If you ask ChatGPT, Claude, and Gemini to name the best marketing agency in Montréal, you won’t get the same list Google gives you. While massive legacy agencies still dominate both, mid-size and boutique firms that rank on page one of Google are almost entirely invisible to AI engines. The reality is simple: optimizing for search engines no longer guarantees you exist in the answers generated by artificial intelligence.
I wanted to know what happens when a potential client bypasses Google entirely and asks an AI for a recommendation. We know the shift is happening. People are treating language models like concierges, asking them for highly specific, localized business recommendations. So, we ran a test. We asked the major AI models a straightforward question: “Who is the best marketing agency in Montréal?”
The results were a wake-up call. For years, the digital marketing industry has operated on a single, unshakeable premise: if you want to be found, you optimize for Google. You build backlinks, you structure your site architecture, you target long-tail keywords, and you fight for that coveted top-three position on the search engine results page. But the landscape is fracturing. The way people search for information, and more importantly, the way they search for service providers, is fundamentally changing.
When a VP of Marketing needs an agency, they aren’t just typing “best marketing agency montreal” into Google and clicking the first link. They are opening ChatGPT or Claude, describing their specific business challenges, and asking the AI to curate a shortlist of agencies that fit their exact criteria. They are treating the AI as an expert consultant.
This shift presents a massive problem for agencies that have built their entire lead generation strategy around traditional SEO. If the AI doesn’t know you exist, you aren’t making the shortlist. You are losing pitches before you even know they are happening.
Here is exactly what the models told us, compared to Google’s traditional top results.
| Search Engine / AI Model | Top Recommendations for “Best Marketing Agency Montréal” |
|---|---|
| Google (Top 10) | Sid Lee, Cossette, lg2, Bleublancrouge, Bloom, Major Tom, Dialekta |
| GPT-5 | Sid Lee, Cossette, lg2, Bleublancrouge, Dialekta |
| Claude | Fg+a (Fortin + Audette), Bleublancrouge, Lg2, Diesel, Nurun, Adviso |
| Gemini | Sid Lee, Bloom, Adviso, My Little Big Web, Digitad |
Data generated from direct queries to GPT-5, Claude, and Gemini, compared against Google Search results for the Montréal region.
The overlap is obvious at the top. If you are Sid Lee, Cossette, or lg2, you are fine. You have Wikipedia pages, decades of press, massive digital footprints, and a legacy that is deeply baked into the training data of every major large language model. But look closer at the discrepancies. Major Tom ranks well on Google but vanishes in the AI responses. Fg+a and Diesel appear in Claude’s recommendations but are absent from the others.
And Crelong Media? We didn’t show up in a single AI response. Not one.
This omission stung, but it also sparked a realization. We are a digital marketing agency specializing in AI visibility and bilingual SEO. If we aren’t showing up in these queries, what hope does a mid-size B2B SaaS company or a local law firm have? The rules of the game have changed, and the old playbook is no longer sufficient. We needed to understand exactly how these models were making their recommendations, what signals they were prioritizing over traditional search metrics, and how we could engineer visibility in an AI-first world.
Why do AI engines recommend different businesses than Google?
AI engines do not crawl the web in real-time the way traditional search engines do to rank links; they synthesize answers based on the training data they have ingested and the specific weights they assign to trust signals. When an AI recommends a business, it is looking for a dense, consistent footprint of entity mentions across high-authority sources, not just a well-optimized landing page.
Google relies heavily on backlinks, keyword density, and technical SEO to determine who gets the top spot. If you build a fast website, satisfy user intent, and get enough authoritative sites to link to you, you can rank. Google’s algorithm is fundamentally a matching engine for documents. It wants to serve the most relevant web page for a given query.
AI models, however, operate on a different paradigm. They are not retrieving documents; they are generating text based on statistical probabilities derived from their training data. They favor entities that are deeply embedded in the knowledge graph. They look for structured data, mentions in reputable publications, Wikipedia presence, and a strong, consistent narrative across the web.
If your business only exists on your own website and a few directory listings, you might as well be a ghost to an AI. The models need to see your brand discussed in context, repeatedly, by sources they already trust. This is why the legacy agencies dominate the AI responses. They have decades of PR, news mentions, and industry awards baked into the training data. Their entity footprint is massive and undeniable.
For mid-size agencies, this is a daunting reality. You cannot simply out-SEO a legacy agency in an AI model. You have to out-publish them, out-PR them, and ensure that your brand is inextricably linked to the specific topics and services you want to be known for. You have to transition from optimizing pages to optimizing entities.
