Key takeaways
- You cannot judge SEO expertise directly. You can judge reasoning, specificity and willingness to be checked.
- Compare proposals on what problem each one claims to solve, not on the list of deliverables.
- Start with a paid audit rather than a long retainer. It is the cheapest way to see how someone thinks.
- Agree before signing what you will look at in month three and what would count as it not working.
The difficulty with choosing an SEO agency is structural. You are buying expertise precisely because you do not have it, which means you cannot directly assess the thing you are paying for. Everyone you speak to will sound competent, because sounding competent is the easiest part of this industry.
So stop trying to evaluate SEO knowledge. Evaluate the things you can actually judge — how someone reasons, how specific they are willing to be, and whether the arrangement lets you check their claims later. Those correlate well with competence and you can assess them without knowing anything about search.
Step 1: Work out what you are actually buying
The phrase "we need SEO" covers at least four different jobs, and agencies are rarely equally good at all of them. Being wrong here is the most common and most expensive mistake, because the work will be competently delivered and still useless.
| What you actually need | The symptom | Who is a poor fit |
|---|---|---|
| Local visibility | Customers are nearby and choose within a few miles, but you do not appear in the map results. | An agency whose case studies are all national ecommerce. |
| Technical repair | Pages are not indexed, traffic dropped without a content change, or a migration went wrong. | A content-led agency with no engineering capability. |
| Content and demand capture | The site is technically fine but ranks for nothing your buyers search. | A technical shop that outsources all writing. |
| Commercial positioning | You get traffic and inquiries, but the wrong kind — small jobs, wrong region, no budget. | Anyone who has not asked about your margins or your best customers. |
Step 2: Shortlist on evidence, not presentation
Websites in this industry are optimized to convert you, which makes them nearly useless for comparison. Ignore the design and look for these four things instead.
- Named, described work. Not a logo wall — an actual account of what was done for someone and what changed. Vague outcomes attached to famous brands mean nothing.
- Reasoning you can follow. Anything they have published should explain why, not just what. If you cannot follow the logic, either it is not there or they cannot teach — and you will be dealing with that for a year.
- Something they refuse to do. Every serious practitioner has boundaries. An agency with no stated position on anything has no position.
- Numbers with a source. A result quoted without saying where the figure came from is not evidence. "Search Console, six months, this property" is.
Step 3: Run the first conversation properly
Most people spend the first call being sold to. Invert it. You are looking for three signals, and none of them require you to understand search.
- 01
Did they look before they spoke?
Ask what they noticed about your site. A competent answer contains two or three specific observations and appropriate hedging — "this looks like it might be X, but I would need to check Search Console." A weak answer describes problems every website has.
- 02
Did they disagree with you about anything?
Bring an idea you have about your own SEO. Someone with expertise will engage with it and may push back. Someone selling will agree enthusiastically and add it to the scope.
- 03
Can you repeat their plan back?
At the end, summarize what you think they proposed. If you cannot, it was either not explained or not there. Both are disqualifying, because you will be approving this work for months.
A longer list of specific questions — with the good and the disqualifying answer written out for each — is in questions to ask an SEO agency. It is worth having open during the call.
Step 4: Compare proposals on the problem, not the list
Deliverable lists are designed to be compared, which is exactly why they are misleading. Four articles against eight articles looks like a decision until you notice neither proposal has said what the articles are for.
Read each proposal looking for one thing: what problem does this claim to solve, and how would we know if it worked? Then rewrite each one in a single sentence of your own. The proposals that resist being summarized are the ones without a thesis.
| What the proposal says | What to ask about it |
|---|---|
| Monthly technical audit | What did the first one find? A recurring audit on a site with no technical debt is billable activity. |
| 8 blog posts per month | Targeting what, and why that many? Volume is a decision only after intent is mapped. |
| Keyword research and tracking | Which terms do you expect to lose, and which are worth winning? Tracking is not strategy. |
| Link building | From where, earned how? This is the line item most likely to transfer risk to you. |
| Monthly reporting | Show me a real one, redacted. Reports are where the relationship actually lives. |
Step 5: Start small, and define failure in advance
The single best structural decision available to you is to buy a diagnosis before buying a retainer. A paid audit is a small commitment that produces the one thing no sales call can fake: a document showing how this person thinks, what they noticed, and how clearly they explain it. It is also useful even if you hire someone else, or nobody.
- Own every account before work begins — Search Console, Analytics, Business Profile, domain, hosting
- Agree the specific indicators you will look at in month three, and where you will look at them
- Agree what would count as not working, in writing, while everyone is still optimistic
- Set a review point at month six with a real option to stop
What this does not tell you
This process will not identify the best SEO in your market. It reliably eliminates the bottom of the field, which matters far more: most money lost on SEO is not lost by choosing the second-best agency, it is lost by choosing one that was never doing anything checkable.
If you want an outside read on a proposal you have already been sent, send it over. We will tell you what we would ask about it — including when the honest answer is that it looks reasonable and you should sign it.