Car shoppers are not only using Google anymore. A growing number of them open ChatGPT, Claude, or Google AI Mode and ask something like “which car dealerships in Chicago have the best reputation” before they ever visit a website or submit a lead. The dealerships that show up in those answers aren’t there by chance, and the ones that don’t may not even enter the consideration set.
Understanding how these platforms decide who to recommend, and what you can do to influence that, is now a real part of automotive marketing strategy. The mechanics are different from traditional search, and some of the assumptions dealers and agencies have been operating on about AI visibility are wrong in ways that cost real money.
This post breaks down how ChatGPT, Perplexity, and Google AI Mode actually work when someone searches for a dealership, what the data shows about which signals drive citations and recommendations, and the specific tactics that move the needle on AI visibility for car dealers.
Recommendations vs. Citations: Understanding the Difference
Before getting into tactics, it helps to understand how AI search platforms are structured, because the terminology matters and the distinction changes how you think about what you are trying to accomplish.
When someone asks ChatGPT which dealerships have the best reputation in their area, the platform produces two distinct outputs. The recommendations are the dealerships it names directly to the consumer as its answer. The citations are the sources it drew from to formulate that answer, often shown in a side panel or at the bottom of the response.
A dealership can appear as a citation without being a top recommendation, and a dealership can be recommended without its own website being a primary citation, because the platform may have built its understanding of that dealer from third-party sources like Yelp, Carfax, Reddit threads, or dealer review aggregators.
Understanding what large language models are and how they actually generate responses is useful context here. These are probabilistic systems, not deterministic ones, which has a direct implication for how dealers should think about measuring success in AI search.
There Is No Ranking in AI Search
One of the most important points Dane Saville makes in this webinar is also one of the most underappreciated: there is no ranking in AI outputs. You are either a recommendation or you are not. And because these systems are probabilistic, the same question asked twice from the same device twenty minutes apart can produce different answers, different ordering, different dealerships named, and even different formatting of the output.
Dane tested this directly. He asked both ChatGPT and Perplexity “which car dealerships in Chicago have the best reputation” twice each, approximately twenty minutes apart. The results were meaningfully different both times, on both platforms. Dealerships that appeared in the first output were absent from the second. The framing of the answer changed. The order changed.
This matters because some agencies are selling AI visibility as if it were a rankings report, claiming they can tell you how many times your dealership showed up as a citation or where you fall in the recommendations. Those claims are not grounded in how these systems actually work. The platforms do not expose what people are actually typing into them, the searches are not standardized, and the outputs vary by design. Any agency telling you they can track your AI citation share with precision is guessing, not measuring.
What Actually Drives ChatGPT and AI Recommendations
When Dane ran the Chicago reputation search through ChatGPT and Perplexity and examined the citations that drove those recommendations, several sources appeared repeatedly: Yelp, Carfax, Car Edge, Reddit, dealer review aggregators like DealerRater and BirdEye, individual dealership websites, and a local Facebook community group where someone had asked for recommendations.
Yelp’s presence is worth flagging specifically. ChatGPT has a licensing agreement with Yelp that gives it access to Yelp reviews, ratings, and business information in real time. In Dane’s test, Yelp appeared as a citation multiple times across both the ChatGPT and Perplexity outputs. For dealerships that have let their Yelp profile go unmanaged, this is a meaningful gap, because negative or thin Yelp data is now feeding directly into ChatGPT’s understanding of whether your dealership is worth recommending.
On the organic search correlation, a Seer Interactive study found that 87% of ChatGPT citations aligned with pages ranking well in Bing’s top results for the same query, reflecting the fact that ChatGPT primarily uses Bing’s index for its web search.
SearchLab’s own AI Mode research across 1,600 dealership keywords in three markets found that 44.9% of citations that helped build AI recommendations were pages ranking in the top five organically on Google, and 60.8% were in the top ten. The relationship between traditional organic performance and AI citation is not perfect, but it is real and consistent enough to matter strategically.
How Google AI Overviews and AI Mode Differ from ChatGPT
Google’s AI products behave somewhat differently from ChatGPT and Perplexity, and the distinction is worth understanding.