Does ChatGPT recommend businesses based on SEO?
No, ChatGPT does not recommend businesses based on traditional SEO metrics like keyword rankings or backlink profiles. While good SEO practices can indirectly influence AI visibility by increasing your brand’s presence on the web, the direct correlation is surprisingly weak.
We dug deeper into the data to test this. We knew that general queries favored the giants, so we narrowed our focus. When we modified the query to ask for B2B-specific agencies, the answers shifted entirely. The models recommended Adviso, Bloom, Parkour3, Conversion, and Groupe Ctrl+A. When we asked for AI-focused agencies, they gave us Dialekta, Sid Lee, Exponential.ai, Tactile, Lg2, Nurun, Adviso, and Bloom.
This tells us something crucial: AI models are highly sensitive to context and modifiers. They aren’t just pulling a static list of “the best.” They are dynamically generating recommendations based on the specific attributes associated with your brand in their training data. If your content footprint doesn’t explicitly and repeatedly tie your brand to “B2B” or “AI-focused,” you won’t surface for those queries, no matter how well you rank on Google for those terms.
Traditional SEO often focuses on capturing search volume for broad terms. AI visibility requires a hyper-specific association with niche concepts. If you want ChatGPT to recommend you as a B2B marketing agency, your brand name needs to appear in close proximity to B2B marketing concepts across the web. It’s not enough to have a “B2B Marketing” service page on your site. You need industry publications, podcasts, and third-party reviews discussing your B2B expertise. The AI needs to learn the association through repeated exposure in its training data.
How do AI models filter for B2B and AI-focused agencies?
AI models filter for specialized agencies by analyzing the semantic proximity of your brand name to specific industry terms across the entire web, meaning your website’s service pages matter far less than how third-party sources describe your expertise. If the broader internet doesn’t explicitly categorize you as a B2B or AI-focused agency, the AI models won’t either.
When we ran our initial tests, the results were dominated by the massive, full-service consumer agencies. But we wanted to see what would happen if a potential client had a more specific need. We modified our query to ask for the best B2B marketing agencies in Montréal. The results shifted dramatically. The models recommended Adviso, Bloom, Parkour3, Conversion, and Groupe Ctrl+A.
Why did these specific agencies surface? It wasn’t because they had a single page optimized for “B2B marketing agency Montréal.” It was because their entire digital footprint is saturated with B2B context. They publish case studies about complex B2B sales cycles. Their executives speak at B2B conferences. They are mentioned in articles discussing B2B lead generation strategies. The AI models have learned, through millions of data points, that these entities are strongly associated with the concept of B2B marketing.
We saw a similar pattern when we asked for AI-focused agencies. The models gave us Dialekta, Sid Lee, Exponential.ai, Tactile, Lg2, Nurun, Adviso, and Bloom. Some of these are legacy agencies that have aggressively pivoted their PR to highlight their AI capabilities. Others, like Exponential.ai, are built entirely around the concept. The AI models picked up on these signals. They recognized the press releases, the thought leadership articles, and the structured data that tied these brands to artificial intelligence.
This highlights a critical vulnerability for generalist agencies. If you try to be everything to everyone, your entity footprint becomes diluted. The AI models struggle to categorize you. In an AI-driven search environment, specificity wins. You need to clearly define your niche and ensure that every piece of content you publish, and every third-party mention you earn, reinforces that specific association. If you want to be known for AI visibility, you have to own that conversation across the web, not just on your own domain.
How do AI models handle localized queries like “Montréal”?
AI models handle localized queries by looking for strong geographical associations within your entity footprint, relying heavily on structured data, local news mentions, and context clues rather than traditional local SEO signals like Google Business Profile proximity. If your brand isn’t consistently tied to Montréal in third-party content, the AI won’t confidently recommend you for a local query.
In our tests, the models clearly understood the geographical constraint. They didn’t recommend agencies from Toronto or New York. But how did they know which agencies were truly the “best” in Montréal? They looked for signals of local authority. This includes mentions in Montréal-based publications, participation in local industry events, and bilingual content that reflects the reality of the Québec market.
For agencies operating in Montréal, bilingualism is a massive entity signal. If your digital footprint is entirely in English, you are missing half the context that AI models use to understand the local market. The agencies that surfaced consistently in our tests—like lg2, Cossette, and Bleublancrouge—have deep, bilingual footprints that firmly establish their presence in Québec. They aren’t just located in Montréal; they are part of the cultural and business fabric of the city, and that reality is reflected in the data the AI models have ingested.
