SEO is dead. Write this on the tin, not on a blog post.

The position, stated plainly.

Search engine optimisation, as a discipline, is over. The mechanic it relied on — that a human types a query, scans a list of links, picks one, clicks it, and reads — has been replaced for most commercial queries by a different mechanic: a human asks a model a question, reads the answer, and either acts on the named business or asks a follow-up.

The replacement mechanic does not reward keyword density. It does not reward backlinks the way Google did. It does not reward the twenty other things SEO agencies have spent fifteen years optimising. It rewards being the page the model trusts enough to quote. The two skill sets overlap, but the centre of gravity is in a different place.

We say "SEO is dead" not as a marketing claim but as a working diagnosis. The agency you're paying for SEO is still doing the work. The work is just no longer the work that pays the bill.

The honest part of the eulogy.

SEO is not dead in the sense that nobody types into Google any more. Plenty of people still do. SEO is dead in the sense that the marginal customer — the next client through your door — has moved. The customer who is in pain and needs an emergency dentist in Bristol now opens ChatGPT first. The customer who wants to understand whether a prenup makes sense in Manchester asks Claude before they ask a friend. The buyer's behaviour shifted in 2024-2025; the discipline that optimised for the old behaviour hasn't caught up.

If you have a strong SEO position today, that position is still worth something. The decay curve is slow. But the new position — being cited in the AI answer — is what determines whether you keep getting customers in 2027.

What replaced it.

Generative-engine optimisation. The same word "optimisation", different output. The unit of value is the cited paragraph, not the click. The discipline is editorial rather than technical. The page that wins is the page a person would quote, not the page a crawler can index.

The five differences that matter, in plain English:

— The model rewards being quotable. The Google bot rewarded being indexable. Different optimisations.

— The model rewards specifics — numbers, places, dates. Google rewarded keyword density. Different writing.

— The model rewards short paragraphs with a single answer. Google rewarded long pages with many headings. Different structure.

— The model rewards bylines and "last updated" stamps that actually change. Google rewarded fresh content cosmetically. Different honesty.

— The model rewards original specifics that nobody else has written. Google rewarded matching what was already ranking. Different competitive logic.

What to do with your SEO contractor.

We get asked this on every onboarding call. Three options, in ascending order of how much you want to keep them.

If they're an agency: keep them on the SEO work for the residual Google traffic, ask them honestly whether they have any plan for the AI-citation surface, and if the answer is vague — most of them are — start budgeting for the replacement.

If they're a freelancer: they can probably learn the new work faster than an agency. Send them the methodology piece on this site and the case study on the Manchester family-law firm. There's nothing in GEO that an editorially-strong SEO contractor can't pick up in a month.

If they're you: stop. The work you've been doing — the keyword research, the backlink outreach, the technical audits — is not the work that pays the next year. The work that pays is editorial: write three citation-grade pieces a month, in your own voice, on the questions your customers ask. That's it.

The phrase that matters.

There is a single sentence we say to every customer who's still spending on traditional SEO: "If your competitor's name comes up before yours when somebody asks ChatGPT about your trade in your city, no amount of rank-one keywords on Google will save you."

That's the diagnosis. The treatment is the editorial pipeline this site sells. The reason to start now, rather than in a quarter, is that the compounding window is open and it won't be forever. The cohort that wins the next twelve months stays won.

• Eli Marsden, Editor.