AI Overviews, which appear above standard search results, showed more consistency in which dealerships were recommended across Dane’s repeated tests than the standalone LLM platforms did, though the ordering still varied. For AI Overviews, SearchLab’s research found a stronger correlation with organic Google rankings than with Bing: pages ranking at the top of Google organically were significantly more likely to appear as AI Overview citations.
Google AI Mode goes further than AI Overviews in producing individualized, in-depth outputs. When Dane ran the same reputation search in AI Mode twice, the results were even more varied than in standard AI Overviews, with only one dealership appearing in both outputs. AI Mode is more probabilistic because it works harder to understand and personalize to the user’s specific intent, which produces greater variance in outputs.
The practical implication is that what separates top-ranking dealerships in organic Google search also provides the strongest foundation for AI visibility across Google’s products, and that foundation matters increasingly as Google continues transitioning toward AI Mode as its primary search experience.
Building Entity Knowledge: What Actually Moves the Needle
The underlying goal of everything that influences AI visibility is the same: building what Dane describes as entity knowledge, or what these platforms understand about your dealership in relation to the topics most relevant to your business. The richer, more consistent, and more accurate that picture is across every platform where your dealership appears, the more likely these systems are to include you in relevant recommendations.
Dane tested this directly with SearchLab’s own AI visibility score around the topic of automotive SEO. Starting at a score of 66 and trailing competitors significantly, the team ran a series of off-site tactics over approximately three weeks, without touching the website at all: adding automotive SEO to social media profile descriptions across all platforms, building out a profile on a relevant industry directory, publishing weekly Facebook posts, weekly LinkedIn posts, daily posts on X, and three weekly Instagram short-form video posts, all around automotive SEO topics. They also produced a podcast episode on automotive SEO fundamentals.
The score moved from 66 to 75 after three weeks of that activity. Then a press release about SearchLab’s GBP study and AI overview research was distributed and picked up by AP, Bloomberg, Google News, National Law Review, and others. A contributed article went live on Digital Dealer’s website. The score jumped to 89, moving SearchLab above both competitors and into Claude’s training data for the first time.
The report generated by the visibility tool confirmed which specific sources were driving that improvement, including the social media profile updates, the Digital Dealer article, and the press release distribution. The cause-and-effect was documented, not assumed.
The broader framework Dane describes for building entity knowledge encompasses reviews and user-generated content on platforms like Reddit, active social media profiles on Facebook, LinkedIn, Instagram, and YouTube, consistent directory and citation presence with accurate descriptions of the business across all profiles, press releases distributed to news outlets, contributed articles to industry publications and local media, and website content that answers consumer questions in depth around the topics most central to the dealership’s business.
For dealerships specifically, this means the Local SEO foundations that drive map pack performance are directly connected to AI citation performance: accurate Google Business Profile data, strong review volume and cadence, consistent NAP across directories, and a website that clearly describes what the dealership sells, where it operates, and what makes it worth choosing. SearchLab’s 2025 automotive GBP study documents how far the average dealership still has to go on those fundamentals.
How to Think About Measuring AI Visibility
Because AI outputs are probabilistic and the platforms do not share query data, traditional ranking metrics do not apply. Dane suggests several more honest indicators to track.
Branded search volume in Google Search Console is one: if your dealership is being recommended repeatedly in AI outputs, some of those people will search your name directly rather than clicking through. An increase in branded search suggests growing name recognition, which AI visibility can drive even when it does not produce a direct click.
Referral traffic from AI platforms is trackable in GA4, though for most local businesses the volume is currently small. Dane estimates most small businesses see under 5% of total site traffic coming from all LLM platforms combined. That will grow, and the infrastructure for tracking it should be in place now.
Conversion volume matters more than traffic volume: more calls, form submissions, and chats from qualified shoppers is the actual goal, and improving AI visibility should contribute to that over time even when direct attribution is difficult.
The most honest measure of AI visibility right now is a tool that shows what AI platforms understand about your dealership in relation to your core topics, benchmarked against your competitors, which is exactly what SearchLab’s AI audit produces.
Ready to See Where Your Dealership Stands in AI Search?
SearchLab can run an audit that shows how platforms currently understand your dealership relative to your core topics and your primary competitors, alongside the foundational SEO audit that underpins all of it. If you want a clear picture of where the gaps are and what is worth addressing first, start with a free discovery call and we will show you what we are seeing.
