GEO for local governments: getting your town hall cited by ChatGPT and AI search

Something quiet shifted in the way people deal with their town hall. Since 2024, when Google began rolling out its AI Overviews, the entry point moved: where a resident used to phone the switchboard or scroll the municipal website, many now open a chat window and ask in plain words: "how do I register my child for school in [town]?", "when does the market open on Saturday?", "what documents do I need for a passport in my commune?"

The AI answers in a sentence or two. And here is the catch for a public body: the model will answer whether or not the information comes from you. If your official page is unclear or hard to reach, the AI reaches for whatever it can find, an old aggregator, a forum thread, a cached page from three years ago, or it fills the gap on its own. The resident then acts on an answer you never wrote. That gap, between what your town actually offers and what the AI says it offers, is exactly what GEO applied to local governments works on.

GEO for a local government, what it means

GEO for a local government means making your municipality, town hall or region the source generative AI tools cite when a resident asks about a public service, instead of leaving the answer to a stale site or a hallucination.

The term Generative Engine Optimization was coined in 2023 by a team of researchers from Princeton, Georgia Tech, the Allen Institute and IIT Delhi, in work presented at the ACM SIGKDD conference. Their finding holds in one sentence: a generative engine no longer returns a list of pages, it writes an answer from several sources and chooses which to cite according to measurable criteria. In their study, applying GEO methods lifted a source's visibility in generative answers by up to 40% (Aggarwal et al., 2023-2024), a gain that was real and reproducible.

Applied to the public sector, that turns into very concrete work: making sure that when someone asks an AI about a service you run, your official information appears as the answer, correct and traceable back to you. This is not keyword stuffing. A local authority carries a duty of clear, accurate public information, and GEO extends that duty to the surfaces where residents now ask their questions. It means taking care of the sources that describe your services (your official pages, your practical information, your public profiles, your structured data) and publishing the few pieces of content that honestly answer residents, newcomers and local businesses. That is precisely what Cicero Studio does: GEO audit, editorial production and automated semantic meshing, in service of your visibility on Google as well as in AI engines.

Why town halls and regions are concerned too

The reflex of "I ask an AI before I call the town hall" is gaining ground for public-service questions, from an opening time to a full administrative procedure. When the AI answers in the resident's place and gets it wrong, the confusion, and the extra calls, land back on your services.

"A town hall has nothing to do with ChatGPT." I hear it in almost every first conversation with an elected official or a communications manager. Public service runs on official channels, front desks and paper, the reasoning goes. And those channels do still matter. But a new habit is settling in alongside them, and it targets exactly the everyday questions that make up the bulk of a town hall's incoming calls. Since 2024 Google has rolled out its AI Overviews, those generated answers shown at the top of results, then added a fully conversational AI Mode built for open questions, of the "how do I do X in my town" kind, as Google documents in its guidance on AI features. In parallel, assistants like ChatGPT now browse the web and cite pages in real time. The context is favourable: according to the Baromètre du numérique, 94% of French people connect to the internet and the use of AI tools is rising sharply. For a local government, the stake becomes clear: being the source the model reads, so the answer is right and it comes from you.

The landscape is also vast, which makes findability both harder and more valuable. France counts around 35,000 communes, alongside its départements, régions and thousands of intercommunal bodies, according to INSEE. In a field that dense, an AI that answers "the town hall of [name] is open until 5pm" has to pick a source, and it will pick the clearest one it can trust, not necessarily the official one. There is also a hard truth on the supply side, one we measure daily: in our base of 50,000 French organization websites, only 65% are genuinely live and 8% carry fewer than 20 indexable pages (Cicero Studio internal analysis). Plenty of municipal sites sit in exactly that thin zone, a nice banner photo and little findable detail, which gives the model very little reliable material to cite.

In the GEO audits we ran for local structures and public-facing organizations, we checked this point query by query, and a pattern keeps coming back: public-service questions are high-stakes for accuracy. The resident wants the right hour, the right document list, the right contact, and a wrong answer costs a wasted trip or a missed deadline. When the AI answers directly and confidently, most people act without double-checking. If the source it used is not yours, or is out of date, you inherit the fallout at the counter. This shift first touches everyday practical queries, those that combine a service and a place. We unpack the local dimension in our analysis of the impact of AI on local search, and the spread of AI answers across ordinary queries is now well documented, as we cover in our piece on AI Overviews on business queries.

