How to Hire an SEO Consultant

Decide what you are buying before you go shopping

Most unsatisfactory search hires were decided badly at the start, by a buyer who had not distinguished between three quite different purchases. Getting this right before the first call eliminates more risk than any amount of vetting afterward.

• Diagnosis. You do not know why the numbers look like they do, or which of six competing theories is right. You want somebody to find out and tell you, with evidence. This is bounded, produces a document, and ends.
• Execution. You already know what needs doing and lack the hands or the specialist skill to do it. You want work performed — pages built, an account managed, a migration run.
• Counsel. You have capable people and a decision to make. You want an experienced second opinion at intervals, not a program.

These have different shapes, different costs and different failure modes, and providers will happily sell you whichever one they prefer to sell. A company that needs execution and buys diagnosis ends up with an expensive document nobody builds. A company that needs diagnosis and buys execution spends a year doing careful work in the wrong order.

If you genuinely cannot tell which you need, that uncertainty is itself the problem, and a small diagnostic project is the correct first purchase.

Write the problem down in a form a stranger can price

Providers quote wide when briefs are vague, because ambiguity is risk and risk gets priced. A page of specifics will get you narrower, more comparable numbers from everyone you approach.

What to include: what changed and when, with dates; which pages or product lines matter commercially, and roughly what each is worth; what the site runs on and who can change it; whether you have a development team and how work reaches them; how many markets and languages; what you have already tried and what happened; who has to approve spending; and what a good outcome looks like in a number you already track.

Two questions here are worth more than the rest. Who will implement? — because whether implementation is included is the single biggest driver of cost, and the most commonly ambiguous line in any quote. And what does the commercial outcome actually depend on? — because a site whose real constraint is that nothing converts once it arrives does not have a search problem, and a decent provider will say so.

Where to look, and what search results are actually showing you

Searching for this yourself is instructive, mostly for what it reveals about the market. The terms return a mix of directory listings, aggregator lists titled “best SEO consultants”, and the service pages of firms who invested in ranking for them. Being on a directory list is evidence of a marketing budget, not of competence. Being ranked first is evidence somebody is good at ranking their own site, which is real but not the same as being good at yours.

Better sources, roughly in order: someone in a comparable business who has actually paid this person and will describe the experience unprompted; a developer or agency who has worked alongside them and has seen the specifications they write; conference talks and written work where the person had to be specific in public and could be contradicted; and public track records where the analysis is checkable rather than asserted.

One warning from Google's own hiring guidance, worth taking literally: be wary of firms that email you out of the blue. Unsolicited outreach claiming to have found problems on your site is a volume business, and the diagnosis in it was generated before anyone looked.

The twenty-minute screen before you take a call

You can eliminate most of a shortlist without speaking to anyone.

• Is anything guaranteed? If a page promises rankings, positions or traffic in any phrasing, stop. Google states plainly that no one can guarantee a number one ranking, and warns against providers claiming a special relationship with Google or a priority submission.
• Is the method described or concealed? Google's guidance is direct: be careful if a company is secretive or will not clearly explain what it intends to do. “Proprietary methodology” with nothing behind it is a refusal to be checked.
• Are the credentials the kind you can verify? A partner badge or a review rating can be checked in the issuing directory. An unattributed award cannot. Note which they chose to display.
• Do the case numbers have a baseline, a date range and a stated source? Most do not. A bare percentage with no window and no measurement method is decoration.
• Can you tell who does the work? If a firm's site names no individual anywhere, ask early, because the person selling and the person delivering are usually different.

Google's hiring guidance is short and worth reading before any of these conversations, including one with me.

What the first call should establish

A good first call is mostly them asking questions. If forty minutes pass and nobody has asked what your business actually makes money on, you are being pitched rather than assessed.

What you want to learn: whether they can restate your problem better than you stated it; whether they will name a probable cause and describe how they would confirm or eliminate it; whether they can describe something they would not do and why; and whether they will tell you the parts of your situation that are outside search entirely.

The most informative question you can ask is about failure. Ask what an engagement of this kind looks like when it goes badly, and what usually causes that. Someone who has done this work has a ready and slightly weary answer — implementation stalled, the sponsor left, the platform could not do what the plan required, the content never got written. Someone who says engagements do not go badly is telling you either that they have not done many or that they are not going to be honest with you when one does.

Ask, too, what they need from you. An engagement that requires nothing from your side is not going to change anything on your side.

Reading the proposal on its commitments rather than its promises

Separate two categories in every document you receive: what is promised about results, and what is committed about work. The first category is worth nothing — nobody controls the ranking system, and Google states there is no assurance that a change will produce noticeable impact. The second is the entire substance of what you are buying.

A proposal is legible when it states: exactly what will be examined or produced; who by, named; on what schedule; what you receive at the end, as an artifact rather than an outcome; what is expressly excluded; what the provider needs from you and by when; how progress will be measured, using metrics defined before work starts; and how the arrangement ends.

Deliverables written as activities rather than artifacts are the most common weakness. “Ongoing optimization”, “monthly reporting”, “content strategy” and “technical improvements” can all be satisfied by doing nearly nothing. “A prioritized technical findings document with implementation specifications for each item, delivered by a stated date” cannot.

Compare quotes only after normalizing them. Two numbers are not comparable if one includes implementation and content production and the other does not.

Structuring the engagement so it can end without damage

Whatever you sign, three provisions protect you and cost the provider nothing if they are competent.

A recorded baseline. Before any work begins, capture impressions, clicks and average position for the pages in scope on a stated date, with the query set defined, plus the conversion figures you will judge against. Without this, every later conversation about whether it worked is a memory contest. Remember that average position in Search Console means the position of your topmost result for a query, not an average across all your pages — so the number needs defining, not just recording.

Review points rather than a fixed term. Search work has a genuine lag, so a one-month judgment measures nothing; Google's own position is that some changes take days and others several months. A minimum useful term is reasonable. A twelve-month lock-in with no exit is protecting the provider's forecasting, not your outcome.

Ownership stated in writing. Your analytics and Search Console properties, your advertising account, your content, your documentation and any tooling configured on your behalf remain yours and stay accessible if the relationship ends. Agencies holding client advertising accounts under their own management identifier is common and usually reversible — but only if you established it up front.

The first ninety days, and what should worry you

Judge early progress on evidence of work and diagnosis, not on traffic, because traffic will not have moved and any provider claiming credit for movement in week three is reading noise.

By the end of month one you should have: a recorded baseline; a diagnosis with the evidence attached; a prioritized order of work with effort estimates; and at least one specification detailed enough to build. By month three you should be able to point at specific changes that actually reached production, on specific dates, and at the leading indicators those changes were supposed to move — index coverage, crawl distribution, field performance data, impressions on the affected templates.

What should worry you: reports full of activity counts with no named change; a provider who cannot tell you what shipped and when; recommendations that stay in a document across two review cycles; any reluctance to write down how something will be measured; and a sudden pivot to a new product category when the original plan has not moved. What should not worry you: a month with little visible output because the correct action was to wait for data, provided somebody says so plainly at the time.

Frequently Asked Questions

Published 2026-08-11