Strategy & Planning · 9 min read · July 15, 2026
Winning the class of 2027: an AI-visibility plan for multi-location driving schools
The class of 2027 does not open with Google. It asks an assistant which driving school near it is any good, and largely trusts the reply. If your school is named there, you have the learner before a competitor knows the conversation happened. Generative engine optimization is now part of running a multi-location driving school, and it rewards different things than SEO does.
How the class of 2027 actually searches
Your next intake is already researching, and Google is not where the research starts. The first question about a licence goes into ChatGPT, phrased more or less like this: 'I live in Regensburg, what does a class B licence cost and which driving school should I go to?' What comes back is a short recommendation, usually with names in it. If yours is not one of them, you do not exist for that teenager, however many five-star Google reviews you have collected.
That is the break with how this used to work. A prospect once saw ten blue links and did the comparing themselves. Now the engine does the comparing and hands back a shortlist. Pew Research, using real browsing data from 900 US adults across 68,879 searches, found people clicked through to a website in 8% of searches where an AI summary appeared, against 15% where none did, and clicked a link inside the summary itself in just 1% of visits. For a school with branches in several towns the consequence is sharp: being findable in one place is not coverage. You need to be named per city and per district, or you lose the learners who live closest to your door.
The class of 2027 is also the opening. It is the first cohort to have had an assistant on hand for its whole school career, and most driving schools in your area are still buying search ads and waiting for the phone. Start now and your name is already sitting in those answers by the time this cohort is eligible to book a lesson.
What GEO means for a driving school
Generative Engine Optimization, or GEO, is the work of shaping whether and how an answer engine names your school. It is not a rebadging of SEO . The overlap is thinner than most people assume: Ahrefs found only around 6 to 8 percent of the URLs ChatGPT cites also sit in Google's top ten for the same query, and roughly 80 percent do not rank in Google's top 100 at all. Engines build answers from what the model already absorbed, from review sites and trade directories, and, when they do search live, from whatever page they can retrieve and quote. Your job is to leave the same set of facts in all of those places.
For a driving school those facts are unglamorous and very specific. Which branches do you actually run? Which classes do you teach at each one, from B up through A, the BE trailer, C for trucks? Do you run intensive courses, holiday courses, automatic-only training, lessons for people who are frightened of driving? Are eyesight tests and first-aid courses done in house or sent elsewhere? These are the questions prospects put to an assistant, and these are the answers that have to exist somewhere a machine can read them.
GEO does not replace your Google presence, it feeds on it. Google says its own AI features are rooted in its core Search ranking and quality systems, and that no AI-specific file or markup is required to be eligible. So a current Business Profile per branch, real opening hours and a price that matches the one on your own site are the raw material. Being sloppy here does not merely go unnoticed. It manufactures conflicting facts, and conflict is the one thing that reliably stops a language model from naming you at all.
One brand, several branches: the location problem
A single-site school never meets this problem. With several branches, the engine has to work out that one brand name in Cologne-Ehrenfeld, Cologne-Sulz and Bergisch Gladbach is one business at three separate addresses. Tip all of it into a single 'our locations' block and the picture blurs, and you get named less often for exactly the questions that carry a place name.
The fix is structure. Every branch gets its own page: its address, its hours, its named contacts, and only the classes you genuinely teach there. Not every site runs motorcycle class A, and that difference is precisely what has to be legible. An engine asked about motorcycle training in Bergisch Gladbach should be able to tell whether you teach it there or only at the main office.
The usual failure is the cloned branch page: one text five times with the place name swapped out. It reads as filler to a person and gives a machine nothing to tell the pages apart. Write the local detail instead. The test routes examiners actually use, the junction everyone dreads, where a learner can park near the office. That is the material a specific, place-level answer gets built from.
Reviews and mentions: your strongest signal
Nothing you can influence moves an answer engine as much as what other people say about you. An Ahrefs analysis of around 75,000 brands found that how often a brand is mentioned across the web correlates with AI citation at about 0.664, roughly three times the correlation for backlinks at about 0.218. Profound's read of roughly 730,000 US ChatGPT conversations explains part of why: only about 18% trigger a web search at all, so most answers come from what the model already absorbed. For a driving school, most of that absorbed material is reviews. The star average matters less than the wording. When students keep writing 'patient instructor', 'passed first time' or 'flexible about lesson times', those phrases become the way you get described.
So make the asking systematic, branch by branch. Every student who passes gets asked, and gets asked for detail: which class, which branch, what actually helped. A line like 'class B at the Sulz branch over the summer holidays, the intensive course was worth it' does far more work than five stars and no text, because it settles three things a prospect would otherwise have asked separately.
Answer the bad reviews too, plainly and with a fix attached. The engine reads the exchange, not only the complaint, and a straight reply to criticism reads as a business worth naming. Silence underneath a one-star review is also a signal, and not one you want repeated back to a parent.
