Why ChatGPT Isn't Recommending Your Montreal Dental Practice

A growing number of people looking for a dentist in Montreal aren’t starting on Google anymore — they’re asking ChatGPT, in French or English. An estimated 45% of people now ask AI tools for local recommendations, a sharp rise from a year earlier. In a bilingual market, a website with thin or generically-translated content gives an AI tool even less to work with — in either language — than a single-language site with the same problem.

Here’s the plain version: imagine someone asks ChatGPT, “meilleur dentiste à Montréal” or “best dentist in Montreal.” Either way, the AI answers using whatever it can find that’s clear and specific. A page that just repeats generic phrases — translated or not — has nothing for it to quote. So it recommends the practice that actually spelled things out, in whichever language the searcher used.

Why this matters especially in Montreal

Montreal is a large bilingual market where clear, fast-loading pages stand out, which makes this specific problem worse for a dental practice:

• Content quality has to hold up in two languages, not one. A vague or thinly translated version in either language gives an AI tool less to work with than a fully fleshed-out single-language site would.
• Patients search in whichever language they’re most comfortable with. A practice needs to be recommendable to both — missing genuine detail in one language means losing that half of the market to AI recommendations.
• A competitive dental market on both sides of the language divide. There’s no shortage of other practices for an AI tool to recommend instead if yours doesn’t give it something specific to say.

Average Maps position — practices AI skips (worse = higher number) 13.2

Average Maps position — practices AI actually names (better) 8.5

How Google Maps ranking differs for dental practices AI actually mentions

Item	Value
Average Maps position — practices AI skips (worse = higher number)	13.2
Average Maps position — practices AI actually names (better)	8.5

Source: Identity Dental, 2026 AI Visibility Study

What actually makes ChatGPT recommend a dental practice, in either language

• Say exactly what you offer, in plain language — in both languages. Not “des soins avancés,” but “nous offrons des plombages, des traitements de canal, et des couronnes dentaires,” with a real page for each, matched by an equally specific English version.
• Answer the questions patients actually ask, in the language they ask them. “Est-ce que ça fait mal de se faire poser un plombage” and “does it hurt to get a filling” both deserve direct, honest answers — not one being an afterthought translation of the other.
• Add the hidden “label” that tells computers what your page is about in the correct language — this is schema markup, essentially a clearly labeled index card behind your webpage.
• Build the site so it loads fast and shows content immediately, in whichever language the page is in — some AI tools won’t wait around for extra code to finish loading before reading the page. This is the same reason a lightweight, Astro-built site tends to out-rank a heavier build.

Websites AI tools recommend vs. websites they skip

What you actually get

• You show up when someone asks an AI tool for a dentist in Montreal, in either language — instead of losing that patient to a competitor before they ever see your name.
• Both halves of your market get equally strong content, rather than one language getting the real effort and the other an afterthought.
• Your site keeps working for regular Google search too, in both languages — every AEO fix here is also a plain SEO improvement.
• The site loads fast on a phone, which matters directly to patients searching in a hurry, in whichever language they think in — the same recall and intake automation covered in our New York dental snapshot guide makes sure that speed carries through to booking too.

What AEO and GEO actually mean, in plain terms

AEO (Answer Engine Optimization) and GEO (Generative Engine Optimization) are two closely related terms for the same shift: making your practice the source an AI tool pulls from — and names — when it answers a question directly, instead of handing someone a list of links to click through. AEO usually refers to the narrower goal of getting pulled into a direct, synthesized answer or a featured snippet; GEO usually refers to the broader goal of being named and described favorably inside a generative AI response from a tool like ChatGPT, Perplexity, or Google’s AI Overviews. In practice, most agencies — including us — use the two terms almost interchangeably, because the work that earns one almost always earns the other.

It helps to be precise about what this is not. AEO/GEO is not a separate discipline that replaces search engine optimization, and it’s not a special trick that works independently of having a genuinely good website. The agencies pitching it as a brand-new category you need to buy separately from “regular SEO” are, for the most part, selling the same fundamentals — fast pages, clear structure, real content, legitimate reviews — with a new label on top. What’s genuinely new is the mechanism an AI tool uses to decide what counts as trustworthy and quotable, and that mechanism rewards a few specific things more heavily than classic keyword-based ranking did: direct, self-contained answers near the top of a page; structured data that spells out facts a model doesn’t have to guess at; and consistency between what your website says and what shows up about your practice everywhere else online — your Google Business Profile, directory listings, and reviews.

