Nearly one in two Google searches in France has local intent, a volume representing more than a billion searches every month. For a long time, the recipe for capturing a share of them was well known: a well-filled Google Business profile, a few customer reviews, and a dedicated page for each targeted city.
That recipe is no longer enough. AI answer engines (ChatGPT, Perplexity, Google's AI Overviews) now answer your prospects' local questions directly, without necessarily sending them to your website. And the technique most used by businesses to become visible locally, multiplying “agency in [city]” pages, is starting to backfire.
We discovered this ourselves recently, on our own website. This guide covers what we learned, with concrete levers you can apply today.
What has changed in local SEO
Local SEO is no longer limited to Google Search and Google Maps. In 2026, generative engines also handle local queries, and they don't work like a classic search engine.
Where Google used to display ten links to choose from, a generative AI often gives a single answer, synthesized from a few sources it deems reliable. According to Gartner projections, up to a quarter of traditional search traffic could shift to chatbots and AI assistants. And the “zero-click” search phenomenon, where users get their answer without ever clicking a link, would already account for 60% of searches according to several recent SEO studies.
For a business with local customers, this shift changes the nature of the game. Being visible is no longer enough: you have to be cited. An AI answering “which plumber should I call in Lyon” or “which web agency should I choose in Bordeaux” will only cite a handful of sources, the ones it judges most reliable, clearest and most up to date.
Why “agency in [city]” pages no longer work
This is where many businesses (us included) got trapped. The practice of duplicating a page for each targeted city, changing only the city name in otherwise identical content, has been a local SEO classic for years.
The problem: Google, and by extension the AI engines that rely partly on the same quality signals, now considers this type of page to be doorway pages: pages created solely to capture traffic on keyword variations, with no real added value for the visitor. It's a practice explicitly targeted by Google's anti-spam rules.
The consequence isn't just a poor ranking. In our case, these pages were simply not indexed: Google crawled them, noticed they offered nothing more than the page next door, and chose not to show them in search results.
Our own experience: the trap we spotted on our own site
We had several pages of this kind on agencenorry.com, targeting cities where we wanted to grow our visibility (Bordeaux, Lyon, Marseille, Paris, Lille), in addition to Poitiers, where we are actually based. When auditing our own Search Console, we realized that none of these pages had been indexed since they were created.
Digging in, the reason was clear: near-identical content from one city to the next, with the same testimonials, the same structure, the same pitch, only the city name changed. Exactly the pattern Google flags as irrelevant.
The fix wasn't to delete everything at once. We sorted them out:
- Poitiers, where we have a real physical presence, was kept and enriched: explicit mention of our local base, an invitation to meet in person, verifiable elements rather than declarative ones.
- Paris and Lyon, where we have real commercial potential without a physical presence, were reworked with honest, differentiating content, without claiming a base we don't have, but explaining concretely how we support clients in these areas remotely. We are currently testing whether this “lighter but differentiated” version is enough to get indexed.
- Bordeaux, Marseille, Lille, where we had neither a presence nor a strong differentiating argument, were removed and redirected to the home page rather than continuing to pollute our indexing with duplicated content.
This choice is directly inspired by what we observe at players more established in this kind of multi-city strategy: even slight differentiation (a unique title, a real anchor point) can be enough when the site's overall authority is already strong. For a younger site, that margin is thinner, hence the value of prioritizing rather than spreading the effort across five cities at once.
Old reflex vs. new reflex: the comparison
| Old reflex (classic local SEO) | New reflex (local SEO + generative AI) |
|---|---|
| One duplicated page per city, only the name changes | A page per city only where there is real legitimacy, with differentiating content |
| Generic claims (“regional leader”) | Concrete proof: case studies, named clients, quantified results |
| A Google Business profile created then forgotten | A profile maintained continuously: photos, reviews, hours, updates |
| Content published once, never touched again | A review at least annually, every six months on strategic pages |
| No structured data, or minimal markup | Complete Schema.org markup (Organization, LocalBusiness, FAQPage) |
| Different NAP information across directories | Strictly identical name, address and phone number everywhere |
| Goal: rank well in a list of links | Goal: be the source cited in an AI-generated answer |
The 5 levers that really work in 2026
1. Proof rather than claims
A generative AI, like a user, places more value on a concrete case (“we supported this Lyon-based client on this topic, with this result”) than on a mere claim of geographic presence. It's verifiable, hard to duplicate from one page to another, and it serves the human user as much as citation by an AI.
