What is SEO consulting and what does a consultant do?
What an SEO consultant actually delivers week to week, and what they do not. Scope boundaries, decision rights, and how it differs from an agency.
By Roozbeh Nazari · CEO
SEO consulting has become an umbrella term covering very different kinds of work. For one company it means an outside supplier that sends a monthly report, for another a team that produces content, for a third a specialist who only runs technical audits. As long as that ambiguity sits on the table, the engagement turns into a clash of expectations in its very first month: one side expects strategy while the other delivers execution, and each thinks the other is not doing their job.
This article describes the concrete content of a consulting relationship. Clarifying what gets delivered, what does not, and who holds which decision is the first thing to do before choosing a service scope. What follows is not an advertisement for agency services; it is a breakdown of a way of working that repeats in the field.
A consultant’s real output is decisions, not deliverables
What an SEO consultant produces is not content or code; it is decisions. Which set of pages gets written, which technical debt gets paid down first, which language takes priority, which page gets merged and which gets removed. Each of these costs far more than the hours of execution behind it: an information architecture set up wrongly means rewriting hundreds of pages six months later.
That is why the first thing to measure in a consulting relationship is not the volume of content produced but the accuracy of the decisions. A good consultant states up front the conditions under which their own recommendation would turn out to be wrong, and agrees on the indicator that would show it. Making a recommendation reversible is worth more than defending it well.
What a consultant’s week actually produces
Consulting is not abstract work; it has tangible weekly outputs. In a typical working week, the things put on the table are these:
- Diagnosis: an observation taken from the live environment and the data it rests on. For example, a particular template showing the wrong title in search results, and the demonstration that this comes from a template variable.
- Prioritisation: open items ranked on impact and cost. The ranking comes with its reasoning, because the reasoning has to be arguable.
- An actionable specification: an item written at a level a developer or editor can apply without further interpretation. "Fix the canonical" is not a specification; which template, under which condition, producing which value — that is a specification.
- Verification: a check on whether the previous week’s items were actually applied in production. An unapplied item comes before producing new ones.
- A decision log: a short, reasoned note of the decisions taken that week and the alternatives rejected. When the team changes, this note is the institutional memory.
Everything outside this list — presentation aesthetics, report page counts, tool screenshots — adds nothing to the value of the relationship. A five-item decision log does more work than a forty-page monthly report.
What a consultant does not do
Describing what falls outside the scope matters as much as describing what falls inside it. In a consulting relationship, the work that typically does not belong to the consultant is this:
- Publishing operations: entering content into the CMS, preparing images, running the publishing calendar. The consultant supplies the template and the criteria; the in-house team or a separate production team runs the operation.
- Development: writing and shipping code. The consultant writes the requirement and the acceptance criteria, the engineering team implements it, and the consultant verifies the result.
- Commercial decisions: pricing, service portfolio, target market selection. The consultant shows what search demand says about those decisions; the business makes the call.
- Outcome guarantees: a commitment to a specific ranking or a specific traffic figure. Why that is not possible is covered separately below.
Agency, consultant, in-house team: dividing the roles
The difference between the three models is not budget but capacity and decision rights. The agency model sells production capacity: writers, developers and specialists arrive together and most of the work happens outside. The consulting model sells decision capacity: production stays in-house and what comes from outside is direction and review. In the in-house model both sit inside, and what is bought externally is at most a periodic second opinion. How these three combine is what we set out, stage by stage, in our way of working.
The choice depends on which resource the company is short of. Selling consulting to a company with nobody to write is pointless; so is selling a strategy project to a company that has no direction problem but a production bottleneck. The most common mistake in practice is mistaking a production bottleneck for a lack of strategy: a new strategy document does not solve an unpublishable-content problem.
A hybrid model is also possible and is usually the most realistic: outside help for technical implementation, an in-house team for content, and a single decision-maker above both. What matters is that the decision-maker is singular. Arrangements where two separate parties set priorities without knowing about each other are the most expensive arrangements there are.
Leading indicators instead of guarantees
A ranking commitment is the least trustworthy signal in a consulting relationship. Google’s own documentation is explicit on this: nobody can guarantee a number one ranking on Google, and providers who guarantee rankings are counted as a warning sign. The same applies to being cited in AI answers; which source gets quoted is not under the provider’s control.
Leading indicators take the place of guarantees. These are measures that move before the outcome and can be influenced directly: the share of target pages indexed, the count of technical errors per template, publishing pace against plan, the internal link depth of pages within the target set. If those indicators improve while the outcome indicator does not, the diagnosis is wrong — and seeing that does not require waiting six months.
The mark of a healthy consulting relationship is this: both sides have defined together, at the outset, the indicator that would show the work is not going anywhere. Without that definition, the success of the engagement can only be defended with a retrospective narrative — which is persuasion, not measurement.
The first thirty days: the diagnostic window
The most critical period of a consulting relationship is the first thirty days, and the output of that period is not content but a shared assessment of the current state. Three things need to happen in this window: mapping the site’s current condition at template level, verifying what the measurement setup is actually measuring, and reaching a written agreement on who decides what.
The second item is usually skipped and gets expensive later. Decisions taken before the measurement setup has been verified rest on the wrong data: in a setup with a broken goal definition the conversion count does not reflect reality, and unfiltered internal traffic looks like organic growth. The consultant’s job in the first month is to audit the measurement itself; every recommendation made without doing so is groundless.
The document that should exist at the end of thirty days is short: two pages at most. It contains a summary of the current state, the items to be closed in the first quarter, the reasoning behind them, and the thresholds that would show the relationship is not working. Without that document, the argument in later months turns on what people remember rather than on what the data says.
Negotiate decision rights, not price
Negotiation over consulting proposals is mostly conducted on the monthly fee, when the item that actually determines the cost of the relationship is where decision rights sit. If applying the consultant’s recommendation requires separate approval every time, the pace of the relationship becomes tied to the approval cycle and the fee stops buying what it should. Conversely, an outside party that can ship to production with no approval at all makes institutional risk unmanageable.
The arrangement that works in practice is to separate decisions by risk level: reversible, narrow changes are applied directly, while structurally expensive decisions go to a single approver. Making that distinction in the working document rather than the contract is enough; what matters is that it is written down.