Digitize Reality: How AI Learns What Your Dealership Is Actually Like

Your dealership does a lot of things that are real.

Your technicians hold certifications that took years to earn. Your service department has a shuttle that runs a set radius. You have a price guarantee with an actual dollar figure attached to it. Somebody on your team has been fixing these cars since the nineties. There is a children's play area in the waiting room because many of your service customers are parents.

All of that is true. Almost none of it is written down.

That is the gap we as GEO researchers spend most of our time closing, and we call it digitizing reality. It is the idea that LLMs can only describe your dealership using what they can read. If a thing exists in your building but not in your website copy, it does not exist as far as any web crawler is concerned.

Here is what we have learned about it from four studies of our own.

What Digitizing Reality Means

It is not writing more content. It is writing down what is already true.

Every dealership has a store of real, specific, verifiable detail that lives in people's heads and in the physical building. The processes. The timelines. The credentials. The perks nobody advertises. What it actually feels like to wait for an oil change there.

A shopper standing in your showroom picks all of that up in ten minutes. A crawler reading your website picks up whatever you typed.

The question is not "what content should we create." It is "what is already true about us that nobody has written down."

Proof One: The Pages AI Actually Reaches For

We audited one dealership's presence in ChatGPT across 290 prompts.

The most-cited source in the entire study was the dealership's own website. It appeared in 100% of answers, and it was the only source with that reach. Reddit came second at 86%, then CARFAX at 72% and DealerRater at 69%.

It goes further than that. Broken out by question type, the dealer's own site was cited in every single prompt in every single category. Six of six on brand and reputation questions. Eight of eight on product questions. Eight of eight on competitive comparisons. Seven of seven on niche questions. No other source came close to that consistency.

If you take one thing from this post, take that. Your website is not a bystander in AI answers. It is the base layer of nearly every one of them.

So we went and looked at which pages were doing the work. Two stood out, and neither one was a vehicle page.

The first was a service center page. Not a stub. Roughly 1,200 words describing what actually happens there, including a waiting area written out item by item: coffee and tea, soft drinks, snacks, wifi, phone charging, a business center, etc. Four different ways to not sit there while your car is worked on, named individually rather than gestured at.

The second was a why-buy page, and it is the better example of the two. A price promise with a specific dollar figure if they fail to beat a written offer. Complimentary tire rotations at set mileage intervals. A shuttle with its radius stated in miles. Lifetime towing. And this line, which is the most quotable sentence on the site: the service department's factory-trained technicians hold over 90 years of combined Volvo experience, broken out as a specific count of expert, master, and certified level technicians.

That is a real fact about real people that existed long before anyone typed it. But we crawlers can access it on this website and repeat it in their answers.

The pages doing the heavy lifting in AI answers are usually not the ones you think. They are the pages that explain what working with you is like.

This Is Not an Automotive Quirk

We wondered whether this was something odd about car shopping, so it is worth knowing that the same pattern shows up in a completely different industry.

Aleyda Solis, a well-known SEO, published an analysis in May 2026 covering 25 leading ecommerce sites across five US subverticals, using 90 days of AI citation and AI traffic data. Her finding on which pages get cited:

"Many highly cited ecommerce pages are not classic product or category pages. They are pages that help answer the user's decision question."

Size guides. Support articles. Policy pages. Buying guides. Not the product pages that everyone in ecommerce spends their entire budget on.

Translate that into a dealership and you get almost exactly what we found. The size guide is your service page. The policy page is your why-buy page. The pages answering the decision question are the ones getting used, in two industries that have nothing in common except that people research before they buy.

This is not a car thing, it’s how these systems work.

The About Us Dropdown Just Became a Strategy

Here is the part worth pausing on, because it overturns twenty years of how dealership websites get prioritized.

The About Us section of the navigation has always been the part nobody optimizes. Our team, meet the staff, why buy here, our facility, community involvement. You would make sure the copy was original so it was not duplicate, tick the box, and move on. It was hygiene, not strategy. Those pages did not rank for anything with volume and they did not sell cars, so nobody put real work into them.

That calculation has changed.

Those are the pages that describe what your dealership is actually like, which makes them the pages that answer decision questions, which makes them the pages LLMs reach for. The section of your website with the smallest budget attached to it is the section doing the most work in AI answers.

Proof Two: Reality That Existed but Wasn't Written Down

The clearest test of this idea came from a GMC dealership in Georgia, a town of about 16,000 people.

The store genuinely delivers vehicles to customers up to 200 miles away, free. It has done so for years. It handles the whole purchase online if you want. All true, all real, and all of it existed on the website only inside homepage banner graphics.

A picture of a sentence is not a sentence.

When shoppers in nine surrounding markets asked ChatGPT which dealer would deliver to their house, the store appeared in 1 of 27 answers.

So the team wrote it down. One page, stating how far, at what cost, with the exclusions and the eligible cities named in text. Nothing about the offer changed. Only whether it was readable to crawlers and wrote all the specific details about the offering down in text.

Eight days later the store appeared in 15 of 27 answers. A month later, 24 of 27. Eight of the nine markets now name the store on all three questions, including cities well outside its own backyard.

Same dealership. Same offer. Same trucks. The only thing that changed was that reality got typed out.

Proof Three: The Same Lesson on a Different Surface

We ran into an identical problem from a completely different direction with a Toyota dealership.

Their lease offers lived on designed graphics with the payment printed right on the image. Good-looking, effective for shoppers, and invisible to crawlers, which had no way to know which image on a page full of images was the one that mattered.

Once the markup told Google which image was the main one, those pages went from zero clicks in Google Images to real traffic inside a few months.

Different surface, different fix, same underlying problem. Something true about the business was locked inside a picture.

Audit your banners. Whatever your best offer is, if it only exists as a graphic, no machine has ever read it.

Proof Four: Changes by Platform

Everything above comes from ChatGPT. We also ran a separate study on Google's AI Overviews, and it is worth knowing that the two do not behave the same way.

AI Overviews are generated by a language model too, but they sit directly on top of Google's search index in a way ChatGPT's answers do not. That difference shows up in the data.

Across all US Volvo dealerships, 79% of AI Overview responses cited at least one dealership, and the top 10% of dealers captured 61% of all citations. A small group takes most of the visibility, which is the same winner-take-most shape you see in organic search.

The predictor is what matters most here. The strongest single indicator of whether a dealership got cited in an AI Overview was whether it already ranked in the top 3 organically, at a correlation of 0.869. 

So on that surface, ranking is the entry ticket. Digitizing reality is what makes you usable once you are through the door, and it does not substitute for ranking there.

On ChatGPT we saw something different, where the dealership's own website was cited in every answer regardless of where it sat in Google. Different gate, different work.

Be careful with anyone who talks about "AI visibility" as one thing. Getting named in ChatGPT and getting cited in an AI Overview are related jobs with different entry requirements, and a strategy built for one will not automatically deliver the other.

What To Digitize First

Start with what is true and unwritten. In rough order of value:

Credentials with counts. How many technicians, at what certification levels, with how many combined years. Master techs by name if they will let you.

Policies with numbers. Shuttle radius. Loaner terms. Warranty specifics. Price guarantees with the actual figure. Delivery distance and cost.

Process and timeline. What happens between dropping a car off and getting it back. How long the average service visit takes. What the paperwork actually involves.

Real-world footprint. Awards with the awarding body and the year. Manufacturer recognition by its proper name. Certifications. Community involvement, sponsorships, how long you have been in the building.

The physical experience. What the waiting room has in it. Whether there is somewhere to work. Whether the kids will be okay.

None of this requires you to invent anything. It requires somebody to walk the building with a notepad.