The proposal is always the same — sixty pages, one per service per city. That is the exact shape of two named policy breaches, and it is also why the pages do not convert.
Key takeaways
A model can draft a service page well, and the constraint is not writing quality. It is that the standard agency proposal — a page for every service multiplied by every city you would like work from — is described in Google’s spam policies twice over. Scaled content abuse is defined there as generating many pages primarily to manipulate search rankings rather than to help users, with using generative AI to produce many pages without adding value listed as an example. Doorway abuse is defined as creating pages to rank for specific, similar queries that lead users to intermediate pages less useful than the destination, with multiple pages targeted at specific regions or cities given as an example.
Read carefully, the policy is about purpose and value, not about who typed the words. So the workable answer is: draft with AI, publish one page per service you genuinely sell, and make each one carry information only your firm has.
Section 74.01 reaches a representation to the public that is false or misleading in a material respect, and separately a statement, warranty or guarantee of performance or efficacy not based on an adequate and proper test — with the proof of the test on the person who made it. Subsection 74.03(5) requires the general impression to be taken into account alongside the literal meaning. Clause 1 of the Canadian Code of Advertising Standards requires all advertising claims to be supported by competent and reliable evidence, disclosed to Ad Standards on request.
Generated copy defaults to confidence. “Fastest turnaround in the province”, “we save clients thousands” and “99% success rate” all appear unprompted, and every one of them sits inside those provisions. The control is mechanical: maintain a claims register, and any sentence containing a number, a comparison or a superlative gets listed with its source before it enters a draft — and comes out if the source is somebody’s recollection. Treadstone Law’s guides to advertising and marketing law and misleading advertising cover the wider obligation.
If your firm has an office in a city, serves clients there, and can say something specific about practising there — the registry you file at, the tribunal you appear before, the local deadline that catches people out — that is a real page. If the only difference between two pages is a place name substituted by a script, that is the pattern the policy describes, and keyword stuffing is described in the same document as including blocks of text listing cities and regions a page is trying to rank for. Four honest pages beat forty generated ones, and they are also cheaper to keep accurate.
A marketing agency is briefed by a mid-sized accounting firm and proposes fourteen city pages across two provinces, generated from one template.
It ships four instead. Each covers a service the firm sells — corporate compilations, personal tax, an advisory engagement, and a payroll service — and each is built from a recorded interview with the partner who runs it. Every page names the province and the filing regime it relates to, sets out the steps and what the client supplies, and carries a byline. The claims register removes nine sentences across the four drafts, including two performance claims nobody could evidence.
The two offices get one location page each, containing genuinely local information rather than a substituted place name. Six months later the firm can still defend every sentence, and updating the pages after a fee change takes an afternoon rather than a sprint.
This page is about the pages a firm publishes to win work. If the underlying question is about handling the enquiries those pages produce, that is intake, and it lives on the professional practice owners page. If it is about your own accounting rather than client-facing marketing, that is accounting automation. Campaign mechanics sit under marketing automation.
Is AI-written content against Google’s rules?
No. The policy targets many pages generated primarily to manipulate rankings without adding value, “no matter how it is created”. Purpose and value are the test.
Should we say a page was drafted with AI?
No rule we could find requires it. Google’s helpful-content guidance does encourage telling readers how content was produced where that is useful, and it strongly encourages accurate authorship information. A named reviewer is the more valuable signal.
How many service pages should a small firm have?
As many as it has services it can currently deliver and describe specifically. For most Canadian professional firms that is between four and ten.
A 30-minute call is enough to tell you whether AI pays for itself here.