In practice: if you have three clients in Lyon, an “our Lyon clients” page with real case studies will always be worth more than a generic “web agency in Lyon” page with no substance.
2. A Google Business Profile that is genuinely maintained
It remains the foundation of local SEO, with or without AI. A Geolid study (2026) shows that a well-kept profile (up-to-date photos, systematic replies to reviews, accurate hours, regular updates) is now a prerequisite rather than a competitive advantage: the national average rating is already close to 4.2/5, and the average number of reviews grows every year.
For businesses without a physical location in each targeted city, Google offers a “service area business” option, which lets you cover several geographic areas without claiming an address you don't have.
3. Consistent structured data
Schema.org markup (Organization, LocalBusiness, FAQPage types) helps engines, classic and generative alike, understand unambiguously who you are, what you do, and where. It isn't a technical gimmick: it's an increasingly decisive prerequisite for being picked up in an AI-generated answer, which needs structured, unambiguous data to “trust” a source.
A page that clearly answers concrete questions (through FAQPage markup, for example) is statistically more likely to be extracted and cited than an equivalent paragraph of free text.
4. Content freshness
An AirOps study (2026, State of AI Search Report) analyzed the pages cited by AI on commercial queries: the vast majority came from content updated in the last twelve months, with a marked peak for content updated in the last six months. A page frozen for two years is statistically less likely to be cited than a recently revised page, even if the substance hasn't fundamentally changed.
In practice, this means local content is no longer handled as “publish once, forget it,” it deserves a review at least annually, ideally every six months on your most strategic pages.
5. Information consistency (NAP)
Name, address, phone (NAP, for Name, Address, Phone), the same information everywhere on the web: your website, your professional directories, your social networks, your Google profile. Inconsistencies (an old address still displayed somewhere, a different number from one directory to another) are a signal of distrust for Google as much as for an AI trying to verify a piece of information before citing it.
Checklist: audit your own local AI presence
Before launching into a redesign, here are a few quick checks you can do yourself:
- Do your “agency in [city]” pages have genuinely different content from one city to the next, or does only the name change?
- Has your Google Business Profile been updated in the last three months?
- Is your NAP information (name, address, phone) strictly identical on your website, your directories and your social networks?
- Does your website include Schema.org markup (Organization, LocalBusiness, FAQPage)?
- Have your most strategic pages been reviewed in the last six to twelve months?
- Do your local pages contain concrete proof (case studies, named customer reviews) rather than generic claims?
- Have you checked in Search Console that your city pages are actually indexed and not treated as doorway pages?
If you answered “no” to more than two questions, there is probably quick room for improvement to seize.
How much does a local SEO + AEO strategy cost in 2026?
Budgets vary widely depending on company size, the level of local competition and the scope of services. As a guide, the ranges commonly seen in the French market in 2026:
- €800 to €2,000 / month for a local SMB needing an audit, monthly content and basic tracking
- €2,500 to €6,000 / month for an in-depth strategy with regular content production and ongoing technical optimization
- €6,000 to €15,000 / month for an aggressive strategy combining local SEO, link building and full GEO/AEO integration
The right indicator isn't the advertised cost, but the expected return: a customer won through good local visibility can represent, over their lifetime, a multiple well above the monthly investment.
Key takeaways
Local SEO isn't dead with the arrival of generative AI, it's changing nature. Quantity gives way to quality and proof. For an SMB or an agency, that's rather good news: the window is still open for those who adapt now, before the practice becomes widespread.


