How to Choose an SEO Agency: 12 questions to ask before you sign
Most guides to hiring an SEO agency are written by SEO agencies who want to be hired. This one is too, so read it with that in mind. What I can offer is that the questions below are the ones I would find hardest to answer if I were bluffing, which is exactly why they are worth asking me and everyone else you talk to.
I am Nabin Thakur. I run SEO recovery work for businesses in the US, Canada, the UK, Australia and India, which means most people who contact me have already hired an agency once and it did not go well. The pattern in those conversations is consistent enough to be useful.
Two of the twelve questions below are ones I almost never hear a buyer ask. They are numbers six and seven, and they are the ones that separate an agency that diagnoses from one that guesses.
How do you choose an SEO agency? Judge three things in order. First, whether they diagnose before they pitch, because an agency that proposes a package before looking at your Search Console is selling inventory, not a fix. Second, whether their reporting connects to money rather than to activity. Third, whether the person in the sales call is the person doing the work. Price is the fourth question, not the first.
Three things to settle before you talk to anyone
Sort these out first and you will waste far less time on calls that were never going to go anywhere.
1. Decide what you are actually buying. Growth and recovery are different jobs. Growth means you rank acceptably and want more. Recovery means you used to rank and no longer do, and the cause has to be found before anything is built. Agencies that are good at one are often mediocre at the other, and most will happily sell you either.
2. Know your real budget, and know what the bands buy. Published benchmarks put a standalone SEO audit somewhere between $1,000 and $10,000, and a full recovery project between $5,000 and $25,000. Retainers vary far more widely than that and are much harder to benchmark honestly, which is why I am not going to invent a figure. What matters is the question underneath: how many hours go into your account each month, and who spends them. Our own pricing is published in full if you want a reference point.
3. Decide how you want to work. Some businesses want a weekly call. Some want to be left alone and read a monthly summary. Neither is wrong, but an agency built around one will frustrate you if you want the other. Say which you are on the first call.
The 12 questions to ask any SEO agency
1. Can you show me three live client sites right now?
Not case studies. Live URLs, on the call. A confident agency shares a screen and shows you. The dodges to watch for are the PDF deck, the promise to send examples later, and the claim that all client work is confidential. Some client work genuinely is confidential. All of it never is.
2. How will you measure success, and does it connect to revenue?
A good answer names organic revenue, qualified leads, or bookings, with rankings and traffic as supporting evidence. A weak answer stops at rankings and impressions. Impressions are the easiest metric in SEO to grow without earning a single customer, which is precisely why struggling agencies lead with them.
3. Who does the work, and what hours do they overlap with mine?
Ask for a name, not a role. Then ask what that person's working hours look like in your time zone. Hiring across borders is completely normal and search engines do not care where your agency sits, but a four hour overlap and a twelve hour overlap are very different experiences when something breaks. The failure mode is not distance. It is discovering in month three that the senior person from the pitch has never touched your account.
4. Do I own the Search Console and Analytics properties?
The right answer is that the properties are yours and the agency works inside your accounts with delegated access. If an agency insists on holding your Search Console, or reports only through their own dashboard, ask why. When that relationship ends you need to leave with your own historical data. People who have been through a bad exit ask this question first.
5. What is your link building approach, and can you name a placement?
Reasonable answers involve editorial placements, digital PR, niche directories, or genuinely useful assets that attract links. Ask them to name one site they placed a link on in the last quarter. Volume promises like "500 backlinks a month" describe a link farm, and the cleanup costs more than the campaign did.
6. What was Google's update calendar doing when my traffic changed?
This is the question almost nobody asks, and it is the most revealing one on the list. Google ran a confirmed update on 503 of the last 1,704 days. Roughly three days in ten, something was moving. So "it was probably an algorithm update" is not a diagnosis, it is a coin flip dressed as expertise.
What you want is someone who opens your Search Console, lines your drop up against the specific confirmed update windows, and tells you which one it overlaps and which it does not. Sometimes the honest answer is that your drop sits in a gap between updates, which means the cause is on your site and not in the algorithm. That answer is worth more than a confident guess. Our update history tool tracks 41 confirmed windows if you want to run the check yourself first.
7. Can you show me whether AI assistants currently cite my site?
An increasing share of buying research now happens inside ChatGPT, Copilot, Perplexity and AI Overviews, where being cited matters more than being ranked. This is measurable. Bing Webmaster Tools reports citation counts and citation share, and the data is free to anyone who verifies the site.
Ask the agency to pull it. If they have not heard of it, that is not automatically disqualifying, but it tells you how current they are. If they answer with vague talk about "optimising for AI" and no numbers, that is a different signal.
