Key takeaways
- Ask what they would do in month one. A competent answer starts with diagnosis, not deliverables.
- Any promise of guaranteed rankings or a named position is a disqualifier — nobody controls Google’s index.
- You should own the Search Console, Analytics and Business Profile accounts. Not the agency.
- Ask which keywords they would ignore. An agency that cannot name any is chasing traffic volume, not customers.
Hiring an SEO agency is unusually hard to do well. The work is invisible for months, most of it happens in tools you do not have access to, and the vocabulary is dense enough that a confident salesperson can talk for an hour without saying anything checkable. So the interview matters more here than in almost any other service you buy.
These are the questions worth asking, grouped by what they are actually testing. For each one there is a version of the answer that tells you the agency knows what it is doing, and a version that should end the conversation. You do not need to understand SEO to use this — you only need to notice which kind of answer you got.
Questions about how they diagnose
Good SEO starts by finding out what is actually wrong. An agency that arrives with a fixed package has decided what is wrong with your site before looking at it.
Diagnosis
1. What would you do in the first month?
Good: an audit, a Search Console review, and a prioritized list of problems — no publishing yet. Bad: “we start with 10 blog posts and 50 directory submissions.” That is a plan written before they saw your site.
2. What do you think is currently holding this site back?
Ask this on the first call. Good: two or three specific observations, hedged appropriately, from an actual look at your site. Bad: generic statements that would apply to any website on earth.
3. Can you show me a real audit you have delivered?
Good: a redacted example with specific findings and the reasoning behind them. Bad: a tool export with a colour-coded score and no interpretation. Anyone can run a crawler; the value is in knowing which of the 400 issues matter.
4. What have you decided not to do for a client, and why?
This is the best single question on this list. A real practitioner has talked a client out of something. Someone who has never declined work is selling whatever is asked for.
Questions about results and promises
This is where most bad engagements are decided, and it is the easiest section to judge, because the disqualifying answers are unambiguous.
Promises
5. Can you guarantee I will rank number one?
The only acceptable answer is no. Nobody has a mechanism to guarantee a position in an index they do not control. An agency that says yes is either overstating what it can do, or planning to rank you for a phrase nobody searches — a long enough phrase ranks first easily and is worth nothing.
6. How long before I see results, and what moves first?
Good: a range, an explanation of why it is a range, and a distinction between early indicators and revenue. Bad: a confident single number, especially a short one.
7. What would you show me at month three to prove this is working?
Good: impressions and average position in Search Console, indexation coverage, and movement on specific tracked terms — leading indicators, because clicks lag. Bad: “traffic will be up,” or a metric that only exists inside their own dashboard.
8. What happens if it does not work?
Good: a straight answer about diagnosis, changing approach, and how the contract ends. Bad: deflection, or a guarantee that quietly redefines “work” as activity delivered rather than outcomes achieved.
Questions about keywords and intent
Traffic is not the product. Customers are. These questions test whether the agency knows the difference.
Targeting
9. Which keywords would you deliberately ignore?
Good: high-volume informational terms that attract researchers rather than buyers, or terms where the results are dominated by a format your business cannot produce. Bad: “we target everything relevant.” Every budget has a limit, and refusing to prioritize means the priorities get set by whatever is easiest.
10. How do you work out what someone searching this actually wants?
Good: they look at what Google currently ranks for the term, because that is Google’s own answer to the question. Bad: keyword volume and difficulty scores with nothing behind them.
11. How many of these searches are likely to become inquiries?
Good: an honest breakdown — most terms will not, a few will, and here is which is which. Bad: a projected traffic figure multiplied by an invented conversion rate and presented as a forecast.
Questions about ownership and access
This section is boring and it is the one that costs people the most money. Access decides whether you can verify anything, and whether you can leave.
| Asset | Who should own it | What happens if the agency owns it |
|---|---|---|
| Google Search Console | You. The agency gets added as a user. | You lose all historical performance data — the only record of whether the work did anything. |
| Google Analytics | You, on your own account. | Your entire traffic history disappears when access is revoked. |
| Google Business Profile | You, as primary owner. | Worst case on this list. Recovering a profile someone else owns is slow and sometimes fails. |
| Domain and hosting | You, always. | Your website becomes a hostage in a commercial dispute. |
| Content produced for you | You, in writing. | Some contracts allow published content to be removed on termination. |
Access
12. Whose name will the Search Console and Analytics accounts be in?
Good: yours, with them added as a user. Bad: any version of “we handle that on our side.” Insist. This is not negotiable and it costs nothing to do correctly.
13. What do I keep if we stop working together?
Good: everything — accounts, data, content, documentation. Bad: hesitation, or content licensing terms that expire.
14. Can I see the raw Search Console data, not just your report?
Good: yes, and you already have access. Bad: a report-only relationship. A report is an interpretation, and you should be able to check it against the source.
Questions about the work itself
Execution
15. Who will actually do the work, and who writes the content?
Good: named people, and a clear answer about how AI is or is not used. Bad: you meet a strategist on the call and never hear from them again.
16. Where will links come from?
Good: earned coverage, genuine partnerships, digital PR, or a frank “we focus on on-site work and do not buy links.” Bad: a monthly quota of links, or anything described as a “network.” Paid links breach Google’s spam policies, and the risk sits on your domain, not theirs.
17. What will you need from me?
Good: a real list — subject-matter input, approvals, developer time, photographs. SEO needs things only you have. Bad: “nothing, we handle everything.” That answer sounds convenient and describes work done without your business in it.
How to score the conversation
You do not need a spreadsheet. After the call, ask yourself three things.
- 1Did they tell me anything I could go and check myself in the next ten minutes? Specificity is the signal.
- 2Did they say no to anything, or disagree with me about anything? People with expertise push back. People selling do not.
- 3Could I explain their plan to a colleague tomorrow? If not, either it was not explained or there was not one.
If you want a second opinion on answers you have already been given, send them over. We will tell you which ones were reasonable — including when the agency you are considering is the right choice and we are not.