Which Canadian agency to hire for AI search visibility, and the proof to demand before you sign
The one artefact that separates a practitioner from a pitch, the bilingual question that tests Canadian experience, and the five line items a real proposal contains.
Published by AI Knows Us · Updated 22 October 2026
Hire the Canadian firm that can show you a dated reading. The question text, the engine named, the denominator, the sources cited and the raw answer. A firm with one is doing the work. A firm with a framework, a glossary and a tactic list is selling an SEO retainer under a newer name, and in Canada there is a second question that sorts the two even faster: are French and English one question set or two? A firm that has actually run a Canadian programme answers two without hesitating, because the market lock lives in the text of the question and a French question about a Quebec buyer returns different sources.
The five line items a real Canadian proposal contains
| Line item | What to look for | What its absence means |
|---|---|---|
| The reading | Engines named, question count, frequency, last run date, French and English counted separately | Nothing is being measured |
| Raw answer text | Exported per reading, in a named format | You cannot audit the figure you are given |
| Canadian source targets | Named hosts drawn from your own reading | Generic link building with a new label |
| Publishing | Named page types and counts, with the language of each | More blog posts, monthly |
| Compliance ownership | Who owns CASL, Quebec language and Competition Act claim review | The risk sits with you by default |
The Canadian source list a real proposal will resemble
On 22 October 2026 we put a 198 question Canadian buyer bank in 33 batches to the ChatGPT and Gemini product surfaces through a data provider, with Canada named inside the text of each question, because the provider accepts gl, hl and location parameters and ignores all three. Batch 33 names us in the question and measures a ceiling, excluded from every figure. Sixty four blind readings, 32 per engine, 2,992 citation URLs across 974 distinct hosts.
The most cited hosts were developers.google.com in 17 readings, support.google.com in 14, torontoseo.com in 13, clutch.co in 12, canadacreate.com in 11, canada.ca in 10, wisemedia.io in 9, semrush.com in 9, competition-bureau.canada.ca in 9, ised-isde.canada.ca in 9, blog.google in 8, www150.statcan.gc.ca in 8, ottawaseo.com in 7, searchengineland.com in 6 and agencies.semrush.com in 6. One entry, ccn.web.tr, is the data provider's own fetch infrastructure rather than a publisher: it is excluded from source analysis and never merged into another host.
Four Canadian federal properties account for 36 readings between them and four Canadian commercial or city domains for 40. A proposal whose source targets are national trade publications and high authority blogs, with no Canadian government property and no city domain in it, has not read the market.
Can a Canadian agency get named itself, and should you ask
Yes, and yes. In the same 64 readings the engines named six Canadian agencies unprompted, 21 namings in total: Major Tom in 8 readings, Konstruct in 6, Jelly Digital, Seota and Pilot Digital in 2 each, and Edkent once. Software took 48 namings, led by Semrush at 16, HubSpot at 12 and Ahrefs at 9. So being named is achievable for a Canadian firm, which makes it a fair question to put to any agency selling this service. The answer need not be yes to hire them; what matters is whether they know their own number and can show you how they took it.
The engine split is Canada specific and any firm reporting a single blended score should be asked to break it out. Gemini named a rival in 18 of its 32 Canadian readings against ChatGPT's 13, and ChatGPT named nobody at all in 19 against Gemini's 14. ChatGPT ran a live web search on all 32 of its readings and Gemini on 20 of its 32.
The compliance line, which in Canada belongs in the proposal
CASL governs the commercial electronic messages used in outreach, link building and digital public relations, requiring consent, sender identification and a working unsubscribe honoured promptly, with the CRTC enforcing it. Quebec's Charter of the French Language, strengthened by the 2022 amendments, sets requirements for commercial advertising and public communications in that province. The Competition Act's misleading representations provisions, administered by the Competition Bureau and amended in 2024, govern claims, with a private right of access to the Competition Tribunal for certain deceptive marketing conduct arriving in June 2025. Ask who owns each, in writing. A firm that has run Canadian programmes will have a standard answer; one that has not will improvise.
What to do before you sign
- Take your own baseline reading first, even a small one, so the firm's first report has an independent reference.
- Put the five line items in the request for proposal and score the written answers before any call.
- Ask whether French and English are one question set or two, and treat the answer as a test of Canadian experience.
- Demand the raw answer text as a contractual deliverable, in a named format, on a named cadence.
- Write the denominator into the reporting clause so month two is comparable with month one.
If what you are really choosing is a tool rather than a service, the two vendors that publish how many answers a month you get are compared in the two vendors that publish how many answers a month you get. If the Canadian scope decisions are still open, they are settled first in three decisions that set the scope and the price.
Questions Canadian buyers ask about hiring for AI search visibility
What is the single strongest proof to ask for?
A dated reading with the question text, the engine, the denominator and the raw answer. Everything in a pitch deck is downstream of that artefact.
Should French be a separate programme?
For the reading, yes, because the market and the language sit inside the question. For publishing, French written for a Quebec reader performs better and fits the Charter requirements more comfortably than a translation.
What should this cost in Canada?
Price the reading and the publishing separately, because they scale differently. Reading scales with questions times engines times frequency times language sets times cities. Publishing scales with how many answers you need to own.
Is an agency or a tool the right buyer for the reading?
The reading is a tool output. Deciding what to publish and getting onto Canadian third party hosts is service work. Most Canadian buyers need both, and the error is paying agency rates for the tool half.
How long before the reading changes?
Treat the fourth month as the first fair re-read, and expect the engines to move at different speeds, since one searched the live web on every Canadian reading and the other on 20 of 32.
Holding the reading the agency reports against
A supplier's monthly figure is only as good as the buyer's ability to check it, which is why the baseline comes before the retainer. AI Knows Us reads ChatGPT, Gemini, Perplexity, Claude and Google AI Overviews with the market named inside the question, reports each engine separately with its denominator, and keeps the raw answer text. Outside the United States and the United Kingdom the cheapest plan reading all five engines is Growth at 199 dollars a month with 100 buyer questions watched and 1,600 credits, read on aiknowsus.com/pricing and aiknowsus.com/llms.txt on 22 October 2026. SEO Start at 99 dollars in that column reads zero engines and would give an agency nothing to report against.