Why your SEO doesn’t ship, and what to do about it

There's a piece doing the rounds today from Bill Hunt about why SEO work doesn't get implemented. The framing is sharp: a senior SEO submits 1,400 tickets over 18 months, almost none of them ship, the team gets laid off because traffic kept declining, and from the business's perspective nothing was being done. The CTO in his story walks him to a whiteboard, points at a faint dotted line, and explains that everything below it doesn't get built this fiscal year. No debate. No negotiation.

The piece is good. It's also describing the single largest failure mode in this industry, and almost nobody writes about it because it isn't a tactic and it doesn't sell a tool.

I want to take Hunt's argument seriously, then push it further. Because the implementation gap isn't a comms problem you can fix with better ticket-writing. It's a structural problem with how SEO is organisationally positioned, and the AI search transition is making it materially worse — not better, as some people are claiming.

The framing most SEOs use — "engineering won't prioritise us" — is the symptom. The cause is that SEO sits in a category of work that no executive has been trained to underwrite.

The line exists in every company I've ever worked with

After 18 years of this, the pattern is depressingly consistent. The clients who get SEO work implemented at pace share two traits: a senior internal sponsor who treats organic search as a P&L line, and a development resource that doesn't have to be re-justified every sprint. The clients who don't have either of those things spend money on audits that sit in Notion forever.

You will lose that fight every time.

I've watched £40,000 audits go to die. Not because the recommendations were wrong — most were correct, prioritised properly, written with implementation notes — but because there was nobody internally whose job it was to convert the document into shipped code. The deck got presented, leadership nodded, the marketing manager who commissioned it bookmarked the file, and engineering kept building the product roadmap they were already building.

Hunt is right that reframing helps. "AI readiness" gets prioritised when "SEO improvements" doesn't. I've used the same trick — calling Core Web Vitals work "performance optimisation for conversion" suddenly gets it through the door when "page speed for SEO" doesn't. The relabelling works because it routes the request through a different stakeholder with a different budget line.

But relabelling is a workaround for a structural problem, not a solution to it.

The structural problem is that most companies have never decided who owns organic search performance at an executive level. Sales has a VP. Product has a VP. Engineering has a CTO. Marketing has a CMO who, in most mid-market businesses, came up through brand, demand gen, or paid acquisition. SEO sits underneath them as a tactic, not as a function. So when the SEO consultant or in-house lead submits a ticket, they're not asking for resource — they're asking the marketing team to convince engineering to deprioritise something the CEO already cares about, in service of a metric the CEO doesn't understand the mechanics of.

You will lose that fight every time.

The four shapes of SEO inside a business

In my experience, SEO sits in one of four organisational shapes, and the shape determines what you can ship more than any other factor.

The first shape is SEO as a tactic owned by content. The content lead owns it. They write briefs, commission articles, and submit technical tickets to engineering through whichever process marketing uses. Implementation rate on technical work: roughly nothing. They have no authority and engineering treats their tickets as background noise. This is where most mid-market businesses live.

The second shape is SEO as a function owned by an SEO lead reporting into marketing. Slightly better. There's a named owner, often someone with real expertise. But they still have no authority over engineering's roadmap, and their boss the CMO is usually paid on demand gen metrics that don't include organic. Implementation rate: maybe 20% of submitted work, usually content-related items because those don't require engineering.

The third shape is SEO as a cross-functional initiative with a senior sponsor. Now we're talking. There's a VP-level person — sometimes the CMO, sometimes a head of growth, occasionally the CEO directly — who has decided organic search is strategic and has been given political cover to allocate engineering resource to it. Implementation rate jumps to 50–70%. This is rare, and it's almost always the difference between an SEO programme that works and one that doesn't.

The fourth shape is SEO embedded in engineering. The SEO lead reports through the engineering org or has a counterpart who does. Tickets land in the same backlog as everything else and get triaged with the same logic. Implementation rate: high, but it requires a company sophisticated enough to recognise that organic discoverability is an infrastructure problem, not a marketing problem. I can count on one hand the mid-market UK businesses I've seen do this.

The honest read on most consultant engagements — including most of mine, historically — is that we are paid by people in shapes one and two to produce work that requires shape three or four to actually ship. That's the loop. And we built it.

Why AI search makes this worse, not better

The current industry narrative is that AI search urgency will finally give SEO the executive air cover it needs. Hunt's IBM anecdote nods at this — work that sat for months suddenly gets prioritised when relabelled as AI readiness. The pitch deck version of this story is everywhere: *AI search is a CEO-level concern, therefore SEO budget and engineering resource will follow*.

The AI search transition is generating executive urgency and simultaneously routing the money around the work that would matter.

