Measurement & Reporting · 9 min read · July 15, 2026

Where am I found? Checking AI answers town by town for painters

Someone in Rosenheim now types "who can paint my facade and actually turn up on time?" into ChatGPT instead of scrolling Google, and gets back one paragraph naming two or three firms. Google's own AI Overviews already reach more than two billion people a month. Whether your painting business is in that paragraph changes from town to town, so this guide shows you how to check it town by town and service by service, find the gaps and close them.

Why AI answers change from town to town

Almost every question that ends in a painting job carries a place name. "Painter in Ingolstadt", "facade coating Landshut", "who hangs wallpaper in Fürth?" – the town is part of the question, and that is where it gets decided whether the answer names you or a firm three villages over. The models are not recommending "a painter" in the abstract. For business queries all the major answer engines retrieve first and write second, pulling in directory entries, review text and pages from the open web before composing the reply. If nothing in those sources ties your name to that town, there is nothing to retrieve, and you are simply not in the answer.

The awkward part is that this visibility is patchy rather than even. It is entirely normal that ChatGPT names you every time for your home town and never once for the district 20 kilometers up the road, where you are just as happy to work. For a painting business whose vans cover half a dozen municipalities, that gap is money. Every town you are missing from is a town where the caller books someone else without ever having heard of you.

So a regional check means one thing: you stop treating your AI visibility as a single score and start measuring it per town and per service. Only when you know which town names you, for which question, can you see where your marketing is doing work and where nothing is landing. That location-precise view is the whole point of GEO for a trades business, and it really is not the same game as nationwide keyword rankings: in one Ahrefs analysis only about six to eight percent of the URLs ChatGPT cited also sat in Google's top ten for the same query, and roughly 80 percent of them did not rank in the top 100 at all.

The questions your customers actually type

Before you can measure anything you need to know what is actually being asked, and for painters the phrasing is refreshingly concrete: "good painter for interiors in Regensburg", "who does facade renovation with scaffolding in Kempten?", "painter to repaint an old-building apartment in Augsburg", "have mold painted over near Nuremberg", "firm for lacquering doors and frames in Ulm". Every one of them welds a service to an object type and a place. Rebuild that combination in your test list, or you will measure something your customers never ask.

So write down the services you actually earn money on and cross them with the towns you actually drive to. A painting business usually lands on ten to fifteen: interior painting, facades, wallpapering, lacquer work, floor coatings, drywall, mold remediation, exterior insulation, concrete repair, decorative finishes like textured plaster. Cross those with eight to twelve towns and you have a test list shaped like the way people look for you, not like the way you describe yourself in a brochure.

Include the blunt, unpolished wording too. Nobody types "facade coating services"; they type "who paints the outside of my house cheaply?", "painter who works Saturdays in Passau", or "fast painter for an apartment handover next week". That everyday phrasing is what the models are answering. Test questions that read like your price list will miss the queries that actually turn into work.

How to run the check, town by town

The simplest method is manual. Take your service-and-town list and put each question, one at a time, to ChatGPT, Gemini, Perplexity and Google with AI Overviews showing. For each answer, log whether you are named, roughly where in the answer, and which competitors sit next to you. It is dull work and it teaches you more than any dashboard. Twenty questions in, the shape is usually already there: named every time in your home town, most times in two neighboring ones, never in a third.

Ask each question in a fresh chat while logged out, so your own history is not quietly handing you the answer you were hoping for. Repeat the important ones on different days, because the answers move. Keep it all in one plain table: a row per town and service, a column per system, and in each cell "named / named, second / not named". Within a couple of weeks you own a map of your own visibility that shows in black and white where you are strong and where you are absent.

Doing this by hand across a dozen towns and a dozen services stops being realistic fast, and that is the point where a tool or a service that fires the queries on a schedule and logs the mentions starts to pay. The technology matters less than the fact that you keep going. One measurement is a snapshot. The same measurement every quarter is an early-warning system that tells you whether your own work moved anything, or whether a competitor is quietly taking towns off you.

Finding the blind spots worth fixing

With the table filled in comes the honest part. A blind spot is any town-and-service pair you want the work from and are named nowhere for. Painters keep finding the same uncomfortable shape: rock solid in the main town, invisible in exactly the leafy suburb full of big old villas where the high-margin facade jobs live. That is not a cosmetic gap, it is revenue walking somewhere else.

Then rank the gaps by what they are worth, because towns are not equal. A two-hundred-person village you visit twice a year matters less than a county seat half an hour out that could send you steady work. Mark the three to five town-and-service pairs with the most upside and start there, instead of spreading the same thin effort across twenty towns at once.

Then work out why the competitor is named and you are not. It is usually something you can see with your own eyes: a genuine page on his site for that exact town, reviews that mention the place by name, entries in regional directories. Nobody outside the labs knows the weighting – OpenAI, Google and Perplexity publish no formula for which retrieved pages become cited sources, and anyone selling you one is inferring. But the evidence trail is visible, and that is what tells you what to build.

Real jobs on paper beat template town pages

Models name a business for a town when they can find solid evidence tying the two together, so the strongest lever a painter has is real, place-specific work written down. Renovated a facade in Freising? Turn it into a short reference: the building, the job, the materials, before and after, two sentences on what the old plaster was doing. Name the town because it belongs in the sentence, not because you are stuffing it in. Those write-ups are the raw evidence a model can retrieve and quote.