Can you manipulate AI recommendations?
You cannot manipulate AI recommendations using traditional black-hat SEO tactics like keyword stuffing or buying cheap backlinks, but you can strategically influence them by feeding the knowledge graph with high-quality, verifiable information. The models are designed to resist simple manipulation, prioritizing consensus and authority over sheer volume of mentions.
If you try to spam the internet with low-quality press releases claiming you are the “best marketing agency in Montréal,” the AI models will likely ignore them. They weigh the authority of the source heavily. A single mention in a reputable business journal carries exponentially more weight than a hundred mentions on low-tier directory sites.
To influence the models, you have to play their game. You have to provide structured, verifiable data. This means ensuring your Google Business Profile is perfectly aligned with your website’s schema markup, your Wikipedia page (if you have one) is accurate and well-cited, and your profiles on major review platforms like Clutch or G2 are robust and active. The AI looks for consistency across all these touchpoints. If your website says you specialize in AI visibility, but your Clutch profile says you are a generalist web design firm, the AI gets confused and will likely choose a competitor with a more consistent narrative.
What happens if you ignore AI visibility?
If you ignore AI visibility, you risk being entirely excluded from the next generation of search and discovery. As more users turn to AI models for recommendations, the businesses that fail to adapt will simply cease to exist in these new ecosystems, watching their pipeline dry up as competitors capture the AI-driven consideration phase.
We are already seeing this shift in user behavior. People are tired of sifting through ten blue links, dodging sponsored ads, and reading SEO-optimized fluff that takes 500 words to answer a simple question. They want direct, synthesized answers. They want an AI to do the heavy lifting of research and curation.
If a potential client asks Claude for a list of the best marketing agencies in Montréal and your name isn’t on it, you aren’t even in the consideration set. You have lost the pitch before you even knew it was happening. You won’t see a drop in your Google Analytics traffic immediately, but you will notice a slow, inexplicable decline in high-quality inbound leads.
This isn’t a future problem; it is a current reality. The data from our test proves it. The mid-size agencies that are currently thriving on Google traffic are highly vulnerable to this shift. If they don’t start optimizing for AI visibility now, they will find themselves entirely dependent on a legacy search engine while their most lucrative prospects migrate to AI-driven platforms.
The Three Moves for AI Visibility
If you want to ensure your business is recommended by AI engines, you need to stop relying solely on traditional SEO and start executing on these three concrete steps today:
- Deploy Comprehensive Entity Schema: Implement detailed JSON-LD Organization and Person schema on your website. Define your brand, your founders, your services, and your social profiles explicitly. Connect the dots for the AI. Do not leave it up to the models to figure out who you are and what you do.
- Build a Third-Party Content Footprint: Stop focusing solely on your own blog. Get your brand mentioned on high-authority, trusted websites. Pitch guest posts, appear on podcasts, and secure digital PR placements. The AI needs to see you validated by others. A mention on a high-authority industry site is worth ten blog posts on your own domain.
- Publish Original, Quotable Data: AI models love original research and statistics. Publish data that no one else has. When other sites cite your data, they reinforce your authority and relevance in the AI’s training data. Become the source of truth for a specific topic in your industry.
Where we were wrong (and what this doesn’t solve)
When we first started looking into AI visibility, we assumed that a strong technical SEO foundation would naturally translate into AI recommendations. We were wrong. We spent months optimizing our site speed, fixing crawl errors, and building traditional backlinks, only to find that it moved the needle on Google but did absolutely nothing for our visibility in ChatGPT or Claude.
This approach also doesn’t solve the problem of immediate lead generation. AI visibility is a long game. You cannot buy your way to the top of an AI recommendation list the way you can with Google Ads. It requires sustained effort, a fundamental shift in how you approach digital presence, and the patience to wait for the models to ingest and reflect your updated entity footprint. It is not a quick fix for a slow quarter.
Find out what the AIs are saying about you
You can’t fix what you can’t measure. If you don’t know whether ChatGPT, Claude, and Gemini are recommending your business, you are flying blind. We built a process to map exactly how your brand appears across the major AI models and identify the gaps in your entity footprint.
Get your free AI visibility audit today and see exactly what the machines are telling your potential clients.
Related in the field: watch the entity-footprint playbook in action as a local trade becomes the only bilingual answer around L’Assomption, then run the 10-minute AI visibility self-audit on your own brand.
WHERE THIS GOES
Ask an AI about your brand. We’ll wait. The audit is free, human-reviewed, and lands within 48 hours.