I'm sceptical. Here's why.

AI search urgency does generate executive attention. It does unlock budget. But the budget is going to the wrong places. It's going to GEO tools, citation tracking dashboards, "AI content optimisation platforms," and consultants selling visibility audits. It's not going to the underlying infrastructure work — schema hygiene, server response time, log analysis, fixing 499s, building the editorial floor — that actually moves the needle.

Mike King's work on the 499 status code is the cleanest current example. He's documenting a real, measurable gating mechanism — pages that time out for AI crawlers don't get cited, full stop — and the fix is engineering work. Origin server performance, edge configuration, retry handling. None of that gets sold by a GEO platform. None of it shows up in a dashboard the CMO can show the board. So it doesn't get prioritised, even though the executive interest in AI visibility has technically created budget that could be spent on it.

The AI search transition is generating executive urgency *and* simultaneously routing the money around the work that would matter. The implementation gap isn't closing. The window-dressing layer is just getting fatter.

This is the part Hunt's piece doesn't quite reach. He's right that relabelling works at the tactical level — call it AI readiness, get it shipped. But at the strategic level, the relabelling is being weaponised against the SEO function. Vendors are selling "GEO platforms" that cost more than the underlying engineering work would, deliver less, and absorb the budget that would otherwise go to fixing the actual problem.

What good actually looks like

If you're in-house and your work isn't shipping, the answer isn't better tickets. It's a re-negotiation of the operating model. Specifically:

You need a named executive sponsor whose performance review includes an organic discovery metric. Not a vanity metric — a metric that's tied to a P&L line they're held accountable for. If nobody at VP level loses money when organic declines, organic will lose every prioritisation argument.

You need engineering resource that doesn't have to be re-justified each sprint. This usually means either embedded SEO engineers (in larger orgs), a standing percentage of engineering capacity allocated to organic search infrastructure (10–15% is the number I see working), or a clear escalation path that doesn't run through the marketing-to-engineering political layer every time.

You need to stop submitting individual tickets and start submitting initiatives with revenue projections attached. "Fix our internal linking" loses to a product launch. "A £180,000 ARR opportunity contingent on internal linking restructure, signed off by the head of growth, with engineering effort scoped at four weeks" might not. The work is the same. The wrapper isn't.

And you need to be honest about what you can ship in the operating model you're actually in, not the one you wish you were in. If you're an SEO lead in organisational shape one or two, you cannot deliver a programme that requires shape three. Either escalate to get the shape changed, scope your work to what's deliverable in your current shape, or accept that you're being paid to produce documents that won't be implemented and decide whether you're comfortable with that.

If you're a consultant or agency, this is harder. You usually can't change a client's org shape. What you can do is diagnose it before you take the engagement, scope your deliverables to match it, and refuse work that requires implementation capacity that doesn't exist. I've started turning down engagements where the buyer is in shape one and the work needed is shape three. The audit will be correct, the recommendations will be sound, and nothing will happen. That's not a win for anyone.

The honest limits

This argument has obvious holes worth naming.

Some companies do execute well from shape one or two — usually small businesses where the founder is technical, or where the marketing lead has unusual political capital. The shapes are a pattern, not a determinism.

Some SEO work genuinely doesn't need engineering. Content programmes, editorial guidance, brand-building, digital PR — these can be shipped by marketing teams without engineering bottlenecks, and the impact on AI search citations is real. If your work doesn't ship, look hard at how much of it actually requires engineering resource vs. how much could be reframed as content or editorial work that you control directly.

And reasonable people will argue that the consultant's job is to deliver the audit and the client's job is to implement it, full stop. I used to believe that. I don't anymore. If the implementation rate on your work is structurally near zero, your delivery model is broken regardless of who's technically responsible.

The actual takeaway

Hunt's piece reframes the problem from "SEO is dying" to "SEO isn't being implemented." That's a more useful frame, and it's correct. But the implication most readers will take from it — *write better tickets, relabel your work, align with executive priorities* — is treating a structural problem as a communications problem.

The work doesn't ship because the organisational design is wrong. Most SEO functions are positioned as marketing tactics requiring engineering execution, owned by people without engineering authority, sponsored by executives who aren't measured on organic outcomes. No amount of ticket-writing fixes that.

The businesses that win at organic search in the AI era are not going to be the ones with the best GEO dashboards. They're going to be the ones that decided, at executive level, that organic discoverability is infrastructure work — and resourced it accordingly. Everyone else is going to keep submitting tickets that sit below the dotted line.

That's the line Hunt's CTO drew on the whiteboard. The question isn't how to write better tickets to get above it. The question is who in your company has the authority to redraw it.