The public-sector queries shifting to AI

The public-sector queries shifting fastest to AI combine a service and a place ("how to do X in [town]"), or ask for practical facts (hours, contacts, documents). These are the ones where a wrong AI answer costs a resident a wasted trip, so the ones you most want to own.

Not all searches migrate at the same pace. For a local government, here are the intents to watch first, because these are where the AI answer weighs most on a resident's day, and on your front desk.

The common thread of these queries: the resident wants a reliable, actionable answer, not ten links to sift. That is exactly the format where a generative AI adds the most perceived value, so the one where it has taken over most. To be the cited source, your local government has to give the model something clean to read: a clear service page per intent, accurate practical information, a public profile that matches your site, and structured data that says plainly you are the official body. When that material exists and is consistent, the AI has every reason to prefer you over a random aggregator.

The four levers that make a local government citable

Four levers stand out for a local government: clear service and procedure pages (one intent per page), accurate practical information consistent across your site and profiles, structured data identifying you as an official public body, and local content that documents your area. AI tools reuse clear, verifiable, official sources.

Where do you start when you run a small communications team, or none at all? It is the first question I get from town halls. From experience, here is the order I recommend, from most profitable to longest term. And a small warning before the list: none of this is magic. I have seen a set of tidy, current procedure pages cut a town's repetitive phone calls noticeably, where a redesigned but shallow website had changed nothing. The difference was never the visual. It was the clarity and the accuracy.

1. Clear service and procedure pages, one intent each

This is the foundation. For every common step a resident takes (ID card, passport, school enrolment, civil status documents, permits), a dedicated page that answers in plain words: who it is for, what documents are needed, where and how to do it, how long it takes. One intent per page, no jargon, the answer up front. A single sprawling "your steps" page that buries ten procedures gives the model almost nothing to extract cleanly. Pages built this way are also exactly what ranks on Google, so you serve both surfaces at once.

2. Accurate practical information, consistent everywhere

Opening hours, addresses, phone numbers, closures: the facts an AI answers with directly. The lever here is not just having them, it is keeping them consistent across your website, your Google Business Profile and any public directory. When your site says 5pm and an old listing says 6pm, the model has no way to know which is right, and may pick the wrong one. A single source of truth, kept current, is what lets the AI answer confidently from you.

3. Structured data that says "official public body"

Behind the scenes, structured data tells engines what your pages are and who publishes them. Google's Search Central documentation on Organization structured data shows that explicit, consistent information about your name, your nature and your contacts helps engines describe your organization accurately, an official local authority rather than a lookalike site. Marking up your pages this way is a quiet but real advantage: it raises the model's confidence that it is citing the genuine source.

4. Local content that documents your area

Beyond procedures, the questions about local life (events, projects, transport, why settle or set up a business here) are the ones that shape a town's image in AI answers. A page that honestly documents what happens in your commune (a market, a cultural season, a local support scheme) gives the model material to describe your area with substance rather than clichés. Google confirms the broader principle in its documentation on AI features: it is helpful, original, well-structured content that makes a page eligible for generative experiences. For a region or an intercommunal body chasing attractiveness, this lever is often the one with the highest ceiling.

The Cicero Studio method for local governments

Cicero Studio brings together three building blocks for a local government: a GEO audit that measures your citability and the accuracy of AI answers about your services, editorial production that fixes your practical information and creates your service, procedure and local-life pages, and automated semantic meshing that ties it all together.

Our starting conviction is simple. AI visibility cannot be decreed, it is built source by source, with method. So how do we actually go about it for a public body? Here is the order, no jargon.

1. GEO audit, measuring on your resident queries

We query AI engines on the real questions residents ask about your town (procedure + place, practical facts, local life, settling in, business support) and we record, query by query, whether your local government is cited, whether the answer is accurate, and which source the model used in your place. From this we draw a starting point and a map of the information to fix or create first. This audit is free and with no commitment. Our GEO audit method details each criterion we evaluate, and our page on AI visibility for an organization sets out what counts for a body whose job is trustworthy information.