For a Montreal dental practice specifically, that last point is doubly true because the “everywhere else online” often exists in two languages. An AI tool checking whether your hours, services, and address are consistent has to reconcile a French listing and an English listing that may have been set up months apart by different people. When they don’t match, that’s a weak signal to the model — not necessarily a feature.

How ChatGPT, Perplexity, and Google AI Overviews actually decide what to cite

None of the major AI answer engines consult a single ranked list the way classic Google search does. Instead, each works through a version of the same process: find content related to the question, decide which of it is trustworthy and relevant enough to use, and then extract or synthesize a specific answer from the sources that pass that bar.

A few concrete things make a dental practice more likely to clear that bar:

• Specific, self-contained answers. A paragraph that fully answers one question — “what does a root canal cost at our practice” or “qu’est-ce qu’un traitement de canal” — without requiring the reader to already know context from elsewhere on the page, is exactly the shape of text these systems are built to lift and quote.
• Structured data (schema markup). This is metadata embedded in the page’s code — invisible to a visitor, but directly readable by software — that spells out facts explicitly instead of making a model infer them from prose. Dentist/LocalBusiness schema states your name, address, phone number, and hours in a fixed format; Service schema lists out what you offer; FAQPage schema marks a block of questions and answers as exactly that, which is one of the highest-leverage schema types for AI visibility because it maps almost one-to-one onto the question-and-answer format these tools output.
• Author and practice credibility signals (E-E-A-T). Pages that show who wrote or reviewed the content, link to a dentist’s real credentials, and match the practice’s registration with the Ordre des dentistes du Québec give a model more reason to treat the content as trustworthy enough to repeat to someone asking a health-related question.
• Consistency across the web, not just your own site. Several of these tools also weigh directory listings, review platforms, and mentions elsewhere — not only your website — so a page that’s internally perfect but contradicted by an outdated Google Business Profile still undermines the signal.

It’s worth being skeptical of anything that frames this as a hack. There’s no special file format or secret markup that gets a page cited on its own; a page still needs genuinely useful, accurate, specific content underneath the schema, or the schema just formally confirms that there’s nothing worth quoting.

A step-by-step look at what makes a dental website AEO-ready

Turning the principles above into an actual build comes down to a handful of concrete, checkable pieces. This is roughly the order we work through them on a new dental site:

Building an AEO-ready dental website, step by step

Step	What it covers	Why it matters to an AI tool
1. Service pages, one per treatment	A dedicated page for each major service (cleanings, fillings, root canals, implants, Invisalign, etc.), not one page listing everything	Gives a model one clear, specific source to cite per question instead of a vague overview page
2. NAP data, locked and consistent	Name, Address, Phone — identical on the website, Google Business Profile, and directory listings, in both languages	Mismatched basic facts are one of the fastest ways to lose trust signals with both AI tools and Google
3. Dentist/LocalBusiness + Service schema	Structured data describing the practice and each service in machine-readable form	Removes the guesswork — the facts are stated, not inferred from prose
4. FAQPage schema, one block per page	A genuine FAQ section on each relevant page, marked up so software recognizes it as question-and-answer pairs	Maps directly onto the format AI answer engines already output
5. Author / practice credibility details	Named dentist(s), credentials, and a link or reference to Ordre des dentistes du Québec registration	Health-related answers get extra scrutiny from AI systems before being repeated to a user
6. Fast, immediately-visible content	A lightweight build (see our note on [Astro vs. WordPress performance](/blog/astro-vs-wordpress-performance)) that renders real content without waiting on heavy scripts	Some AI crawlers and human visitors alike will not wait around for a slow page to finish loading
7. Reviews, displayed and kept current	Recent, real patient reviews surfaced on the site and kept current on Google	Reviews are one of the few third-party trust signals AI tools weigh alongside the website itself

None of these steps is exotic on its own. The practices that actually show up in AI answers are usually the ones that did all seven, consistently, rather than the ones that did one of them perfectly.

A buyer’s guide: what to ask an agency before paying for “AEO”

AEO and GEO have become popular enough as terms that a lot of agencies now offer to sell it without materially changing what they do. A few direct questions separate the agencies that understand the mechanics from the ones re-labeling a standard SEO package:

• “Show me schema markup you’ve actually deployed for a real client.” If they can’t point to a live example, they likely haven’t implemented it themselves.
• “How would you test whether our practice currently gets mentioned by ChatGPT, Perplexity, or Google AI Overviews for our key searches?” This is a test that takes minutes to run live — a capable agency should be comfortable doing it in front of you.
• “What’s actually different about your process for AI search versus plain SEO?” A vague answer, or one that treats the two as identical with no distinct steps, suggests the term is being used as a sales label rather than a real methodology.
• “How do you measure this, given that most AI citations never produce a click?” Honest agencies will acknowledge that AEO/GEO measurement is less mature than click-based SEO reporting, and will talk about citation tracking or direct testing rather than promising clean analytics dashboards that don’t really exist yet for this channel.

