12 questions to ask before hiring an SEO consultant

Choosing an SEO provider gets reduced, in most companies, to comparing proposals: price, duration, number of line items. Yet proposals largely resemble one another; the difference shows up in how the provider answers questions. The twelve questions below can be asked in about forty minutes in a single conversation, and they tell you far more than the proposal document.

The questions fall into three groups: diagnostic ability, scope and working method, and measurement. Under each question is what to listen for in the answer. You can ask these of our own service scope too; the aim is not to favour one provider but to make the comparison standard.

Diagnostic ability: 1-4

The first four questions measure whether the provider actually looked at your site before the conversation, and how they interpret what they saw.

• 1. What are the three most expensive problems you can see on our site right now, and why in that order? What to listen for: a specific page or template should be named, and the ranking should be built on a cost-and-impact axis. A generic list of items is the sign that nobody looked at the site.
• 2. How did you arrive at that diagnosis, and what data did you use? What to listen for: not the name of the tool but the place they looked. The source of the live page, its appearance in search results, the server response — verifiable observations.
• 3. Which of these problems is specific to our sector and which is the same on every site? What to listen for: the ability to tell the two apart. Declaring everything sector-specific and distinguishing nothing at all are equally incomplete.
• 4. Where might you be wrong? What to listen for: a genuine area of uncertainty. A provider who answers "we are not wrong" will go on defending a wrong diagnosis in the months ahead.

Scope and working method: 5-8

The second group clarifies the daily mechanics of the relationship and the load on each side. Ambiguity here turns into an invoice in the third month of the contract.

• 5. Which work is yours and which is ours? What to listen for: a written, singular list. "We run it together" means, in practice, work nobody does.
• 6. How does the plan change if our developer capacity is limited? What to listen for: the ability to gather low-dependency items into a separate set. A plan where every recommendation needs a developer never gets applied in most companies.
• 7. Who will write the content and where will the subject-matter expertise come from? What to listen for: a concrete procedure for obtaining expert input. In health, legal and finance, this question is not negotiable.
• 8. What will you have delivered in the first ninety days? What to listen for: deliverables, not outcomes. An answer promising rankings in ninety days makes all the remaining questions unnecessary.

Measurement and reporting: 9-12

The last group determines whether the relationship can be audited. If the definition of measurement is not set at the outset, the argument about results becomes a contest of narratives.

• 9. Which indicator will we measure success by, and where will it be read from? What to listen for: a source you can access too. An indicator visible only in the provider’s dashboard is not auditable.
• 10. How will we know the work is not going anywhere? What to listen for: a threshold and a date defined in advance. A provider with an answer to this question separates themselves seriously from one without.
• 11. Which decisions will have their reasoning recorded in the report? What to listen for: a decision log. A report showing only metrics will not remind you, three months on, why you took that path.
• 12. If the contract ends, what are we left with? What to listen for: documentation, access, and transferable assets. Arrangements locked inside the provider’s own infrastructure raise the cost of leaving invisibly.

Warning signs in the answers

Some answers, regardless of their content, tell you in advance how the relationship will go. Google’s documentation on choosing a provider names several of them outright: guaranteeing rankings, implying a special relationship with Google, promising priority submission to the search engine. The moment any of these phrases appears, there is no need to evaluate the rest of the proposal.

There are three quieter signs as well. First, the provider who asks no questions during the diagnostic stage: diagnosing without asking means applying a ready-made template. Second, the provider who ties every answer to a tool screenshot; the tool shows the finding, not the interpretation. Third, the provider who offers only names as references without describing the scope of the work — a brand name carries no information when nobody knows what was done. That is why, on our side, we write out what work was done and at what scope on our results page.

Finally, the point of these twelve questions is not to put the provider through an exam. The point is to be sure both sides are describing the same job. At the end of the conversation, notice what the provider asked you: a party that asks nothing is selling their own template, not doing your work.

How to structure the conversation

These twelve questions are not a survey but the skeleton of a conversation. Order matters: the diagnostic questions come first, because that is where you find out fastest whether the provider genuinely looked at the site, and that determines how the remaining answers should be read. Scope questions sit in the middle; measurement questions are left to the end, because they only get meaningful answers once scope is clear.

It also helps not to go in alone. Having one person from marketing and one from the technical side listen together means the answer is judged by two different sets of ears; an answer that is technically weak but persuasive usually only catches the technical ear. Having both take notes separately after the conversation and compare them produces a noticeably more accurate assessment than judging alone.

There is a timing question too: it is more productive to ask these before requesting a proposal than after receiving one. Once the proposal is written, the provider moves into defending their own document; beforehand, the same questions turn into a real diagnostic conversation, and the proposal then gets written on top of it.

Making the answers comparable

When you speak to three providers, comparing the answers from memory is impossible; the winner ends up being whoever spoke last. A simple fix is to keep a three-level note for each question: concrete and verifiable answer, general but reasonable answer, evasive answer. The note records the type of the answer rather than its content, and in that form it can be compared directly across providers.

When that table is complete, the pattern that emerges is usually clearer than expected. Evasive answers clustering under a particular heading — most often measurement and exit terms — carry more information about how that provider works in that area than a reference list does. A single evasive answer is not grounds for elimination; three in a row under the same heading show you in advance the subject you will be arguing about in the months to come.

Finally, run the comparison not only between providers but against your own expectations. Laid side by side, the answers to these twelve questions lead most companies to see that what they were actually looking for was not an SEO provider but a missing internal role. Seeing that early is far cheaper than spending twelve months in the wrong relationship.

Sources

• Do You Need an SEO? — Google Search Central
• Google Search Essentials — spam policies and technical requirements