The decisions that get made in a week, with nobody qualified in the room
A developer proposes moving the blog to a subdomain. An agency recommends a disavow file. A platform migration is scheduled for the same quarter as a content launch. Somebody wants to consolidate four regional sites into one. A vendor sends a proposal promising a first-page placement. Each of these gets decided within days, by people who are competent at their own jobs and are not search specialists, and each one is expensive to reverse.
That is the gap this fills. Not more work — a seat that is already briefed on your situation and can be asked, this week, what the search consequence of a decision is, before the decision is made rather than after it shows up in the numbers.
It suits organizations that can execute perfectly well. They have developers, they have writers, they may have an agency or two. What they do not have is anyone senior enough in search to say that one is fine, this one will cost you eighteen months, and they are not going to hire one, because the volume of that judgment does not fill a job.
This is not execution, and the distinction is the whole point
Nothing is implemented inside this engagement. No changes are made to your site, no content is written, no advertising account is touched, no links are pursued. It is counsel, and it is priced and scoped as counsel.
That boundary is not modesty, it is what makes the advice worth having. An advisor who also delivers has a stake in what gets recommended, and the recommendation drifts toward the delivery whether anyone intends it or not. Keeping the two apart means I can tell you an engagement is not necessary, that the incumbent agency is doing good work, or that the correct answer this quarter is to spend nothing on search and fix the checkout.
It also means the engagement does not grow of its own accord. If a piece of work needs doing, it gets specified and handed to whoever executes — your team, your agency, or a separately scoped project. What does not happen is a standing advisory arrangement quietly becoming a delivery retainer over six months, which is the usual life cycle of these arrangements elsewhere.
The four things this actually gets used for
Across the situations where this arrangement earns its keep, the work sorts into four categories.
Decisions, as they arise. The subdomain question, the migration timing, whether to consolidate, whether a rewrite needs a redirect map, whether a proposed URL structure will hold at ten times the current page count. Usually answered in a call and a short written note, and the note matters — the reasoning has to survive the person who asked it leaving.
Roadmap review. Reading the plan before it is funded, in the order it is written, and saying which items are dependent on others, which are premature, which are solving problems the site does not have, and which are missing. Cheaper than discovering the same things by executing them.
Vendor selection. Sitting in on pitches, reading proposals, and asking the questions the buying team does not know to ask. This is the highest-value hour in the whole arrangement for most clients, because the cost of the wrong vendor is a year.
Review of work done by others. An assessment of whether what is being delivered matches what was sold, and whether it is having any effect that can be demonstrated.
Reviewing someone else's work without turning it into a fight
Second opinions on an incumbent provider are delicate, and the useless version is easy to produce: a list of everything the other party did differently from how I would have done it. That helps nobody and it is mostly taste.
The version worth paying for asks a narrower set of questions. Does the work match what was contracted? Is there a dated record of what changed and when, so that any effect could in principle be attributed? Do the recommendations rest on evidence from the site, or on tool defaults applied to any site? Would anything here be considered a spam policy violation — undisclosed paid links, content generated at scale without adding value, pages published on the site purely to trade on its reputation? Is anything being reported that the underlying data cannot actually support, such as click-through from AI features, which Google does not report at all?
Where the answer is that the work is sound, that is the finding, and it gets stated as plainly as a criticism would be. A review that never clears anyone is not a review. Where the answer is that it is not sound, the output is specific enough for the provider to respond to — which is a more useful outcome for you than a recommendation to fire them.
What to ask a vendor, and whose questions to use
When this seat is used for vendor selection, the framework is not one I invented. Google publishes the questions it suggests asking a prospective search provider, and using them has the advantage that no vendor can dismiss them as a competitor's framing: ask for examples of previous work and successes, ask whether they follow Google Search Essentials, ask what results and timeframe they expect, ask about their industry experience, ask how changes and recommendations will be communicated, and check whether they are genuinely interested in understanding the business.
The same source lists the warning signs, and they are worth quoting exactly, because they are more emphatic than most buyers expect. "No one can guarantee a #1 ranking on Google." Be wary of anyone claiming a "special relationship" with Google or advertising a "priority submit." Be wary of firms that "email you out of the blue." Avoid providers who talk about "link popularity schemes" or submitting a site to thousands of search engines. "Be careful if a company is secretive or won't clearly explain what they intend to do." And, memorably: "You should never have to link to an SEO." Google's hiring guidance is short, and reading it before a pitch changes the meeting.
What I add in the room is the follow-up question. A vendor who clears the published list can still be selling a fixed monthly deliverable that has nothing to do with your actual constraint, and finding that out takes a specialist asking about scope, sequencing and who does the implementing.
What a month looks like from your side
A standing call at an agreed cadence, access between calls within an agreed response time, and a written record of anything decided. The amount of access is bounded and set at the start, because an unbounded arrangement either gets abused or gets rationed quietly, and both are worse than a stated number.