A few answers are reliable red flags on their own: claiming AI search “isn’t really measurable yet” as a reason to skip reporting entirely, pitching a completely separate, expensive “AI package” layered on top of your existing site with no clear deliverable, or promising a specific ranking position inside an AI answer — these tools are non-deterministic enough that no one can honestly guarantee a specific outcome.

Common AEO mistakes dental practices make

Most of the practices that struggle here aren’t doing nothing — they’re doing a few things that look productive but don’t move the needle, or actively work against them:

• Treating AEO as a bolt-on instead of a foundation. Buying “AI optimization” as an add-on service while the underlying site is slow, thin, or poorly structured skips the part that actually matters; AEO/GEO builds on the same fundamentals as good SEO, it doesn’t replace them.
• Over-indexing on schema while under-investing in content. Perfect structured data wrapped around a vague paragraph just tells a model, accurately, that there’s nothing specific to quote.
• Publishing generic content instead of specific answers. A blog post that could apply to any dental practice in any city, with the name swapped, gives an AI tool nothing a competitor’s equally generic page doesn’t also offer.
• Only focusing on the website itself. AI tools often weigh directory listings, review platforms, and third-party mentions alongside a practice’s own site — a perfect website with a neglected or inconsistent Google Business Profile still sends a mixed signal.
• Letting French and English drift out of sync. In a bilingual market, the most common version of this mistake is a fully-built English site next to a thin, auto-translated French equivalent (or the reverse) — which quietly caps how well the practice can be recommended to half its potential patients.
• Ignoring mobile. A majority of local searches, including the conversational ones typed into an AI chat app, happen on a phone — a page that’s slow or awkward on mobile undermines everything else that was done correctly.

What this actually costs, and the realistic payoff

There’s no industry-standard price tag for “AEO” as a separate line item, and that’s partly the point — the work overlaps so heavily with a well-built website and ongoing content that pricing it as a brand-new category is usually more about the agency’s positioning than a genuinely distinct deliverable. A more useful way to think about the cost is in two parts: the one-time cost of building (or rebuilding) a site with the right structure, schema, and bilingual content, and the ongoing cost of keeping listings, reviews, and service pages current as the practice changes.

The realistic payoff isn’t a guaranteed flood of AI-driven bookings — no one can honestly promise that, given how new and non-deterministic this channel still is. The more defensible case is that the same work compounds across three channels at once: a practice that fixes thin content, adds real schema, and keeps NAP data consistent tends to see it reflected in regular Google rankings, in richer Google Maps visibility, and in AI-tool mentions, because all three systems are rewarding overlapping signals. Viewed that way, the spend is closer to “build (or fix) the website properly, once, in both languages” than to a recurring specialty service — which is also why a practice that’s never invested in its site architecture usually sees the largest relative gain.

Montreal-specific context that actually matters here

A few facts about the Montreal dental market are durable enough to shape how a practice should actually handle its website, rather than being generic filler:

• Most adults in Quebec have no RAMQ dental coverage. RAMQ’s dental benefits are largely limited to children under 10 and people on certain financial assistance programs with a valid claim slip — the large majority of working adults and seniors pay out of pocket, carry private insurance, or now qualify for the federal Canadian Dental Care Plan (CDCP) if their household income is under the program’s threshold and they have no private coverage. That makes cost transparency and clear information about accepted insurance and the CDCP genuinely more valuable to a Montreal patient researching a dentist than it would be in a jurisdiction with broader public dental coverage, and it’s exactly the kind of specific, practical detail an AI tool can quote when someone asks what a procedure actually costs.
• Every licensed dentist in Quebec is regulated by the Ordre des dentistes du Québec, the professional body that governs licensing and handles patient complaints about cost or care quality in the province. Referencing a dentist’s registration with the Ordre, alongside their name and credentials, is a legitimate, verifiable trust signal — the kind of author/practice credibility detail that both patients and AI systems weigh more heavily on health-related content.
• Montreal is a genuinely bilingual market, not a market with one dominant language and an afterthought translation. A large share of residents are equally comfortable searching in French or English, and many move between the two depending on context — which is why a thin, machine-translated version of a site in either language shows up as thin content, full stop, rather than as an acceptable shortcut.
• The dental market is dense and competitive, particularly on the island of Montreal, with a high concentration of practices clustered in central and west-end neighborhoods compared to the surrounding off-island suburbs. In a market with this many alternatives for a patient — or an AI tool — to choose from, a vague or generic web presence has more competitors ready to be recommended instead, in whichever language the search happened in.

