across 381 unique dealer domains. Which sounds like a lot until you realize it took 381 different dealers to combine for roughly four times what a single YouTube domain delivered, and to still lose the per-domain race to YouTube, Reddit, and Edmunds.

In classic SEO, dealers fought other dealers for "best truck 2026". In AI Overview, you are fighting a 19 year old's TikTok-quality YouTube review and a 200-comment Reddit thread on r/whatcarshouldIbuy.

The dealers who broke through did one thing: they answered a specific question better and longer than anyone else.

Finding 4: Reddit Is the #1 Cited Source for Branded Searches About Your Dealership

We added the branded query bucket specifically to test something dealers obsess over: does Google's AI ever say my dealership's name in its summary?

The short answer is yes, but here is the unsettling part. When a shopper searches "best Honda dealership in Charlotte" or "AutoNation Tampa reviews", the #1 source Google's AI cites is not the dealer. It is not the OEM. It is Reddit, with 41 citations across our 100 branded queries.

The full top-5 cited sources for branded queries:

1. Reddit with 41 citations
2. Yelp with 35
3. Carfax with 28
4. DealerRater (Cars.com's review aggregator, not a dealer) with 26
5. Cars.com with 24

A single dealer broke through this in a meaningful way. A California Chevrolet dealer got cited 18 times in branded queries, more than any other actual dealer site in the entire study. They figured out something the rest have not, with one twist: the page Google's AI is citing has a URL ending in __trashed, which is WordPress's marker for soft-deleted content. The dealer literally moved this page to their CMS trash, and Google is still citing it as the official answer for "Carvana vs dealership" in 18 different US cities. The accidentally-archived page is outperforming most dealers' active websites.

The lesson is not "delete your pages and hope for the best." The lesson is that one well-written comparison page, indexed once, can keep paying out for months. Most dealers will not write the comparison because Carvana is the enemy. This dealer wrote it honestly, and they are still being cited for it.

What this means: your reputation strategy needs to push great content onto the third-party review platforms Google's AI trusts, not just your own About Us page. Your owned content can win research queries. Third-party content wins branded queries.

Finding 5: Market Size Does Not Matter

We sliced AIO coverage by approximate DMA rank to answer the question dealers always ask first: "does this apply to MY market?"

Tier	Description	AIO Coverage
T1	Top-10 DMAs (NYC, LA, Chicago, Houston)	58.4%
T2	DMA 11-25 (Phoenix, Seattle, Charlotte)	56.4%
T3	DMA 26-100 (Nashville, Toledo)	57.3%
T4	DMA 100+ and small markets (Lubbock, Fargo, Monroe, Bowling Green)	61.4%

The pattern is essentially no pattern. AIO coverage is flat across market sizes. The smallest markets in our study actually saw slightly higher AI Overview coverage than the largest.

If you were hoping your market was too small for Google to bother with AI Overview, the data does not support that hope. AI Overview is rolling out evenly across DMA tiers.

What Dealers Should Actually Do With This

We are not going to give you the usual "publish more content" advice. Here is what the data actually suggests.

1. Audit your existing top-of-funnel pages for AI Overview citation potential. Look for pages on your site that answer a specific buyer question (leasing vs buying, best time of year to buy, what credit score you need, what to look for in a used car). Most dealer sites have these as buried blog posts. The two dealers cited most in our study had clear URLs, clear titles, and 1,000+ word answers. If your existing FAQ page is 300 words of corporate filler, rewrite it.

2. Stop measuring blog content by traffic to that blog. The Virginia dealer's lease-vs-buy page is getting cited in cities they do not sell cars in. That visibility may never produce a click. It is producing brand impressions in Google's AI Overview to qualified buyers nationwide. That is a different KPI than blog pageviews and your reporting needs to catch up.

3. Branded reputation work belongs on third-party sites. Our branded query data shows Google's AI prefers to summarize Reddit, Yelp, Carfax, and DealerRater over a dealer's own About Us page. Investing in your DealerRater, Cars.com, and Google Reviews profiles is not extra credit anymore. It is what determines whether your dealership shows up in branded AI Overviews when someone Googles your name.