8. What is the contract, and how do I leave?
Read the exit clause before the deliverables. Long minimum terms are not automatically predatory, since real SEO work does take months to show, but a long lock-in with no performance conditions attached is all risk on your side. Ask what happens to your content, your links and your account access if you leave in month four.
9. Who writes the content, and how much of it is AI-generated?
Ask directly, and expect a straight answer. AI in the drafting process is now normal and is not by itself a problem. Publishing unedited AI output at volume is a problem, and it is one of the most common causes of the slow decay I get called in to reverse. Ask to read one piece they published last month, then judge whether a human clearly had the final say.
10. Have you worked in my industry and my market?
Industry matters most where the rules are strict. Medical, legal and financial sites carry real E-E-A-T constraints, and a generalist can quietly cause damage. Market matters less than people assume, with one exception: if your customers are local, the agency needs to understand your local search landscape specifically, not just search in general.
11. What happens if my rankings drop?
Every site drops sometimes. The answer you want describes a process: investigate the cause, separate an algorithm effect from a site problem, then fix it within the existing scope. The answer to walk away from is that your rankings will only ever go up.
12. What is excluded from the price?
Most proposals are vague here deliberately. Push for specifics. Is long-form content extra? Are developer hours included when the fix needs code? Who pays for tools? Who pays for link placements? The gap between what you assumed you were buying and what was actually scoped is where most agency relationships sour.
Five warning signs
Google says plainly that nobody can guarantee rankings. Anyone offering one is either uninformed or counting on you being uninformed.
Rather than name a universal floor, do the maths yourself. Divide the monthly fee by a plausible hourly rate in the agency's own market. If the answer is three or four hours a month, you are buying an automated report and nothing else.
Tasks completed, keywords tracked, backlinks built. None of it answers whether the business made more money. Ask what changed for the client, not what the agency did.
Twelve months is defensible if there are milestones you can hold them to. Twelve months with no milestones is a subscription, not an engagement.
Ask what specifically they will do in the first thirty days. Anyone who has done this work has a clear answer. Anyone who has not will retreat into strategy language.
Five things that should reassure you
1. They look before they quote. An agency that asks for Search Console access, or runs a diagnostic, before naming a price is treating you as a problem to solve rather than a slot to fill.
2. They tell you something you will not pay them for. The willingness to say "this part is fine, do not spend money on it" is the single most reliable honesty signal I know.
3. They disqualify themselves sometimes. An agency that says your budget does not fit, or your problem is outside their competence, is protecting your money and their reputation at once.
4. Their published work has numbers in it. Case studies with dates, figures and specifics are checkable. Case studies with percentages and no context are decoration.
5. They put pricing in public. Not every good agency does this, so it is not a rule. But an agency that publishes rates has removed its own room to price you by how desperate you sound.
How to compare agencies fairly
Proposals are designed to be hard to compare. Every agency bundles differently, names things differently and emphasises whatever it happens to be good at. The fix is to stop reading proposals side by side and instead score all of them on identical rows.
| What to score | Agency A | Agency B | Agency C |
|---|---|---|---|
| Hours on your account per month | |||
| Named person doing the work | |||
| Who owns Search Console | |||
| What the monthly report shows | |||
| Minimum term and exit terms | |||
| What is explicitly excluded | |||
| Answer to question 6 |
Three is the right number to shortlist. Two gives you no baseline. Five turns into a research project and you will pick on gut feel anyway.
Five expensive mistakes
1. Buying growth when you needed recovery. If your traffic fell, adding new content on top of an undiagnosed problem usually makes the problem bigger. Find the cause first. This is the single most costly mistake I see, and it is almost always made in good faith.
2. Choosing on price alone. The saving is small and immediate. The cost is large and delayed, which is exactly the shape of decision humans get wrong.
3. Letting the agency hold your accounts. It feels like a convenience at signup and like a hostage situation at the exit.
4. Expecting results on a schedule the algorithm does not follow. Across fourteen measured update-to-update gaps, the wait between a fix and a visible recovery has ranged from weeks to several months. Any agency promising a date is promising something outside its control.
5. Not asking what happens when it goes wrong. Every engagement has a bad month. You want to know in advance what the agency does in it.
Final thought
The best agency for you is the one that shows you real work, tells you things you did not want to hear, prices in the open, and gives you a straight answer to question six. That is a low bar in principle. In practice it removes most of the market.
If you are based in India and paying in rupees, the pricing landscape and the agency market are different enough to deserve their own guide. Read the India version instead.
We answer all twelve in writing before you pay anything, question six included. If the honest answer is that we are not the right fit, we will say so.
Start a conversation See our pricingRelated guides
- How to choose an SEO agency in India, the rupee-priced version of this guide
- How long Google recovery actually takes, measured across fourteen update gaps
- Google update history, 41 confirmed windows you can check your own drop against
- What SEO recovery work involves