Is SEO dead in 2026?

Google's generative features are built on its core Search ranking and quality systems, and use retrieval-augmented generation and query fan-out to surface material from the Search index. A page that cannot rank cannot be surfaced either. Sound SEO is the foundation of AI visibility, not its casualty.

What has changed is the return. Questions with a short factual answer are increasingly answered in place, so the traffic that used to arrive from them does not. In exchange, a well-built page can now be one of the sources an answer is assembled from — which produces no session, no referrer and no line in your analytics.

So the investment is not wasted. The measurement of it is incomplete, and the gap between those two statements is where most of the confusion in this market lives.

Source: Optimizing your website for generative AI features on Google Search, Google Search Central.

We rank on page one and get nothing. What is that?

Usually one of two things, and they are worth telling apart before anyone spends money on either.

The first is that you rank for the wrong questions. Plenty of B2B sites rank well for terms that describe what they sell and badly for the questions a buyer actually types before he knows who to call. "Palletiser manufacturers" is a term. "What does a palletiser cost to install on an existing line" is a question, and it is the one that precedes an enquiry.

The second is that the questions you rank for are now being answered above the results. Your position is intact and fewer people travel past it. Both produce the same symptom — healthy rankings, quiet inbox — which is why the report looks fine while the quarter does not. We have set out how to tell them apart in your B2B website stopped producing enquiries.

What ranking still buys you

Eligibility, mainly, and it is not a small thing.

Being in the index is the precondition for being retrieved. A page that is not crawled, not indexed, or too slow and broken to process, is not a candidate for anything — an AI answer included. Every unglamorous technical job that has always mattered still matters, for a new reason.

You cannot be quoted from a page nobody can read.

It also still buys clicks, on the questions that end in a decision rather than a fact. Nobody chooses a capital equipment supplier from a paragraph. They start there and then they go and look.

What ranking no longer buys you

A complete picture of whether buyers are finding you.

Position measures where your page sits in a list of links. It does not measure whether your company is named in an assembled answer, and the two can move in opposite directions. You can hold position three for a term and be absent from every answer to the questions that term represents.

That second measurement has a name — AI Visibility, how often you appear at all in answers to the questions your buyers ask — and it is defined on Baseline. It is not a replacement for rank tracking. It is the reading rank tracking was never designed to take.

Which SEO work still pays

Most of the work that was always genuinely useful, and very little of what was fashionable.

Answering real questions completely and early on the page still pays, because a passage that makes sense lifted out of its page is a passage that can be quoted. Covering a subject thoroughly rather than publishing one strong page still pays — that is topical authority, defined on Build. Technical health still pays. Being described consistently everywhere still pays, and pays more than it used to.

Structured data still pays, for narrower reasons than it is usually sold on, which we have written up separately in does schema markup help you get cited by AI.

Which SEO work has stopped paying

Volume without coverage. Twenty thin posts aimed at twenty keywords never built an argument a machine could follow, and now they do not even collect the clicks that used to excuse them.

Writing to a search volume figure rather than to a question also stops paying, because the questions buyers ask an AI are longer, more specific and frequently have no volume attached to them at all. And chasing the tactics Google has publicly said do nothing for its generative features — AI-specific files, forced chunking, AI-targeted rewrites of good pages, manufactured mentions — pays nothing by definition.

How to tell which kind you are doing

Take your last six months of published pages and ask one question of each: what single question does this answer, and would the first paragraph make sense quoted on its own?

If most of them answer a question completely in their opening, you have been doing the durable kind and the transition costs you less than you fear. If most of them are built around a term rather than a question, that is the work to change — not because AI arrived, but because it was never the stronger version.

So why does everyone keep saying it is dead?

Because "SEO is dead" has been a reliable headline for fifteen years, and because a genuine shift in what the work earns is easily mistaken for the end of it.

There is also a commercial incentive. A discipline declared dead needs a replacement, and replacements are sold with new names. Google's own position is that optimising for generative AI search is optimising for the search experience, and is therefore still SEO. We agree with that, and we sell measurement rather than a new acronym.

What this article cannot tell you

It cannot tell you how much traffic you have lost, or will. Published figures measure other companies' sites in other categories, and none of them is yours.

It cannot promise the balance holds. AI systems are probabilistic and change continuously, and the guidance underneath this article was published in May 2026 and updated in July. The title carries a year; the page will be updated rather than replaced when the year turns.

And it cannot tell you whether your own SEO investment was well spent. That depends on what was built, which is knowable, and on what is currently being surfaced, which is measurable.

What to do next

Stop asking whether SEO is dead and start asking a narrower question you can actually answer: when a buyer asks an AI who to consider in your category, does your name come back?

You can test that yourself in an afternoon. If you would rather have it measured properly, across all six systems and against the competitors you name, that is what the AI Visibility Audit does — and the answer tells you whether the work you have already paid for is reaching anyone.