2. Editorial production, fixing and creating

We first put your practical information back in order and make it consistent, then write the content that answers your residents' questions directly, in the format AI tools cite: a clear answer up front, the concrete steps, the exact documents, an honest scope. For a local government, that means clean service and procedure pages, current practical information, and a few local-life pieces that give your area substance. Each page is also built to rank on Google and Maps: we do not separate the two. The editorial quality stays that of an agency, the pace is that of a tool. That is what we sum up as "agency-quality work, software-grade productivity".

3. Automated semantic meshing, making the whole thing work

An isolated procedure page is fine. A network that links your service pages to your practical information, your local-life content and your profile, where each page points to the others and signals your official status, is what makes a local government hard for AI to get wrong. On our own content programs (1,849 tracked pages), the average search position we hold is 10.3 across the board (Cicero Studio internal analysis, aggregated and anonymized), and that steady presence owes a lot to this structural meshing. We organise your pages into thematic clusters (a pillar that frames a domain such as civil status or local life, satellites by procedure and by topic, natural internal links) and we maintain this meshing automatically as new content ships. Our GEO vs SEO page explains why this structural work serves both Google and the AI engines at once.

This automation is not a gimmick: it is what lets us hold a pace a traditional agency would bill much more for, without sacrificing human writing or fact-checking, which matters doubly for public information. For a small team juggling communications alongside a dozen other duties, that is precisely what makes the thing sustainable.

Common mistakes among local governments

The three most common GEO mistakes for local governments: letting practical information drift out of date, burying procedures in a single unclear page, and treating the website as a brochure rather than a service. All three reduce both citability and local visibility.

• Letting practical information drift out of date. Wrong hours, an old phone number, a closure never announced: this is the first thing an AI gets wrong, and the first thing residents notice. It is also the easiest to fix, and the highest return.
• Burying procedures in one unclear page. A single wall of text covering ten steps gives the model nothing clean to extract. Splitting it into one clear page per procedure, answer first, changes everything for both the AI and the resident.
• Treating the website as a brochure. A site built around the mayor's welcome message and pretty photos, with no findable service content, does not help a resident act or a model cite you. The website is a service, not a leaflet.
• Spinning up empty pages. Creating dozens of near-identical thin pages to "cover" topics is counterproductive: Google and AI alike detect the padding. A few genuinely useful, current pages are worth far more.
• Expecting a switch to flip overnight. Assuming a redesign alone will fix AI answers is a trap: it is the clarity, accuracy and structure of the content that move the needle, over a few months, not the visual refresh.

At bottom, the right lens is still that of serious, current, resident-first work. To see how an organization becomes a source AI cites, read our method on getting cited by ChatGPT, and to broaden the picture across engines, getting cited by Perplexity and our guide to optimizing for Google AI Overviews. For a step-by-step walkthrough on a single page, our practical method to appear in ChatGPT and AI Overviews shows the mechanics in detail.

What GEO will not do for your local government

For honesty, and because this is exactly the kind of transparency AI tools reward, here are the limits to know before getting started.

GEO is powerful for a local government with real services to explain and current information to share. For a structure with no minimal online presence and no clear service pages, no optimization will work miracles: it is first a job on the practical information and on the procedure pages. AI surfaces in Europe also operate within a clear regulatory frame, set out in the EU regulatory framework for AI, which is worth keeping in mind as these tools handle public information and residents' questions.

Alexis Dollé

CEO & Founder of Cicero Studio

Growth and SEO content strategist, I founded Cicéro to help organizations, and public-facing bodies like local governments in particular, build lasting organic visibility, on Google and in AI-generated answers alike. Every page we produce is designed to give the resident the right information from the right source.

LinkedIn →

Resources to go further

We document our approach publicly, it is our best proof. When a communications manager asks me where to start, I always point to these pages before the first call: they let you understand the logic of GEO and judge for yourself, without jargon. The pillar frames AI visibility in general, the GEO vs SEO page clarifies what really changes for you, and the practical guides show the concrete mechanics, query by query.

Here are the most useful pieces for understanding a local government's visibility in AI engines:

French-speaking reader? Start with our French pillar on agence GEO, then our field guide on the impact de l'IA sur la recherche locale.