How to choose an SEO agency
The questions worth asking, the answers that should worry you, and the two checks almost nobody runs.
Start with what you need
Most bad agency choices are made before any agency is contacted, by not deciding what the job is.
A technical problem, a content problem and an authority problem need different agencies, and an agency strong at one is often ordinary at the others. If your site is slow, broken and invisible, you have a technical job. If it is fine and nobody links to it, you have an authority job. Those are not the same purchase.
The questions worth asking
Who actually does the work? The account manager is usually somebody else. Ask for the name and how much of their week you are buying.
Show me a case study for a business like mine, with the period stated. Not a percentage on its own. A percentage with no start date and no client name cannot be checked by anyone, which is usually why it is written that way.
What will you measure, and what will you report? If the answer stops at rankings and traffic, ask how that connects to leads or sales. If it cannot, you will be arguing about it in month nine.
What happens in month seven? The revealing question. Month one is always an audit. Month seven is where you find out whether there is a plan.
What would make you tell me to stop spending? A good agency has an answer. It is one of the fastest ways to tell an adviser from a supplier.
Answers that should worry you
A guaranteed position. Nobody controls the results page. A guarantee means either a term so obscure that ranking first is worthless, or a promise they cannot keep.
Proprietary methods they cannot describe. Every agency uses broadly the same tools and the same levers. Vagueness here is usually covering for either link buying or very little work.
Case studies with no client named and no period given. One is a confidentiality agreement. A library of them is a pattern.
Reporting that only ever goes up. Real campaigns have bad months. A report that never shows one is a report that has been curated for you.
Two checks almost nobody runs
Check a case study against the data. Take a named client from their site and look at that client’s own traffic trend over the period claimed. We did this for every agency in our ranking and it separated the field more than anything else.
Sometimes the claim is real and the window is simply earlier than the page implies. Sometimes the growth is a third of what is claimed. Both are worth knowing before you sign.
Ask for a client who left. Every agency has them. How they talk about why someone left tells you more than three happy references will.
Before you sign
Get the notice period in writing and check it is not twelve months. Get the deliverables specified as quantities, not adjectives: four pieces a month, not comprehensive content support.
Agree the baseline before work starts, and agree who owns the accounts. If the agency owns your Analytics property or your Search Console, leaving becomes much harder than it needs to be, and some of them know that.
FAQ
Common questions
How many should I talk to?
Three is usually enough to calibrate and few enough to compare properly. More than five and the process becomes its own project.
Should I pick the one that ranks first for "SEO agency"?
It proves they can rank a page in a competitive market, which is worth something. It does not prove they can do it for a business like yours, and their own site is the one client they can take every liberty with.

About the author
Sarah Rice
Editor
Sarah Rice is a PR specialist & digital marketing expert who has spent more than 30 years working with SEO & GEO agencies, PR agencies and brands across a huge range of sectors. She is Director of Rice PR and Director & Editor of The HR World. Earlier in her career, she was Head of Public Affairs and Press Manager at RAC and Head of Campaigns at JBP, and has also worked with DC Thomas and a number of charities.