AI SEO agency
Search work scoped for a results page that answers the question itself. Published prices, sampled measurement, and a written limit on what we claim.

Most agencies selling AI SEO are selling four workstreams under a newer name. That is not the criticism it sounds like. The workstreams are real, and three of the four existed long before ChatGPT did.
What is genuinely new is the scoreboard. Search Console will tell you your impressions rose and your clicks did not. It will not tell you that an AI Overview quoted your pricing page and answered the question on the spot. Nothing will, in the sense of a feed you can query. Any agency handing you an exact monthly citation count has estimated it and rounded the estimate in its own favour.
We run this as one programme with two scoreboards and publish what it costs: $299 CAD a month for reporting with no execution, $1,999 CAD a month for active work across search, AI, paid media and social, and $2,000 CAD once for the full audit if you would rather buy the diagnosis before the treatment.
At a glance
Diagnosis before retainer
Most engagements start with the $2,000 audit, and it is never bundled with the implementation. An agency that has already sold you the remediation has a reason to find remediation-shaped problems.
Prices on the website
Both retainers and the audit fee are published, with what each one excludes written next to it. You should not have to sit through a discovery call to learn a number.
A panel you can audit
Your buyers' actual prompts, run on a schedule across ChatGPT, Gemini, Perplexity, Claude, Copilot and AI Overviews, with every named brand recorded. It is a sample. We label it as one.
Nothing traded for it
AI-facing changes get made without touching what already ranks. No H1 or URL rewrites on a winning page without a redirect plan and a 2-6 week decay window budgeted in advance.
What you are buying, itemised
Technical reachability: confirming that each retriever gets the same document a browser does. This is a harder bar than passing a Googlebot check, because several AI crawlers do not execute JavaScript, and content that only appears after hydration is content they never see.
Passage structure: retrieval operates on chunks, not pages. A section that answers one question completely, with the brand named inside it, is liftable. The same information spread across four subheadings and a table is not.
Entity resolution: one unambiguous machine-readable identity, so an engine knows which company the answer belongs to. This is where most multi-location and rebranded businesses quietly lose.
Corroboration: third parties saying the same things about you that you say about yourself. It is the slowest of the four and the one no amount of on-site work substitutes for.
Why the measurement is a sample and stays labelled as one
There is no analytics feed for AI answers. Generative results are non-deterministic, personalised, and rewritten between model versions, so the same prompt on a Tuesday and a Thursday can name different companies for reasons nobody outside the lab can reconstruct.
So we sample. A maintained panel of prompts, a fixed schedule, six engines, and a record of which brands each answer names. Run it long enough and the trend is real even though any single reading is noisy. Reported next to Search Console, never merged into it.
Rising impressions against flat clicks is the pattern that gets misread most often. It usually means you are being read and not visited. That is a GEO outcome, not an SEO failure, and confusing the two leads teams to fix the wrong thing for a quarter.
What our published numbers prove, and what they do not
Osoba Renos, a GTA renovation contractor: search clicks up 56.1% comparing the most recent three months against the prior three, brand queries at position 1, four non-branded service and near-me queries on page one. Valve Appliance Repair: impressions up 631% and clicks up 331% across its first seven weeks, at an average position still in the high twenties. Both figures come from the clients' own Search Console with the source period published on each case study, and absolute volumes are withheld where the client asked.
Read them honestly and the Valve numbers are a young site surfacing broadly and converting almost none of it. We publish that one because week seven looks like week seven, and an agency showing only percentages would have hidden it.
What none of these prove is a citation count. NovelMaker is the clearest AI-search case on the site, and its result is stated as a capability rather than a number: a product in a category with no established search term, now structured so an assistant can describe and recommend it. Nobody can hand you a verified before-and-after on how often ChatGPT names a brand, ourselves included.
Who this is a poor fit for
Anyone who needs a guaranteed placement inside a named model. That is not a product, and a vendor selling it is either misunderstanding how generative retrieval works or counting on you to.
Sites with no reachable content, no distinct entity, and no third-party mentions, on a three-month budget. The corroboration workstream runs on other people's publishing schedules, not ours, and no retainer compresses that.
Teams that need paid search to carry the quarter. We run Google Ads and Meta in-house on the $1,999 retainer, but if the honest answer is that organic will not move fast enough for your runway, we would rather say so on the call than sell twelve months of it.
AI SEO agency: common questions
What does the first 90 days actually look like?
Month one is technical and structural: crawl and rendering checks across every retriever, then page restructuring so single sections stand alone as answers. Entity consolidation follows and takes 4-8 weeks to propagate. Corroboration is the slow workstream, because it runs on other people's publishing schedules, and the prompt panel usually needs 3-6 months before a change in it is durable rather than noise. Anyone quoting weeks for the whole programme is quoting for the fast part only.
Is this different from what my current SEO agency does?
Overlapping, not separate. Roughly two thirds of the work is technical SEO, structure and entity clarity, which a good SEO agency already does. The difference is the second scoreboard and what it changes about priorities: chunking pages for standalone retrieval, checking non-Google crawlers specifically, and treating off-site corroboration as a ranking input rather than a PR nicety. If your agency is doing all four, you do not need us.
Can we buy the audit and stop there?
Yes, and a fair number of clients do. The $2,000 audit is never bundled with the implementation, on purpose: an agency that has already sold you the remediation has a reason to find remediation-shaped problems. You get the review, the prioritised backlog with effort estimates and owners, the 90-day plan and the 12-month roadmap, and you are free to hand all of it to an in-house team or another agency. If you do start a retainer within 60 days, the fee is credited against month one.
Do you work with clients outside Canada?
Yes. We are based in Toronto and work remotely across Canada and the United States, with per-market pages for several US states and cities. Where a market needs its own prompt panel and its own local corroboration, it gets one, because a national average hides exactly the market you are losing.
What access do you need to start?
Search Console and Analytics, read access to the site or its CMS, and whoever owns the templates on one call. The $2,000 audit's 3-business-day turnaround starts from the day that access lands, not the day the invoice clears, which is the condition that makes the turnaround keepable.
Do I need AI SEO if my Google rankings are fine?
Check the gap between impressions and clicks over the last two quarters before deciding. Rankings holding while clicks fall is the signature of answers resolving the query above your result. If both are steady, your category has not been compressed yet and the honest advice is to keep doing what works and re-check in a quarter.
AI SEO for Toronto businesses
SEO vs AI search: what's the difference
GEO across Canada
Related services
+4
Case studies using this work
+3
Want to be the brand the models name?