Your Buyer Asked ChatGPT First. Your Website Still Opens With a Form.

The number that walked in the door

A buyer walks into a German showroom having already asked ChatGPT what a fair price is for a 2023 Peugeot 308 with 40,000 kilometres. They have a number. They trust it. They are ready to be disappointed if yours is higher.

That number might be right. It might also be an average pulled from a different market, on a different configuration, from data that is eighteen months old. The buyer does not know which, and here is the part that matters: it does not feel like an estimate to them. An AI answer arrives with no “maybe” and no “it depends.” It reads as a fact, and they walk in treating it as one.

This is the buyer European dealers are now selling to. Not a different person, but the same person arriving in a different state. Better informed, more certain, and quicker to treat the first price they hear that contradicts their AI answer as evidence the dealer is working an angle.

The instinct is to worry about whether the AI number is accurate. That is the wrong thing to worry about. The thing to worry about is what your first digital touchpoint says back to a buyer who arrives in that state, because for most brands the answer is: a contact form.

The data, and why the German number is the one to watch

The German car dealers’ association, the VAD, ran a study this spring that is worth reading closely, partly because it is one of the few large neutral samples on this and partly because it resists the easy headline.

Between mid-February and early March 2026, puls Marktforschung surveyed 2,393 German car buyers and prospective buyers on behalf of the VAD. The core finding is not the one the AI-hype cycle wants. It is that 89% of buyers research online and then go to the dealer for advice, a test drive, and the purchase. Only 11% buy fully online, and that split holds steady across every age band from 18 to 75.

Inside that online research, AI is still a small slice. 9% of buyers consult it while orienting themselves; 2% use it for the final choice. The dealer and salesperson remain the single most important touchpoint, for both first preference and final decision, far ahead of any AI tool.

So the easy story, that AI is about to disintermediate the dealer, is not what the data says. The dealer is not being replaced. The dealer is being approached differently. That is a smaller claim and a more useful one, because it is actually happening now rather than in a keynote about 2030.

89% research online, then buy through the dealer. Only 11% buy fully online.

VAD-Studie 2026, puls Marktforschung for the Verband der Automobilhändler Deutschlands, n=2,393. Stable across all age groups, 18 to 75.

The paradox hiding in the same study

The most important number in the VAD study is not about AI at all. It is about the gap between what buyers want and what they get, and AI is making that gap more expensive.

Buyers were asked what matters to them and how satisfied they are with it. The two things they value most are the two things they are least satisfied with. 72% say cost advice is important; 51% are satisfied with the cost advice they receive. 75% say trust in the salesperson is important; 56% are satisfied. The topics that carry the most weight are exactly the ones where the experience most often falls short.

Now put the AI buyer on top of that. A better-informed buyer does not lower their demands for honesty and precision. They raise them. They arrive with a price anchor and a sharper sense of what a straight answer sounds like, into a conversation that was already underdelivering on cost transparency and trust before AI entered the picture.

This is the real shift, and it has nothing to do with whether ChatGPT quotes the right figure. The buyer’s expectations went up. The first touchpoint most dealers offer stayed exactly where it was.

What a form says to a buyer who already did the work

Picture the sequence from the buyer’s side. They have spent an evening researching. They have a shortlist, a configuration, and a price in their head that an AI stated with total confidence. They land on the dealer’s website at 22:00, ready to move.

What greets them is four fields and a submit button. Name, email, phone, message. The form does not know they have already chosen the model. It does not know their price anchor, or that it might be wrong, or that this is the moment to have that conversation while they are still open to it. It collects four pieces of data and promises a callback.

By the time that callback comes the next afternoon, two things have happened. The buyer has configured the same car on two competitors’ sites, and the window in which their AI price anchor could have been calmly reframed has closed. They are now more attached to the number, not less, because nobody engaged with it while it was still a question.

A form is a reasonable tool for a buyer at the start of their journey. It is the wrong tool for a buyer at the end of theirs, and AI is pushing more buyers to arrive at the end.

Meeting an informed buyer where they actually are

The fix is not a better form. It is replacing the form, at least for buyers who signal real intent, with a conversation that can do what a form cannot.

A conversation can ask where the price came from. If a buyer opens with a figure, the useful response is not to counter it, which reads as exactly the dealer manoeuvre they were braced for. It is to ask: did you get that from ChatGPT, or a pricing site? That single question turns a standoff into a shared problem, and it is a question a well-built conversational layer can ask at 22:00 without a human being awake.

A conversation can add the context the AI number is missing. Bare list prices compared online rarely account for trade-in, warranty, delivery, or service. Making those explicit does not argue with the buyer’s number. It gives the number a frame, which is what closes the gap between the AI estimate and the real one.

And a conversation can qualify the buyer while it does all this, so that when a human does step in, they arrive knowing the model, the anchor, the trade-in, and the timeframe. The buyer who did their homework should not have to do it again for your BDC. That is the fastest way to confirm their suspicion that the dealer was not really listening.

This is the qualification layer of the model we published in The Automotive Conversion Engine, where the handoff from a scored conversation to a briefed human is the joint most European networks still run on office hours.

Why the human matters more, not less

It would be convenient, selling what we sell, to argue that the informed buyer wants a fully automated purchase. The data does not support it and neither does experience.

The VAD study is blunt on this: the salesperson is the decisive touchpoint for 45% of first preferences and 26% of final choices, against 9% and 2% for AI. A separate panel of European leasing executives put the same point more plainly earlier this year, that AI produces better decisions and scenarios people would not think of on their own, but the final decision stays with people and the personal relationship cannot be lost.

On a purchase this size, the presence of a competent, straight-talking person is not a nice-to-have. It is the thing the buyer is actually evaluating, underneath the price. AI has raised the bar for what “competent and straight-talking” has to look like, because the buyer now has a second opinion in their pocket. It has not removed the need for the person. It has made the person’s first thirty seconds matter more.

We went deeper on where AI should qualify and where a human has to close in Chatbot vs AI Sales Agent.

The uncomfortable version, for European brands specifically

Most of the loudest data on AI car buyers comes out of the US, and it tends to frame the informed buyer as a threat to margin. The European picture is different and, if anything, more actionable.

European buyers still overwhelmingly want the dealer. The relationship, the test drive, the face-to-face. What they are bringing to it is a higher standard of preparation and a lower tolerance for a first touchpoint that treats them like a cold lead. The brands that win here will not be the ones with the cleverest AI. They will be the ones whose first digital interaction is good enough to match a buyer who spent last night getting good.

That is a solvable problem, and it is solvable now, on the site you already have. The buyer already did the hard part. The only open question is whether the first thing they meet on your website is ready for them.

Our own research points the same way as the VAD data. The Car-Buyer Experience Report 2026 surveyed 310 German car buyers on how they research, where the booking process breaks, and what actually decides the purchase.

Download the Car-Buyer Experience Report 2026

Or if you want to see what an informed buyer meets instead of a form, book a demo and we will run one through your own site.