Make the site readable by machines
Machines guess less when a page labels itself. That is the whole point of structured data. Schema markup sits in the page source and states outright what a human reader can see: this is a driving school, these are its branches, these are the hours, this is the price. DrivingSchool or LocalBusiness, marked up once per branch, covers most of it. Be honest about what it buys you, though. Google has said since 2018 that structured data is not a direct ranking factor. What it does is make your facts easy to extract and hard to garble, which is how you stop an engine inventing a price or a closing time on your behalf.
An FAQ block on each branch page earns its keep for the same reason. Answer the things teenagers and their parents actually type: what a class B licence costs at that branch, how long the training takes, whether you can start at 17 with accompanied driving, whether automatic-only training is an option. Answer them in plain sentences on the page itself, and the page becomes quotable.
Keep prices and terms current and identical everywhere. If your homepage carries one figure, your printed price list another, and your Google profile still shows last year's, you have manufactured exactly the contradiction that gets you dropped from an answer. Faced with three different numbers, an engine reaches for the competitor whose figures agree with each other.
Write the pages an engine can quote
Learners research for weeks before they sign anything: how the process works, what the theory test is like, what the whole thing costs, what to do about nerves. Pages that answer those questions properly are the pages that get retrieved and quoted. An honest guide to accompanied driving from 17, or to passing theory first time, positions you as the school that knows the subject rather than one more listing in a directory.
Specificity is the whole trick. Generic explainers exist in their thousands and get passed over. Local detail is scarce: the three hardest test routes in Augsburg, why so many candidates come unstuck at the north roundabout. Only a school that teaches there can write it, and that scarcity is exactly why it gets picked up.
Write for the parents as well. A good share of the first research is done by a mother or father checking the cost and whether you look serious. An article that sets out honestly what a licence costs today and what makes up that total earns trust from the person paying, and at the same time hands an engine figures it can cite instead of guess at.
Measure what the engines say about you
You cannot steer what you never check. Once a month, put your customers' questions to the main assistants yourself: best driving school in Cologne-Ehrenfeld, where to get a motorcycle licence in Bergisch Gladbach. Write down whether you appear, how you are described, and who appears in your place. It takes an hour and it is the only honest picture of where you stand.
Read the description closely. Are you 'the cheap one', 'good with nervous learners', 'the intensive-course school'? Those adjectives tell you which signals actually arrived. Where they miss the positioning you want, you know what to write next and which reviews to ask for. Check the plain facts hard too, because these systems state wrong things with total confidence. Columbia Journalism Review's Tow Center ran 1,600 queries across eight AI search tools and found more than 60% of the answers incorrect; ChatGPT misidentified 134 of 200 articles while flagging uncertainty only 15 times. A branch you closed two years ago can keep being recommended until you correct the sources underneath it.
Log the results per branch in one unglamorous spreadsheet, month after month. That is how you tell a real shift from a fluke. Visibility in these systems accumulates over months of steady maintenance, not over a launch week. If autumn 2027 is the target, those months are the ones you still have.
A roadmap to autumn 2027
Start with the dull part. Over the next few weeks, get every Google Business Profile straight: address, opening hours, the classes taught there, photographs, the price you actually charge. Then walk your own site and make every figure agree with it. Nobody will congratulate you for this step, and everything after it depends on it. It costs attention rather than budget.
Then build out the branch pages and the review habit. Each branch gets a genuinely local page with its own FAQ. Every student who passes gets asked for a specific review, not just a rating. Add local guide pages steadily, one real question at a time. If the time or the know-how is not there, hire someone who has done this for local businesses; the cost is small measured against a single cohort won.
After that it is maintenance. Check your AI visibility every month, keep the reviews arriving, and change a price everywhere on the day it changes. GEO is a habit rather than a project with a completion date. Keep it up and by autumn 2027, when the new cohort starts asking which school to book, you are the name in the answer, chosen before a competitor has noticed the ground moved.
Common questions
Is a well-kept Google profile enough on its own?
It is the foundation and it is not sufficient. Answer engines pull from several places at once: reviews, your own pages, structured data, forums and directories. They name you confidently when those sources agree with each other. Google's guidance is clear that a page needs only to be indexed and eligible to show with a snippet to be usable as a supporting link, so no AI-specific file will rescue you. A price that contradicts your website, or a branch listing that is a year out of date, keeps you out of the answer regardless of how tidy the profile looks.
How do I get an engine to tell my branches apart?
Give every branch its own page with its own address, hours, contacts and the classes genuinely taught there, and never clone one text with the place name swapped. Add detail only a local would know, like the routes examiners use or the junction that catches people out. On the technical side, mark each branch up as its own DrivingSchool or LocalBusiness entity. That will not lift a ranking, but it makes each place unambiguous and much harder for an engine to garble into one blurred address.
How long before this shows up in answers?
Months, not days. Engines rebuild their picture of you gradually, through reviews, site changes and whatever they retrieve when they do search. Ahrefs, looking at 1.4 million real ChatGPT prompts, found the pages it cites had a median age of around 500 days, which tells you plenty about the timescale. That is the argument for starting now: to be in the answer by autumn 2027, the data cleanup, the branch pages and the review habit have to begin this quarter. Monthly checks are how you tell whether it is working.
Read on