Avoid the one big mistake: a batch of near-identical town pages with the place name swapped out. Google's own guidance on its AI features says the opposite of what the template sellers claim – no AI-specific markup or files are needed, and it warns explicitly against writing separate content "for AI", which risks being treated as scaled content abuse. Search engines and answer engines both spot the pattern, and it costs you more than it gains. A few real references per important town win instead: three honest project write-ups from Erding do more than twenty interchangeable blocks that would fit any town in the country.

Do the same with depth of service. "We paint all over Upper Bavaria" gives a model nothing to hold on to. "Interior painting and mold remediation in Dachau, Fürstenfeldbruck and the surrounding area, including advice on vapor-permeable paints in old buildings" hands it a service, a place and a piece of real expertise in one sentence. That is the kind of line a model can lift into an answer without guessing.

Reviews and listings are the strongest local signal

Reviews are the strongest local trust signal a painting business has, and answer engines read them. In an Ahrefs analysis of around 75,000 brands, how often a brand was mentioned across the web correlated with its AI citation rate at about 0.66 – roughly three times stronger than backlinks did. What matters for you is the place name inside the review text. "Mr. Bauer painted our apartment in Straubing on time and left it spotless" carries exactly the signal that makes you findable for Straubing, so it is entirely fair to ask happy customers to mention the town and the job, gently and without handing them a script.

Keep your Google Business Profile genuinely tidy, because it is one of the sources these systems lean on for local facts: full service list, correct hours, real project photos with captions that say where they were taken, a service area you have actually defined. Structured data on your own site helps in the same way, making facts machine-readable and cutting down on invented hours or prices, though Google has said since 2018 that schema is not itself a ranking factor. Then keep name, address and phone identical everywhere, because contradictory listings are exactly what produces a confidently wrong answer about you.

Local directories often outweigh the big anonymous portals: the painters' guild, the chamber of crafts, regional trade registers. A painter listed with his guild, appearing there with a town and a trade beside his name, is producing the independent regional confirmation an answer engine can lean on. It is unglamorous admin, and it is the ground your location-precise visibility stands on.

Turning the table into a quarterly habit

Measuring changes nothing by itself, so give it a rhythm. Once a quarter, put your most important town-and-service questions again and write the results into the same table. Then you can see movement: has a blind spot closed? Did a new town start naming you after you published two references from there? That trend is worth far more than any single reading, because it is the only thing that connects what you did to what changed.

Set one or two goals a quarter rather than trying to fix everything. For example: "by the end of the quarter I want to be named for 'facade renovation Rosenheim' in at least two systems." Then do the work that could plausibly cause it – publish two real Rosenheim references, collect three reviews that mention the town, put actual detail about facade systems on the site – and check at the end whether it moved. Give it time, though: in one Ahrefs study of 1.4 million ChatGPT prompts, the pages ChatGPT cited had a median age of around 500 days.

Pull your crew in, because the best material is made on site. The journeyman who has just finished lacquering an awkward old-building staircase in Bamberg has, with two phone photos and three sentences in a group chat, handed you the next reference. Organize that route from scaffold to website once and the regional evidence keeps building itself, because you are constantly producing new, real, place-specific proof.

What the measurement can and cannot tell you

Be honest about the limits. These answers are not stable, moving between days and between models, and they are not reliably right either. Columbia Journalism Review's Tow Center put 1,600 queries through eight AI search tools, asking each to identify a source article, and more than 60 percent of the responses were wrong – from 37 percent for the best tool to 94 percent for the worst – usually with no hint of doubt attached. So one missing mention is not a verdict on your business, and one flattering answer is not proof of anything either. Only a pattern, across several rounds and several systems, means something. Read your table as a trend, not as truth.

AI visibility also does not stand in for the work or for your name in town. One analysis of roughly 730,000 ChatGPT conversations found only about 18 percent triggered any web search at all; the rest were answered from what the model had already absorbed. That is the long game, and it is won by being mentioned widely and consistently over years – which happens when you turn up on the day you said you would, leave the place clean and get recommended. GEO amplifies real quality. It does not stand in for it.

The most useful advice is the smallest: start narrow. Take your five most important towns and three core services, run them once this week, and you will already know more about your regional AI visibility than almost every firm you bid against. The system grows from there. That head start is real for as long as most painting businesses still do not know they can appear in these answers at all.

Common questions

I work in ten towns. Do I need a separate website page for each one?

No, and a batch of near-identical town pages actively hurts you – Google treats scaled, templated content as spam. Real reference reports from real jobs in your most important towns work better. Three honest write-ups from one town, each naming the building, the service and the place, beat ten interchangeable pages. Rank the towns by job potential and work down the list.

How often should I re-run the check?

Once a quarter is enough to see a trend, plus a quick spot check after you have added new references or reviews for a town. Consistency beats frequency: the same questions, the same table, the same logged-out setup, so what you see is real change rather than the ordinary day-to-day drift in the answers.

Why does a smaller firm from the next town get named and I don't?

Usually because of visible regional signals, not size. Check whether they have place-specific references on their site, reviews that name the town and the job, a properly maintained Google profile and entries in regional directories. That is the evidence these systems retrieve. The good news is that all of it is buildable: how often a brand is mentioned across the web tracks AI citation rates far more closely than backlinks do, and mentions are something you can earn.

Read on

Strategy & Planning