Local & Industries · 9 min read · July 15, 2026
What People Ask AI Before Buying a Car: Data and Patterns for Dealers
Before they ever call you, plenty of buyers put the question to ChatGPT, Gemini or Perplexity first. They don't type keywords, they type situations: "family estate up to 25,000 euros with room for two child seats" or "is a used diesel still worth it in 2026?" What comes back is a short answer naming two or three options. This is how you make sure yours is one of them.
Why the First Question Now Goes to a Chatbot
The road to a new car no longer starts with Google alone. Plenty of your customers now open ChatGPT, Gemini or Perplexity first and ask the thing they would never have typed into a search box: full sentences about their life, their budget and what they are afraid of getting wrong. What comes back is a recommendation, a shortlist, sometimes a named dealer. That is the moment your dealership is either in the conversation or nowhere near it.
That is not the old search experience with an extra step. On a page of ten blue links the customer did the choosing; an AI answer hands them a conclusion with two or three names in it. Pew Research Center tracked the real browsing of 900 US adults across nearly 69,000 Google searches: when an AI summary appeared, people clicked through to a website in 8% of visits, against 15% without one, and clicked a link inside the summary itself in just 1%. (Google disputes the methodology.) If you are not in the summary, most of those buyers never reach your website, your phone or your lot.
Buying a car is one of the most expensive, advice-heavy decisions most people make, so the first answer carries real weight. And the audience is not niche: Alphabet has put AI Overviews at more than two billion users a month, and ChatGPT was reported at 900 million weekly users in late February 2026. Being the source an engine reaches for when someone describes their budget and their family is now part of how trust gets built, months before anyone meets you.
The Questions Behind a Car Purchase, in the Buyer's Words
Spend a Saturday at your own counter and you will hear the questions people are now typing into a chatbot, only longer. There is the parent: "which used estate under 20,000 euros is reliable with two child seats?" The commuter: "for 40 kilometres a day, does an EV beat a diesel?" The hesitant one: "should I lease new or buy nearly new?" None of that is a keyword. Each one is a decision handed over in a full sentence, constraints attached.
What stands out is how much worry is packed in. "What is my 2016 Golf with 120,000 kilometres actually worth?" "What will a used car with 100,000 kilometres cost me in the first year?" "Can I trust a dealer with no reviews?" People use a chatbot as an adviser with nothing to sell, which is exactly why they put the blunt, slightly naive questions to it that they would never say out loud to a salesperson.
For you that is a content brief written by your own customers. Each of those questions deserves one page that answers it properly. Publish the clearest honest answer anywhere to "EV or diesel for a 40-kilometre commute" — with the numbers, the caveats and the case against — and that page becomes the thing an engine has to work from when the question comes up.
What Generative Engine Optimization Means for a Dealership
Generative Engine Optimization, GEO for short, is the work of getting named and quoted inside AI answers, and it overlaps less than you would expect with classic SEO. An Ahrefs analysis found only about 6 to 8% of the URLs ChatGPT cites also sit in Google's top ten for the same query, and roughly 80% of them do not rank in Google's top hundred at all: citation is a different selection, not a reward for ranking. The one peer-reviewed study of GEO tactics (Aggarwal and colleagues, ACM SIGKDD 2024) ran its tests on a simulated engine rather than live search, and the rewrites that helped most added quotations, statistics and cited sources, while keyword stuffing showed no meaningful gain. Simulated results are not proof, but the direction matches what you already know: be quotable, not louder.
In practice that means pages that take a real customer question and answer it better than any generalist could. A guide called "Used-car check: the twelve things our master mechanic inspects before a car goes on the lot" is worth more to an engine than a page titled "Your dealership in Munich." One solves a problem and can be quoted; the other asserts a position and cannot. Google's own guidance says its AI features run on the same core Search ranking and quality systems, and warns against spinning up a separate set of content "for AI" — that route risks being treated as scaled content abuse. It is one job, not two.
Honesty here is load-bearing rather than decorative. A page that says plainly when a diesel is still the cheaper answer and when it stops being one gives an engine something checkable to lift. A page where every car is a great deal gives it nothing. Credibility is the thing actually being retrieved.
Local Visibility: When Someone Asks for a Dealer Nearby
A large share of car questions is really a local question. "Where do I find a used-car dealer near Regensburg I can trust?" "Which dealer around here does decent financing for a 21-year-old?" Answering those needs the engine to know your name, your patch and your terms — and to find the same version of them everywhere it looks.
So the unglamorous work carries the weight: a Google Business profile that is genuinely maintained, the same address and opening hours on every portal, real reviews with real replies under them. The strongest evidence for any of this points outward rather than at your own pages. Ahrefs looked at around 75,000 brands and found that how often a brand is mentioned across the web correlates with its AI citation rate at about 0.66, roughly three times the correlation for backlinks at 0.22. Being talked about by local press, forums and directories moves more than anything you can edit on your own site.
Then add content that is local without being clumsy about it. "Charging an EV in Regensburg: what our customers actually use" ties a real subject to a real place, and it gives an engine a reason to file you as the local expert rather than one more listing. Write a handful of those and location questions stop being a coin flip between the two dealers with the biggest ad budgets.
Your Machine-Readable Trail Is the Raw Material
AI answers get assembled from whatever a machine can parse without guessing. For a dealership that means the boring layer: one clear heading per question, vehicle facts in structured data fields rather than buried in a photo of a window sticker, and markup clean enough that a crawler sees the same page a customer does. What cannot be read cleanly cannot be quoted.