The written record is the part people underestimate. Advice given in a call evaporates; six months later nobody can remember why the subdomain was rejected, and the proposal comes back. Every decision of consequence gets a short note stating what was asked, what was advised, what it depended on, and what would change the answer. Those notes accumulate into something genuinely useful: an institutional memory for a function that usually has none.
What does not happen is a monthly report of activity. There is no activity to report — the deliverable is judgment, and padding it with a slide deck would be an admission that there was not enough of it.
How it is engaged, and what makes one seat cost more than another
It is retained monthly, rolling, with no fixed term, because counsel that is locked in for a year is counsel with less reason to be candid. The drivers of the figure are the volume of access, not the size of the company: how much bounded access is agreed, how often the standing call runs, how many providers are in play and therefore how much work is being reviewed, how many markets and platforms are in scope, and whether the seat includes attending vendor pitches, which are episodic and time-consuming.
A meaningful driver, and an unusual one: whether the organization wants the reasoning transferred. A seat that teaches an in-house marketer to make these calls without asking costs more per month and less per year, and it is the version I would choose in your position. The goal of an advisory arrangement should be to become less necessary.
When this is the wrong purchase, and how you will know it is working
It is the wrong purchase if nobody can execute. Advice delivered into an organization with no development capacity and no content capacity produces a queue of unactioned specifications and a growing sense that search does not work. Buy execution — your own, an agency's, or a fixed-scope project — and add counsel afterwards.
It is also the wrong purchase if you have one clear problem. A migration, a manual action, an audit: these are bounded pieces of work with a definable end, and buying them as a project is cheaper and finishes. Standing counsel is for a stream of decisions, not a single one.
The measures of whether the seat is earning its place are behavioral, and they are visible within a quarter. Decisions get made faster, because the specialist input arrives before the meeting rather than after the mistake. Fewer decisions get reversed. Vendor proposals get shorter and more specific, because the questions being asked have changed. The reasoning behind past calls is recoverable months later, from the notes. And in the best case, the work commissioned from other providers gets better, because someone is reading it against what was promised — which is the outcome nobody thinks to ask for and the one that tends to pay for the arrangement on its own.
Frequently Asked Questions
What does a fractional search advisor actually do?
Provides standing senior judgment without delivering the work. In practice that is four things: answering decisions as they arise, reviewing the roadmap before it is funded, sitting in on vendor selection and reading proposals, and giving second opinions on work delivered by other providers. Everything is written down, so the reasoning survives staff turnover. Nothing is implemented inside the engagement — no site changes, no content, no advertising account access — which is what keeps the advice free of an interest in what gets recommended.How is this different from an SEO retainer?
A retainer buys delivery: diagnosis, specifications, changes made or handed over, and verification of what shipped. An advisory seat buys judgment only, and explicitly excludes execution. The practical difference shows up in what you get invoiced for and what you can ask for. Under a retainer you can ask for work; under an advisory seat you can ask for a decision, a review, or an opinion on somebody else's work, and if a piece of work is needed it gets specified and handed to whoever executes.Can you review the work my current agency is doing?
Yes, and that is one of the most common uses of the seat. A useful review is narrow rather than sweeping: does the work match what was contracted, is there a dated record of changes so effects could be attributed at all, do the recommendations rest on evidence from your site or on tool defaults, is anything at odds with published spam policies, and is anything being reported that the underlying data cannot support. Where the work is sound, the review says so plainly. That happens more often than clients expect.Will you sit in on vendor pitches?
Yes, and it is usually the highest-value hour in the arrangement, because the cost of choosing the wrong provider is a year of the roadmap. The base framework is Google's own published list of questions to ask a prospective search provider, which no vendor can dismiss as competitive framing. On top of that come the specialist follow-ups about scope, sequencing, who does the implementing and how the work would be judged — the questions a buying team has no reason to know to ask.How much access is included each month?
A bounded amount, agreed at the start, along with a standing call at a set cadence and a response time for anything between calls. Bounding it is deliberate. Unlimited arrangements either get overrun or get quietly rationed by the advisor, and both are worse for the client than a stated figure everyone plans around. If the volume of decisions turns out to be consistently higher or lower than expected, the arrangement gets adjusted rather than endured.What happens if the agency is replaced? Do you take over the work?
Not usually, and it is worth being blunt about why. One practitioner has finite capacity, and taking over delivery for a client whose vendor selection I advised on is exactly the conflict this arrangement exists to avoid. The useful role is helping specify what the replacement needs to do, running the selection properly, and reviewing the new provider's first quarter against what was agreed. Where a bounded piece of work is genuinely better done by me, it gets scoped and priced as a separate project rather than absorbed.How is a fractional advisory seat priced?
As a monthly retained seat, rolling, with no fixed term. What moves the figure is the volume of judgment rather than the size of the business: how much bounded access is agreed, how often the standing call runs, how many providers are in play and therefore how much external work is being reviewed, how many markets and platforms are in scope, and whether attending vendor pitches is included. A seat that also transfers the reasoning to an in-house marketer costs more per month and tends to cost less per year.Published