What an AEO-ready service page actually looks like

The step-by-step checklist above covers the structural pieces, but the part that actually determines whether an answer engine quotes a page is the writing on it. The difference is concrete enough to show side by side.

A typical thin service page says something like: “Nous offrons des soins dentaires complets et personnalisés pour toute la famille” — or the English equivalent, “We offer comprehensive, personalized dental care for the whole family.” It’s not wrong, exactly, but there is nothing in that sentence an AI tool can lift and hand to someone asking “which Montreal dentist does same-day crowns” or “qui offre des couronnes le même jour à Montréal.” It’s true of every dental practice in the city, which means it distinguishes none of them.

An AEO-ready version of the same page states, plainly, what the service actually is, roughly what it costs, how long it takes, and who it’s for: “We offer same-day CEREC crowns, typically completed in a single 90-minute visit rather than the usual two-appointment process, with pricing discussed upfront before any work begins.” That sentence is self-contained — a model doesn’t need outside context to understand or repeat it — and it directly answers a specific question a patient (or an AI tool on their behalf) might actually ask.

The same logic applies to the FAQ section on each page, not just the opening paragraph. A weak FAQ answer restates the question (“Oui, nous offrons ce service” / “Yes, we offer this service”). A strong one gives the actual information someone needs to decide whether to book: what’s involved, what it typically costs or what insurance covers, and what to expect at the appointment — written out fully enough that it would make sense to someone who had never seen the rest of the page.

How to check whether your practice is already showing up in AI answers

Before paying anyone to fix this, it’s worth spending ten minutes finding out where your practice currently stands. The test is simple and doesn’t require special tools:

• Ask directly, in both languages. Open ChatGPT, Perplexity, or Google and type something close to what a real patient would ask — “best dentist near Plateau-Mont-Royal,” “meilleur dentiste proche du Plateau-Mont-Royal,” “dentist that does Saturday appointments in Montreal.” Note whether your practice comes up at all, and if it does, whether the details the AI tool states about you (hours, services, address) are actually correct.
• Try a few different phrasings. The exact wording of a question can change which sources an AI tool pulls from, so testing three or four variations gives a more honest picture than a single query.
• Check what it says about competitors too. If a nearby practice consistently gets named with specific, accurate detail and yours doesn’t (or gets described vaguely, or with outdated information), that gap is the actual problem to fix — not an abstract concept.
• Re-check periodically, not once. These systems update what they retrieve from on an ongoing basis; a practice that fixes its site should expect to see the change reflected over weeks, not overnight, and should check back rather than assuming a single early test is the final verdict.

This kind of direct testing is also the fair way to hold an agency accountable: ask them to run this same check live, before and after their work, rather than taking a vague promise of “AI visibility” at face value.

Keeping both languages equally strong over time

Building a bilingual, AEO-ready site once is the first step; keeping both versions equally current is the part that actually protects the investment. A few habits make the difference between a site that stays useful and one that quietly decays back into the “thin translation” problem within a year or two:

• Update both language versions together, every time. When a new service, price, or hour changes, treat the French and English pages as one update, not two separate tasks on two separate timelines — a stale version in either language is exactly the inconsistency that weakens AI trust signals.
• Write new FAQ entries from real patient questions, in whichever language they were actually asked. The questions people type into a chat window are rarely identical phrasings of the same idea in French and English; a direct translation of an English FAQ often misses the way a French-speaking patient would actually ask it, and vice versa.
• Review schema markup after every content change, not just when the site is first built — a Service schema block that still lists an old price or a FAQPage block that no longer matches the visible text on the page undercuts the very consistency that makes structured data useful in the first place.
• Revisit your Google Business Profile and directory listings on the same schedule as the website, in both languages, since AI tools weigh those sources alongside — not instead of — the site itself.

None of this requires a large team; it requires treating the bilingual site as one connected asset rather than two separate projects that happen to share a domain.

Bringing it together

Being recommended by ChatGPT isn’t a separate marketing channel to chase — it’s mostly a byproduct of having clear, specific, fast content, in whichever language a patient is searching in. A Montreal practice that gives equal, genuine attention to both languages — instead of treating one as a thin translation of the other — positions itself to show up in AI answers on both sides of the market. Our Houston dentist AEO guide and our full AEO/GEO explainer cover the same fundamentals in more depth.