Search “how to choose an SEO company” and you will get the same checklist eight times: define your goals, set a budget, review case studies, ask about reporting, beware of anyone guaranteeing a #1 ranking.
None of that is wrong. It is just back-to-front. Almost every item on that list is something you can only assess once you are deep in a sales conversation, being shown material the agency chose to show you. By the time you are evaluating a case study, you have already spent the scarce resource — your attention — on a shortlist you assembled from almost no information.
So this is the checklist rewritten in the order you actually encounter the decision: what you can verify before contact, what to ask on the call, and what to hold them to afterwards.
Stage one: what you can check before you talk to anyone
Three things are usually knowable without a conversation, and they are enough to build a shortlist.
Published rate band. Just over half of the agencies in our directory publish one. It tells you the cost base of the people who will do the work — a firm publishing $150–$199/hr staffs differently from one publishing under $50. It does not tell you quality. It does tell you, quickly, whether you are in the same universe on budget. (The full distribution is in what SEO actually costs.)
Team size. Of the 1,799 agencies that publish a team-size band, 81% have fewer than fifty people and 35% have fewer than ten. The word “agency” covers a two-person shop and a 900-person organisation, and the difference matters more than the label. Small usually means you get the senior people and less process. Large usually means more process and a production floor.
Whether they specialise in your situation. An agency that mostly does local service businesses and one that mostly does ecommerce catalogues are doing genuinely different work. Our lists are split this way on purpose — local SEO and ecommerce SEO are separate rankings from the general SEO list because the shortlists barely overlap.
Filter on those three and a market of thousands becomes a list of six. That is the entire job of stage one.
A word about reviews, including ours
Here is something agency directories do not usually say out loud.
Of the 753 client reviews in our directory, 716 are five stars — 95.1%. Thirty-five are four stars. Two are three stars. None are lower.
That is not a distribution that describes reality. Nobody’s client roster is 95% delighted. It is the distribution you get when reviews are solicited by the agency being reviewed, from clients the agency chose to ask, at a moment the agency chose to ask them. Every agency directory that lets firms request their own reviews has some version of this problem, ours included.
This does not make reviews worthless. It makes the score worthless and the text valuable. A five-star review that describes a specific problem, a specific approach and a specific outcome is evidence. A five-star review that says “great communication, highly recommend” is evidence only that a client existed. Read for detail, not for stars — and treat any agency’s average rating as a formality rather than a ranking signal.
The same logic applies to our rankings, which is why the methodology is a page and not a footnote.
Stage two: the four questions that are hard to answer vaguely
Most interview questions have a prepared answer. These four do not, which is the point.
“Who specifically works on my account, and how many other accounts do they carry?” The gap between the people in the pitch and the people on the account is the single most common source of disappointment in agency relationships. A confident answer names roles and gives a number. An evasive answer talks about “the team.”
“How many hours a month does this retainer represent, and how are they split?” Retainers are priced from hours whether or not the proposal says so. An agency that will not decompose it is quoting your budget, not your work. This one question does more filtering than the rest of the call.
“What would you do in the first 90 days, and what would you expect to have changed by then?” You are not testing the plan — you cannot evaluate it yet. You are testing whether they will commit to a checkable expectation this early. The honest answer usually involves the words “probably not rankings yet,” which is the correct answer and a good sign. (How long SEO takes covers what a reasonable answer looks like.)
“What does a bad quarter look like in your reporting?” Every agency will show you a good month. Very few have thought about how they will tell you when it is not working. The ones that have will describe a leading indicator they watch and a threshold at which they would change approach.
Stage three: what to hold them to
Once you have chosen, the failure mode shifts. It stops being “did we pick the wrong agency” and becomes “how long will it take us to find out.”
Three things prevent that:
- A named set of leading indicators, agreed before work starts. Pages published, pages indexed, impressions on target terms, technical issues closed. These move in weeks. Rankings and revenue move in quarters, and if they are the only things on the report you have no early warning at all.
- A standing definition of the hours. If the retainer is twenty hours, the monthly report should be legible as twenty hours of work. Drift here is usually the first symptom of an account being quietly de-prioritised.
- A review date with teeth. Six months, calendared at signing, with an explicit question: given what we now know, would we start this engagement again on these terms? It is far easier to ask that on a date you agreed in advance than to raise it cold.
The shortest version
Filter on rate band, team size and specialisation before you speak to anyone. Read review text and ignore review scores, including ours. Ask the four questions that resist preparation. Then agree, in writing, on the leading indicators that will tell you early.
Everything else on the standard checklist is real, but it is stage-two work — and most buyers are still stuck at stage one, choosing between agencies they found by searching and clicking. The rankings exist to make that first filter cheaper: SEO companies, local SEO, and by location if proximity matters to you.