Specifics are what get quoted. "A 2019 Skoda Octavia runs 15,900 to 18,500 euros with us, depending on trim and mileage" is usable; "fair prices, top service" is not. Schema.org markup helps here for a narrower reason than most vendors claim: it makes facts like price, address and opening hours extractable and harder to garble. Google has said since 2018 that structured data is not a direct ranking factor, and its AI guidance is blunt that no special markup, schema or AI-specific file is required to show up in AI Overviews or AI Mode.
That same logic kills the shortcuts. Google's John Mueller confirmed in 2025 that no Google Search system reads llms.txt, and an Ahrefs review of roughly 137,000 sites publishing one found about 97% got no measurable referral traffic from it — file it under cheap hygiene, not strategy. Being readable is necessary rather than sufficient, too: across 1.4 million real ChatGPT prompts, Ahrefs found the engine cites only about half the URLs it actually retrieves. Every honest FAQ, guide and price statement is one more thing it can quote you on, and that stock compounds.
A 90-Day Plan You Can Run Yourself
Start with a stocktake, not a purchase. Write down the ten questions your sales floor hears most and put them, word for word, into ChatGPT, Gemini and Perplexity. Are you named? Is a competitor named instead? Is anything said about you simply wrong? Run each prompt more than once, in a fresh chat, because the answers move. Worth knowing before you panic: Profound's look at roughly 730,000 US ChatGPT conversations found only about 18% trigger a web search at all, so much of what you see is the model working from memory rather than reading your site today. An hour of this tells you more about your position than any dashboard.
Then write for the gaps. One thorough, honest guide a week, each aimed at a question you failed. The topics are free, because they are the questions being asked at your counter, and the voice is already in the building: write the way your best salesperson explains things, patiently and without pressure. Do not build a parallel "AI version" of the site — one set of pages, written for people, is what Google asks for and what an engine can use.
In parallel, get the base signals straight: the business profile maintained, reviews asked for and answered, vehicle data structured properly, hours and address identical everywhere they appear. Then repeat the test at day 90. The change in what those three engines say about you is the most honest scoreboard available, and it costs an hour rather than a retainer.
The Honest Limits: What GEO Cannot Do
So nobody sets the wrong expectation: GEO is not a switch you flip. Content takes weeks to months to be picked up and used, and the pages engines lean on tend to be older than whatever you publish next — Ahrefs put the median age of a ChatGPT-cited page at roughly 500 days. Anyone selling you AI visibility by next week is selling you something else.
Equally honest: you do not control the sentence the machine produces. None of OpenAI, Google, Perplexity or Anthropic publishes a formula for which retrieved pages become cited sources, so anyone quoting you weighted signals is inferring rather than reporting. And these systems get things wrong with total composure. Columbia Journalism Review's Tow Center ran 1,600 source-identification queries across eight AI search tools and found more than 60% of the answers incorrect; ChatGPT misidentified 134 of 200 articles while hedging only 15 times and never once declining to answer. Assume the same confidence when it describes your financing terms.
The fair part: this applies to your competitors too, and the one thing you fully control is consistency. When your hours read Mon–Fri on one portal, Tue–Sat on another and daily on your own site, the machine does not flag the conflict — it picks one and states it flatly. The dealer who answers real questions and keeps a single version of the facts everywhere beats the one still producing glossy phrases.
Conclusion: Be the Answer, Not Just a Link
A car purchase now starts with an honest question put to a machine. Know those questions, answer them better than anyone else in your area, and you become the source the answer gets built from. There is no trick in it. It is the translation of what a good dealership already does: real advice, straight answers, expertise a customer can check.
Start small, but start. Ten questions, three chatbots, half an hour — that test tells you more about your next two years than any paid study. And the direction of travel is not subtle: SparkToro and Similarweb found 68% of US Google searches ended with no click to any website in early 2026, up from 60% in their comparable 2024 study. The clicks are not coming back. The answers are where the buyers are.
The upside is that none of this work is spent only on the algorithm. A guide explaining what a used-car inspection actually covers earns trust with the person reading it too, weeks before they set foot on your lot. In a business built on trust, that is a rare thing: an investment that pays twice.
Common questions
Which questions do car buyers ask the AI most often?
The pattern is situational rather than keyword-shaped: which car fits this budget and this family, whether an EV beats a diesel on my particular commute, lease or buy, and what my current car is worth. Alongside those sit the trust questions — is this dealer legitimate, what will year one really cost, what is the catch on a high-mileage used car. No public dataset breaks this down by market, so treat your own sales floor and your own prompt testing as the source, not a vendor's list.
How does my dealership become visible in ChatGPT or Gemini?
By being the best available source on the questions your customers actually ask. Publish honest, thorough guides on things like used-car checks and financing, keep the Google Business profile and your opening hours accurate everywhere, ask for reviews and answer them, and structure vehicle and site data so it can be read without guessing. The most evidence-backed lever points outward: Ahrefs found brand mentions across the web correlate with AI citation roughly three times more strongly than backlinks. Google's guidance is clear that no AI-specific markup or file is required, and this runs over weeks and months rather than overnight.
Is GEO worth it for a small, local dealership?
It pays off most for local dealers. Many car questions carry a place in them, and being cited is not the same contest as ranking: in Ahrefs' data roughly 80% of the URLs ChatGPT cites do not appear in Google's top hundred, so a national portal's SEO budget does not automatically win the answer. Clean location data, genuine reviews and content anchored in your region can establish you as the local authority, usually with less work than a large anonymous group would need. The real advantage is that most dealers in your area are still ignoring this entirely.
Read on