If you ask ChatGPT, Claude, and Gemini to recommend a local service business right now, they will likely give you three completely different lists, and there is a high chance you are not on any of them. AI engines are highly confident when recommending software platforms, but when it comes to local services and consultants, their answers are inconsistent, hesitant, and often entirely inaccurate.
I spent the last week running hundreds of queries across the major AI models to see how they handle service recommendations. I wanted to know what it takes for a brand to actually show up when a potential client asks an AI for help. What I found was a massive gap between how these models treat products versus how they treat people and agencies.
The AI Recommendation Confidence Gap
| Query Type | AI Response Pattern | Confidence Level |
|---|---|---|
| “Marketing automation platforms” | All three models (GPT, Claude, Gemini) decisively named HubSpot, Mailchimp, and ActiveCampaign. | High |
| “Digital marketing services in Montreal” | Listed 5-8 agencies each, with almost zero overlap between the different models. | Low |
| “Best SEO consultant in Quebec” | GPT hedged and asked for context. Claude listed generic names. Gemini gave specific, but questionable, recommendations. | Very Low |
The data is clear. If you sell software, the AI knows who you are. If you sell a service, you are probably invisible. But you do not have to guess where you stand. You can find out in about ten minutes.
How do I check my AI visibility?
You check your AI visibility by running five specific prompts as a potential client would, across the three major models, and scoring how often your brand appears in the answers. This is the exact diagnostic process we use to establish a baseline before we start any optimization work.
Most founders assume that because they rank well on Google, they will naturally be recommended by AI. This is a dangerous assumption. Search engines index pages based on keywords and backlinks. AI models synthesize answers based on entity relationships, brand mentions across the web, and semantic authority. They are entirely different systems.
To see what the AI actually thinks of you, you need to step into the shoes of your buyer. Open a fresh session in ChatGPT, Claude, and Gemini. Do not give the AI any context about who you are. Just ask the questions your best clients are asking.
What are the five queries for the self-audit?
The five queries test different stages of the buyer journey, from broad category discovery to direct brand evaluation, revealing exactly where your visibility drops off. You need to run all five to get a complete picture.
Here is the framework. Run these prompts exactly as written, substituting your own details:
- “[Your category] + [your city]” — Are you named? This is the baseline local discovery query. If you are a commercial architect in Toronto, ask exactly that.
- “Best [your service] for [your ideal client type]” — Are you recommended? This tests niche authority. “Best B2B accounting firm for SaaS startups.”
- “Compare [you] vs [competitor]” — Does AI know enough to compare? This is crucial. If the AI hallucinates details about your business while accurately describing your competitor, you lose the deal right there.
- “[Your brand name]” — What does AI say about you specifically? Look for accuracy in your core offerings, your location, and your value proposition.
- “Who should I hire for [problem you solve]?” — Are you in the consideration set? This is the ultimate intent query. “Who should I hire to fix my bilingual SEO architecture?”
How do I score my AI visibility?
You score your visibility by counting the number of times your brand is explicitly mentioned and accurately described across the five queries, giving you a score from zero to five. This gives you a hard metric to track over time.
Do not give yourself partial credit. If the AI mentions you but gets your core service wrong, that is a zero for that query. If it lists you as being in Vancouver when you are in Montreal, that is a zero. Accuracy matters just as much as presence.
Here is the scoring rubric:
* 0 mentions = Invisible. The AI does not know you exist, or it hallucinates your details so badly that it actively hurts your brand. You are losing word-of-mouth referrals that are being vetted through AI.
* 1-2 mentions = Emerging. You have some semantic footprint, usually tied to your direct brand name, but you are not seen as an authority in your category.
* 3-4 mentions = Visible. You are doing well. The AI understands what you do and recommends you for specific problems. You are likely winning business from AI-assisted research.
* 5 mentions = Dominant. You own your category in the AI’s training data. When people ask about your space, you are the default answer.
What if my score is zero?
If your score is zero, you need to stop relying solely on traditional SEO and start building semantic authority by getting your brand mentioned in high-trust, unstructured data sources that AI models actually read. A zero score means your digital footprint is too thin or too confusing for the AI to parse.
This is the reality check most businesses fail. They look at their zero score and immediately think they need to stuff more keywords into their website. That will not work. AI models are not crawling your site looking for keyword density. They are looking for consensus across the web. They want to see your brand mentioned in podcasts, in industry reports, on authoritative forums, and in digital PR.
You have to feed the machine the right kind of data. You need clear, consistent messaging across all your digital assets. Your schema markup needs to be flawless. Your entity relationships need to be explicitly defined. If you are a digital marketing agency in Montreal, every piece of content you produce needs to reinforce that specific entity relationship.
The Three Moves to Build AI Visibility
If you want to move from invisible to visible, you need a systematic approach. Here is what you can do right now:
- Fix your foundational schema. Ensure your website has robust Organization and LocalBusiness JSON-LD markup. This is the most direct way to hand-feed facts about your business to AI crawlers.
- Publish opinionated, original data. AI models love citing original research. Stop writing generic “how-to” posts and start publishing data that only you have. (Like the query data table at the top of this article).
- Clean up your digital footprint. Ensure your brand name, address, phone number, and core service descriptions are identical across every directory, social profile, and press release. Inconsistency breeds AI hesitation.
Where does this self-audit fall short?
This ten-minute audit is a blunt instrument; it only measures your baseline presence and accuracy, completely ignoring the deeper technical and structural factors that dictate long-term AI visibility. It tells you if you have a problem, but it does not give you the blueprint to fix it.
We use this exact five-query framework as the starting point for our clients, but it only covers two of the eight dimensions we actually measure. It does not look at your content architecture, your entity associations, your sentiment analysis across different models, or how your brand performs in multi-turn conversational prompts. It is a diagnostic tool, not a cure. If you score a zero, you know you are bleeding, but you still need an MRI to find out why.
Ready for the full picture?
That ten-minute check covered exactly two of the eight dimensions of AI visibility. If you want to know how you perform across the other six—including entity association, sentiment bias, and conversational retention—you need the full diagnostic.
Get the complete AI Visibility Audit and see exactly what the machines are saying about you when you aren’t in the room.
Related in the field: start with what 5 AIs answered about Montréal’s best agency, then see a bilingual entity footprint in practice in the Émondage Élagage Expert Signal File.
WHERE THIS GOES
Ask an AI about your brand. We’ll wait. The audit is free, human-reviewed, and lands within 48 hours.
