Category: Uncategorized

Everyone advising on search right now is learning it at the same time as everyone else. The technology under the work changed inside eighteen months, and anyone claiming settled expertise in something that new is telling you something about their sales process rather than their practice.

Robert Spinrad, who runs an agency, put the honest version on the Unscripted SEO podcast:

“I think a good term I want to use is we’re humble experts. We know what we know, but we also know we don’t know, and that’s OK. We’ll keep learning.”

That is the disposition to hire for. Here is how to check for it in a conversation that lasts thirty minutes.

Ask what they have changed their mind about

The question does more work than any credential. Somebody doing this seriously has revised a position in the last two years, because the ground moved and the evidence moved with it. They will name the thing, say what changed it, and tell you what they do differently now.

An answer that contains no revisions is the answer. Either they have not been paying attention, or they are unwilling to say so in front of a prospect, and both matter.

Ask what they would need to see to know it was working

A good advisor names a metric before the engagement starts and accepts that it might not move. A weaker one describes activity — posts published, links built, audits delivered — which is a description of effort, not of outcome.

Press one level further and ask what would make them tell you to stop spending. An advisor who cannot describe the conditions under which their own service is the wrong purchase has not thought about your business, only about their pipeline.

Ask how they treat what they cannot measure

Attribution is worse than it was. Dark traffic, assistant answers that never produce a click, branded search that was caused by something untracked three weeks earlier. The question is not whether an advisor can solve that — nobody can — but whether they will say so.

Spinrad’s related point was about the specific temptation of the moment:

“We encourage clients to be wary of people saying they are AI experts, because this whole thing is brand new. We’re all learning.”

The advisor worth hiring will mark the boundary between what they measured and what they inferred, unprompted, in the first conversation.

Ask who does the thinking

Much of the mechanical work is now cheap. Crawling, clustering, first-draft briefs, the audit itself — those can be produced quickly by anyone with a subscription. What is not cheap is the judgement about which of the findings matters for a business with your constraints, and the willingness to discard the rest.

So ask who reads the output before it reaches you, and what they removed. An audit delivered at full length, every issue flagged, nothing prioritised, is a tool’s output with a logo on it.

What none of this tells you

These questions test disposition, not competence. Somebody can answer all four well and still be wrong about your market, and a blunt operator who fails the humility test can still be the best technical mind available to you. Treat it as a filter for how an engagement will feel when something does not work — which it will — rather than as a prediction of results.

More on how that shapes an engagement: SEO advisory services and fractional SEO management.

You cannot report what you cannot measure, and you can measure less every year. So the reporting has to change shape.

There was a moment when search marketing was better than PR and better than newspapers, because we had full attribution. We knew which links we built, which rankings moved, how many people came, what they searched and whether they converted. Then Google shipped not provided. Then Search Console started showing a fraction of the query data behind the clicks it reports. Now assistants recommend brands without linking to them at all.

That is three separate hits to the same evidence chain, and the invoice still has to go out at the end of the month.

The gap is bigger than most reports admit

Pull up the pages report in Search Console, find a URL with two or three hundred clicks, then drill into the queries for that URL and add them up. The numbers do not match. They are not close.

“I’ll go on the a pages report and be like, okay, each one of these pages has two hundred, three hundred clicks, then I go drill down, to c take a look at a a query report for that URL, and I’m like, Where where’s all my clicks? Who moved my clicks?”

— Rob Bonham, on Unscripted SEO

I have a blog post ranking for a query about Search Console data being unreliable. It has around fifty clicks. I have query data for four of them. The page proves its own thesis, which is funny for about ten seconds and then becomes a reporting problem.

Rob’s answer is not a better dashboard

“I honestly think it it’s a matter of telling the right story, right? I mean, and I’m not talking about being manipulative, but using data to your benefit to show how you might impact things or how you have impacted things.”

— Rob Bonham

The distinction he draws matters. Telling the right story is not spin. It is choosing the unit of analysis where you can still see something true. Two decades of local, e-commerce, SaaS and home services work has left him with a specific version of that:

“Of course, make sure you’re paying attention to how things are performing at a URL by URL level, but also you might be getting better visibility for your branded searches.”

— Rob Bonham

Technical fixes and content work land at a page level and then show up somewhere else — brand queries, homepage traffic, conversion rate on entrances from organic. The connection is a correlation and should be labelled as one. Labelling it does not make it worthless.

Say what you did, then say what moved, then keep them apart

The reporting discipline that survives the attribution collapse looks like this:

• Name the source and the window on every number. “Search Console, 1 June to 31 August” beats “recent data” in every conversation you will ever have with a CFO.
• Separate the measured from the inferred. Clicks are measured. The revenue you attribute to a technical fix is inferred. Say the word estimate out loud.
• Report the boundary condition in the same breath. A percentage lift without its denominator is a claim, not a finding.
• Show the work, not just the outcome. Deliverables shipped on time are evidence of a functioning engagement even in a month where the graph is flat.

Rob’s framing of what an SEO is actually claiming is honest about its own limits: because we changed this here, we are seeing growth in this area, and it is reasonable to connect them because of how the site works. That sentence contains a hypothesis and admits it. Clients respect that more than a confident number they cannot verify.

The bigger risk is not under-reporting. It is over-producing.

There is a second thing happening at the same time, and Rob thinks it is what will actually cost agencies their accounts. Clients are looking at a ten thousand a month retainer and doing the maths on doing it in-house with AI. Some of them will. What they will not price in is the downside:

“it might be super enticing to you know publish that 50 pages in a month or 100 pages in a month. But think about how that will be perceived by not only you know you know LLM bots, search engine bots, but perceived by the users too.”

— Rob Bonham

And the timeline on the reckoning:

“You’re gonna work yourself into a hole. And you might be you know looking like a like a hero, you know, for maybe a month, a few months, who knows, maybe six months out. But you know, as soon as as soon as that site starts to tank, and it it and it surely will after you know a period, then then they’re gonna start looking at their their agency sideways”

— Rob Bonham

Fast results that come from volume are borrowed. We set out what a credible timeline looks like instead in how long SEO takes to work.

The line to hold

His whole argument comes down to one distinction that is worth more than any dashboard:

“in in a way that is actually adding value instead of just perceived value of, they knocked out five blog posts for me this month.”

— Rob Bonham

Perceived value is what you sell when you cannot measure the real thing. It works right up until it does not, and then it takes the account with it. Measure less, claim less, and be specific about which is which.

Quotes are verbatim from the raw Riverside transcript, trimmed for false starts and filler only. Advisory work and fractional SEO leadership at The SEO Advisory; the full conversation is on Unscripted SEO.