How to choose an SEO agency, using checks you can run yourself.

Why this decision is unusually hard to make well.

Most purchases let you inspect the thing you are buying. This one does not. The work is technical, the results arrive slowly, and the person selling knows far more than the person buying. That gap is the entire reason bad agencies survive.

The usual advice makes it worse. Ask for case studies, and you receive selected wins with no baseline. Ask for references, and you speak to the happiest client. Ask about method, and you get a diagram with the word discovery in a box.

The way through is to stop evaluating expertise you cannot judge and start evaluating evidence you can check. An agency’s own visibility, its willingness to publish a price, its clients’ live sites, and its handling of your baseline are all verifiable without knowing anything about search.

What is the fastest way to evaluate an SEO agency?

Do they rank for what they sell?

An agency’s own site is the one project with unlimited creative control, no client constraints, and every incentive to excel. If it does not rank for the service it sells in the market it claims, that is the most informative single fact available to you.

This is not a perfect test. A firm can be excellent and deliberately invisible, working entirely on referral. But it is the cheapest test there is, and a firm that is invisible for its own category owes you an explanation rather than a deflection.

Run it specifically. Not the brand name, which any site ranks for, but the service plus the market. If they sell SEO in a metro, search that.

Will they tell you the price?

Published pricing is rare in this industry, and the reason is negotiation rather than complexity. Disclosure invites comparison, and comparison compresses margin. The explanation offered is that every business is different, which is true and is not why the number is hidden.

The practical effect is that a buyer cannot assemble a shortlist without booking several calls, each a sales conversation before it is an information one. That asymmetry is the product.

You do not need the price published to proceed. You need the agency willing to state one in writing before a call. Ask, and treat the answer as data.

Can you verify the clients?

A portfolio of screenshots proves ownership of a screenshot tool. Live links prove the work exists and let you test speed on your own phone, check whether the business ranks for anything a buyer would type, and read whether the pages answer real questions.

Confidentiality is a real constraint and a smaller one than it is used as. Some clients genuinely cannot be named. An entire roster that cannot be named is a different situation and worth asking about directly.

When you can open the sites, spend thirty seconds each. Load on cellular. Search one term a customer of that business would actually type. That is enough to tell you whether the work functions.

What single question separates good agencies from bad ones?

How do they handle your baseline?

This is the one that matters most and the one almost nobody asks. A baseline is a record of where your site stands before any work begins, across the terms and markets that matter to you. Without one, every later claim of improvement is unfalsifiable, which is why it is the first thing to ask for.

Without it, progress is whatever the report says it is. With it, every claim is checkable against a document that predates any incentive to exaggerate. The agencies that record one are usually the agencies doing the work, because a baseline is only uncomfortable if the work is thin.

Ask what the baseline covers, who produces it, and when you receive it. Ask to see a sample from another engagement with the client details removed. The response tells you a great deal.

Run the same four checks here.

We publish pricing at 100 and 250 dollars per month, billed monthly or annually, so check two is answered before you contact anyone. Client sites are linked live from the portfolio rather than shown as images, so check three takes thirty seconds each.

Check one you should run yourself rather than take our word for. Search the services on this site in the markets it claims and see what you find. Check four is answered by the free audit, which documents your baseline before any proposal exists and leaves the findings with you either way.

None of that makes us right for you. It makes us checkable, which is a different and more useful claim.

What you should own regardless of who you hire.

Your hosting and your domain registrar, in accounts registered to you. This sounds obvious and is the single most common thing agencies quietly hold, which turns leaving into a negotiation rather than a decision.

Your analytics and Search Console properties, with you as an owner rather than a delegated user. If access was granted to you rather than created by you, it can be withdrawn.

Your content and your site, built on standard tools any competent professional can maintain. A proprietary platform makes leaving expensive by design, and that expense is a feature for whoever built it.

Ask about all three at the start. The answers are short, and an agency that hesitates on any of them has told you what the exit looks like.

What a bad engagement actually costs.

The obvious cost is the fee, and it is the smallest part. A year of a thin retainer at a modest monthly rate is recoverable. What is harder to recover is the year.

Search compounds, which cuts both ways. Twelve months of correct foundational work produces positions that keep earning. Twelve months of reporting theatre produces nothing to build on, and the competitor who started at the same time is now a year ahead.

The worst outcome is active damage. Purchased links, scaled thin content, or a migration that discarded earned equity all leave you behind where you started, and unwinding that takes longer than the original work would have.

This is why the checks matter more than the price. The difference between a good and a bad engagement is not the monthly fee, it is whether the year was spent or wasted.

What honest reporting looks like, so you can recognise it.

Everything reads against a baseline recorded before any work ships. That single document is what turns a monthly report from an assertion into something checkable, and its absence is the most common reason a client cannot tell whether they were served well.

AI search adds a wrinkle worth knowing about. Ask the same question twice and the cited sources can differ, so a single check proves very little. The method that survives that is a fixed panel: agreed questions, rerun on a schedule, logged with dates. Fixed questions cannot be quietly swapped for easier ones.

Single months are noise and direction across months is signal. Both get reported, including the flat ones. A report showing progress every single month is describing something other than reality, and recognising that is a useful skill whoever you hire.

What a real engagement produces early, at any agency.

Month one should produce two artifacts: a documented picture of what your buyers actually search, and a baseline recording where you stand today. If month one produces a strategy deck and nothing else, ask what happens to it.

Months two and three should produce shipped work you can open in a browser. Pages that exist, technical faults closed, profile and entity facts made consistent. Work that produces no artifact is difficult to distinguish from work that did not happen.

After that the program compounds, and the monthly read should tie back to the baseline rather than to a fresh set of numbers each time. Use this as a checklist against whoever you hire, including us.

What this practice will not do, whatever the incentive.

No purchased links, no exchanges, and no private network placements. All three are named directly in Google’s spam policies as link schemes, and the consequence attaches to your domain rather than to us.

Two companion pages cover the rest of this decision: SEO agency vs freelancer and SEO agency red flags. Our own numbers are on the pricing page, the roster is on the portfolio, and the proof page sets out the evidence standard.

No guaranteed rankings and no guaranteed timelines. Nobody controls a search engine. What can be committed to is the work, the measurement, and the honesty of the reporting, and those are the only commitments worth having.

No content published at volume for its own sake. Scaled thin content is named in the same spam policies, and it is also simply a waste of the budget it consumes.

No client kept by paperwork. Month to month, no notice period, no early termination fee. If the work stops earning its place you should be able to act on that the same month.

An agency writing the buyer’s guide is a conflict of interest.

There is an obvious problem with an agency publishing advice on choosing an agency. Every criterion can be selected to favour the author, and most pages of this kind are exactly that exercise.

The defence is to use criteria that are checkable independently and that we could fail. Whether a firm ranks for what it sells, whether it publishes a price, whether its clients are verifiable, and whether it records a baseline are all things you can confirm without asking us anything.

Read this page assuming it is self serving, then run the checks on us first. If we fail one, that is more useful to you than the page is.

Straight answers, in depth.