A national SEO programme here has a complication that buyers in most other markets never have to price: the country runs in two languages, and organic results behave differently in each of them. Budgets are usually written as though there were one audience. Then somebody notices that a third of the addressable market is being served by a machine translation of the English site, and the conversation about scope has to start again halfway through the year.
Canada is two search markets, and most SEO plans budget for one
Decide at the outset whether the French language market is in scope, out of scope, or deferred with a date attached. All three are defensible. What is not defensible is leaving it undecided, because the answer changes the writing budget, the site architecture, the agency you should hire and the way you measure results.
If it is in scope, the second decision is who writes. An agency that subcontracts French to a translation vendor will produce grammatically correct pages that miss the phrases people actually type, because search vocabulary is not a translation problem. Ask directly whether there is a French speaking practitioner on the team, what they have worked on, and whether they do keyword research in French from scratch rather than translating an English list.
Translation is not localisation, and results expose the difference quickly
The tell is always the same. A translated page ranks for nothing because it uses the formal term for a product category while buyers use the colloquial one, or it reproduces an English sentence structure that no one would ever type. Meanwhile the English page keeps performing and everyone concludes that French demand is weak.
Ask for evidence rather than assurances: two examples of phrases where the French and English demand diverge in your category, and what the agency would do about it. A competent SEO team can produce that in an afternoon. A translation workflow cannot produce it at all.
Domain and targeting decisions you only make once
Country level domains, subdirectories and subdomains all work, and the choice matters mostly because reversing it later is expensive. Ask each prospective supplier to state a recommendation and a reason, and to say what the migration would cost if you took their advice and were wrong. The same applies to language annotations: getting them wrong causes the wrong page to be served to the wrong reader, and the symptom looks like a ranking problem rather than a configuration problem.
There is a cross border version of this too. Businesses here routinely find themselves competing in domestic results against much larger operations based elsewhere, with content budgets they cannot match. That is not a reason to give up on those phrases, but it is a reason to ask the agency which battles they are advising you to decline.
National SEO reach or provincial depth
The supplier market divides fairly cleanly. There are larger groups with technical, content and digital PR functions in house, capable of running a bilingual programme and priced accordingly. There are strong regional independents who know one province, one language and one set of industries extremely well. And there are solo consultants, often ex in house, who are excellent at diagnosis and constrained on delivery.
Pick against your constraint rather than against your ambition. A business selling into one province does not need a national agency and will pay a coordination premium for one. A business genuinely operating coast to coast will exhaust a boutique within two quarters.
What a monthly SEO retainer has to state before you can compare two
Four disclosures, requested in the same words from everyone: named people and how much of their week you are buying, the deliverables that exist in an ordinary month counted in specifics, the reporting rhythm and who attends, and the notice period alongside what you keep on exit. Ask for the fee to be separated into strategy, production and running the programme, and if French is in scope, ask for the French production line to be costed on its own.
That last request settles more arguments than any other. It reveals immediately whether the bilingual offer is real or a line in a deck. If assembling identical answers from a national field is the part you would rather skip, send the requirement once and ask for replies in a fixed shape.
Accounts, glossaries and content rights
Analytics and console properties, the tag container, the content management system and the domain registration belong to accounts you own, with the agency granted access. Articles and page copy transfer to you on payment, working files included. Two SEO specific artefacts deserve naming in the contract: the redirect map from any restructure, and the bilingual terminology glossary. The glossary is the accumulated knowledge of how your buyers refer to your products in each language, it takes a year to build properly, and it is the single easiest thing to lose in a handover.
One commercial measure, reported by language
Agree a single headline number stated in your terms rather than the platform's: qualified enquiries and what each one costs. Then split it by language and by region, because a national average will hide whichever half of the programme is failing. Keep position data in the report as a diagnostic, below the commercial line, not above it.
Be patient in the right places. Technical remediation shows within weeks. Content and citations take a quarter or more in either language, and a serious SEO engagement looks flat before it looks good. Decide in advance what early evidence would justify continuing, so the review is about facts rather than mood.
If the programme turns out to be regional in practice, it is worth comparing your shortlist against firms in the largest metropolitan market or agencies on the west coast. For the wider picture of how these suppliers are structured and charged, start from the overview of SEO agencies, and if the diagnosis points at the site itself, talk to development teams working nationally or firms running several channels together.