I've sat through enough SEO audits to recognise the species on sight, and it almost never begins with SEO. It begins with Sales. Prospects can't find us. Somewhere back, whoever wrote the brief had nailed a KPI to it — "increase leads" — as if a website produced leads on its own, rather than the thinnest slice of a funnel that also runs through marketing, sales, product and price. Nobody wants to be the one who says the KPI was hung on the wrong hook. So instead someone says the magic sentence.
We should do a proper SEO audit.
Not out of any sudden passion for canonical tags. Because an audit produces a document, and a document is very good at exactly one thing: making responsibility blurry.
I've watched this from the client side more times than I'd like to admit. The deck lands — serious deck, serious fonts, a slide titled "Quick Wins" listing things like rewrite the page titles and fix the redirect chains. All true-ish. All quick only because the engineers who'd have to do it aren't in the room. Then the deck sits there. Nothing ships. Next quarter there's a follow-up deck about why the first one didn't move the numbers.
I've watched it from the agency side too — the same film from the opposite angle. Someone says we need to show value. Someone says we need a framework, and the deck starts to grow like a living organism. There is always a slide with a four-quadrant matrix. Nobody can tell you what the axes mean, but everyone agrees it looks responsible. And the moment anyone asks the only question that matters — who is actually going to implement this? — the room develops a sudden, sincere interest in "prioritisation alignment."
The funny part is that everyone is behaving rationally. The client wants cover. The agency wants to be retained. The project lead wants a PDF that proves the thing was done. The engineers want everyone to stop inventing new work on a Friday afternoon. Put those four incentives in a room and you don't get a scandal — you get an audit. It isn't a moral failure; it's incentive design working exactly as designed. None of this is unique to SEO — the subject changes, the reflex doesn't.
So let me say the thing the deck was built to avoid. The audit was never the hard part. The analysis that looked like wizardry from outside — turning a crawl export into a list of problems, grouping them, dressing them in the right vocabulary — was always the cheap part. The hard part was implementation: shipping the changes without breaking the templates, measuring whether anything moved, being honest when it didn't. That never fit on a slide, which is precisely why it kept not happening.
And here is where the joke turns on the deck. The whole audit economy rested on scarcity — the tools fragmented, the language gated, the output expensive-looking. Take that away and the deck becomes what it always secretly was: a formatting exercise with a margin. I'm not an SEO expert, and I no longer need to be one to get the work the expert used to sell. I hand a model the crawl export, the Search Console data and a few pages, and it comes back with rewritten titles and meta descriptions, the schema it would add, the canonicals quietly fighting each other, and a read on what to fix first — then it points at the next thing. The loop the retainer used to sell by the month, now for the price of a subscription. None of it is magic; it's fast, structured pattern work, the one thing these models are unambiguously good at.
And the audit was never bespoke to begin with — one template, your logo on slide one, your "opportunities" from the same drop-down as everyone else's. The model, for all it can't do — it can't merge the pull request, and it will hallucinate with total confidence if you let it — does the one thing the deck never did: it works for your site, and only your site.
You were always renting the expertise; that never changed. What changed is what you rent it from: no longer a profession you hired blind, sold to ten companies at once and gone the day the retainer lapsed, but a subscription that works your site and answers to no one else. Expertise didn't disappear. It stopped being a profession and became a subscription.