4. Local SEO is still the cash machine. A 17.2% AIO coverage rate on buying queries means more than 82% of local-intent searches still send shoppers to the classic SERP and to Google Maps. Do not let "AI is killing search" panic distract you from the boring work that still pays the bills.

5. Treat content creation as a 6 to 12 month investment, not a 30 day experiment. The dealer pages getting cited 39 times in our study were not written this quarter. They have been compounding for a while. Start now and you will be in next year's study.

We Are Re-Running This Every 30 Days for the Next 3 Months

The internet is moving fast. AI Overview coverage in November 2025 was different than what we measured this month, and what we measure next month will be different again. Static one-time research has limited shelf life in this category.

Starting now, we are committing to re-running this same 1,100-query study every 30 days for the next 3 months and publishing the deltas. You can expect updated posts at roughly the same time each month showing:

• Whether AIO coverage is expanding or contracting by phase
• Whether the same dealer domains keep getting cited or whether the citation pool is rotating
• Whether new market tiers are seeing different patterns
• Whether the research-vs-buying split is converging or diverging

If you want a free audit of where your dealership currently shows up in AI Overview, get in touch.

Methodology and Limitations

We want this study to be challengeable. Here is exactly what we did, what we did not do, and where the findings could be wrong.

The query set:

• 1,100 queries total, broken into 500 research, 500 buying, and 100 branded
• The same 25 research questions and 25 buying patterns were run across all 20 markets (with city-name substitution where applicable)
• The 5 branded query templates per market test how AI Overview behaves when a shopper searches for a specific dealer or dealer-comparison query
• The full 1,100-query list is available on request. Researchers and partners who want to audit our query selection or reproduce the study can request the dataset.

What we tested:

• 20 US markets across 17 states, spanning DMA #1 (New York) to DMA #143 (Lubbock, TX)
• All searches run via SerpAPI against live Google results with geo location set per market
• Desktop device, US locale, no cache, English only
• AI Overview citations counted by unique domain per query (a 10-citation page is not double-counted within the same search)

What we did NOT test:

• Mobile-only Google results (AI Overview behaves differently on mobile)
• Service department queries ("oil change near me", "Toyota repair Tampa")
• Parts queries
• Used-vehicle-specific long-tail queries ("Honda CR-V under 15k", "Toyota Camry 2018 reliability")
• Spanish-language queries
• Competitor-brand-comparison queries beyond the branded set
• Queries that use specific dealer names (we tested the patterns but not actual dealer-name strings)

These gaps matter. AI Overview behavior is highly query-dependent. The patterns in this study reflect the broad shape of automotive search but a narrower or more specific corpus would produce different results. If your dealership's traffic comes mostly from queries we did not test, this study does not describe your funnel.

Dealer-domain detection:

The "381 unique dealer domains" figure is heuristic. We excluded OEMs, large aggregators (Cars.com, AutoTrader, Carvana), review platforms (DealerRater, Yelp), enthusiast forums, B2B dealer-vendor sites (dealer.com, vAuto, DealerSocket), editorial publications, finance institutions, government sites, and trade press. A handful of edge-case domains may still be miscategorized.

Reproducibility:

The full query list and the raw per-query result data are available on request. If you disagree with our conclusions, ask us for the dataset and run the same study yourself. We will publish a correction if you can show us where we are wrong.

• Snapshot date: May 21, 2026
• Next re-run: approximately June 20, 2026

Want the underlying dataset? Request access and we will share the 1,100-query list, the cited-domains CSV, and the aggregate summary with serious researchers, journalists, and dealer-group analysts.

---

Savvy Dealer is a digital marketing agency working with automotive dealers across the US on AI search visibility, Generative Engine Optimization, and the kind of content that gets cited by Google's AI Overview. If you want a free audit of your dealership's current AIO citation footprint, reach out.