How to Choose and Audit an SEO Provider Using Your Own Semalt Data

Hiring SEO carries an uncomfortable asymmetry: the client cannot evaluate what they bought until six months later, having already paid six invoices. This guide tries to reduce that asymmetry to an afternoon of work, and it is written from the buyer's side rather than the seller's.

We are an agency, so it is worth stating plainly: several of the questions below are uncomfortable for us too. That is exactly why they work.

Before requesting proposals: measure yourself

The biggest advantage a buyer can have is arriving with their own data. You do not need to be technical: you need half an hour and the free side of an analytics platform.

With that photograph in hand, every proposal reads differently. And crucially: six months later you can check for yourself whether anything moved, without depending on anyone's report.

The five questions that filter

1. What will you do in the first thirty days?

A useful answer talks about audit, priorities and concrete fixes. A bad answer talks about “holistic optimisation” and “360 strategy”. If they cannot describe month one with nameable actions, they have no method.

2. Which metrics will you report, and how often?

The good answer includes positions for an agreed query set, clicks compared with the same period last year, and work delivered. The bad answer talks about total traffic and “domain authority” as a goal.

3. What do you need from us, and by when?

SEO fails more often from internal blockers than from external incompetence. An experienced provider will ask for CMS access, time from a technical person and availability of someone who knows the product. If they ask for nothing, they plan to touch nothing.

4. What will you not do?

The most revealing question of all. Someone who knows their trade has clear limits: no promised specific positions, no exact timelines, no work on structurally broken sites without fixing them first.

5. Can I speak with two clients in my sector?

Not to hear praise, but to ask three concrete things: how long before they saw anything, what they had to do themselves, and what went worse than expected.

What to ask in the monthly meeting

Three questions are enough to make a monthly meeting useful: what was done this month, what changed on the agreed queries against the baseline, and what happens next month. If the answer to the second depends on a metric that was never defined at the start, that is the problem to fix before anything else.

When the problem is on your side

Not every stalled project is the provider's fault. If recommendations sit unimplemented for months, if nobody supplies the product knowledge the content needs, or if priorities change every few weeks, no supplier can deliver. Before replacing a provider it is worth checking honestly which of the agreed inputs the company itself never delivered - that check has saved several relationships that looked finished.

Red flags that should end the conversation

The last one deserves an honest caveat: link building exists and is part of the work in competitive sectors. What is problematic is not doing it, but selling it as the objective rather than a means, with no relevance criteria and no verification that the pages carrying those links ever get crawled.

In house, agency or managed campaign

There is no single answer, but there are clear criteria by situation.

In the mixed model, which is the one we most often see working, the company keeps judgement and content and outsources volume: AutoSEO at $149 a month includes AI keyword selection starting from queries the site already touches, daily link building from a network of over 230,000 partner sites, on-site recommendations and live reporting; FullSEO, from $500 a month, adds a dedicated team, manual query control and manual donor approval. The comparison is on the pricing page.

How to read a commercial proposal

SEO proposals look very similar on the surface. The fast way to tell them apart is to look for four concrete elements and see which are missing.

A prior diagnosis, however brief. If the proposal could be sent to any other company by swapping the name, no analysis happened. A serious proposal names at least two or three specific problems on your site, and names them with data.

A calendar with named deliverables. Not “continuous optimisation”, but what is delivered in month one, month two and month three, and in what format.

An agreed definition of success. Which queries, which metrics and against which baseline. Without it, in six months the discussion will be about perceptions.

What the proposal does not include. Almost no proposal writes this down, and it is where most conflicts start: who writes the copy, who publishes it, who decides on site architecture, what happens if the client's development team is delayed.

The three moments when the relationship breaks

From experience, almost every relationship that ends badly breaks at one of three points, and all of them are predictable.

The first is month three, when initial enthusiasm has faded and results are not visible yet. It is survived with an honest report showing leading indicators - top-fifty entries, new queries with impressions, content published and discovered - instead of promising it is “almost there”.

The second is the technical blocker: the provider delivers recommendations and the client's team does not implement them. It is avoided by agreeing a fast lane for small changes up front and accepting that unimplemented recommendations are shared responsibility, not a unilateral excuse.

The third is a change of contact. When the person who hired leaves, the project loses context and usually gets cancelled for lack of memory of what was agreed. The defence is documentary: the dated baseline photograph, the agreed metrics and the report history stored somewhere the company can reach, not only the provider.

How to audit the provider you already have

If the relationship exists and doubts have appeared, verification requires no confrontation: it requires data. Three checks suffice.

Did the agreed query set move? Not total traffic, which can rise for unrelated reasons. The queries that matter, compared with the starting point.

Was the committed content published and discovered? Check it in the indexing module: if a page exists but never received verified crawler visits, something failed in execution, not in strategy.

Did the gap to competitors shrink? Growing while the market grows faster means losing relative ground, and it is the situation that takes longest to detect when only your own metrics are watched.

What must stay in your hands whatever happens

Part of the relationship does not depend on provider quality but on how things are set up from the start. These are the assets that must always be in the company's name, even when somebody else configures them.

The domain, registered to the company with your own administrative access. Hosting and the CMS, with an administrator account that does not depend on the provider. The Search Console property, with the company as owner and the provider as an added user, never the other way round. And the rank tracking panel, with the agreed queries, accessible internally.

Setting it up this way does not express distrust: it expresses continuity. A professional provider proposes it themselves, because it spares them awkward conversations on the day the relationship ends - and it always ends at some point.

Honest timelines and reasonable expectations

The first measurable signals - top-fifty entries, new queries with impressions - appear between weeks four and eight. Movement on commercial queries, between months three and six. Clear financial return, between months six and twelve, depending on competition and starting point.

Any promise that drastically shortens these timelines deserves one question: at the cost of what? The answer almost always involves risks the client will carry without knowing.

The cost of switching providers

Switching has a real cost worth calculating before deciding: four to eight weeks of ramp-up for the new team, plus the risk of losing context about what has already been tried. That is why the correct comparison is not “the new one promises more”, but whether the current one failed on execution or on judgement.

An execution failure - content not published, recommendations not delivered - can be corrected with one conversation and clear milestones. A judgement failure, where the work happens but points in the wrong direction, is what justifies switching, because more volume of the same work will not fix it.

Frequently asked questions