SEO in the age of AI: what still works

The phrase "SEO in the age of AI" is often read as a farewell, as if the discipline were being escorted out to make room for something new. That reading gets the situation backwards.

Very little of the actual work has been retired. What has happened is that the work now serves a second purpose as well as its original one.

The teams that understand this are quietly winning on both surfaces, while the teams chasing a brand-new "AI-era" playbook rebuild foundations they already had. The honest summary is less dramatic than the headlines and more useful: the fundamentals hold, and one thing on top of them changed.

In the age of AI, SEO's fundamentals did not stop mattering; they became the precondition for being retrievable at all. Crawlability, structure, internal linking, and authority are what let an AI system find, read, and trust your pages before it can cite them.

What changed is the goal: not only ranking a page but being the source an answer is built from, which rewards extractable answers on top of the old base. This piece is about what still works, why it works, and the single genuine addition the AI answer layer requires.

The SEO fundamentals became the entry ticket

Everything that made a page rank in classic search still has to be true, and now it is load-bearing for a second reason.

A crawlable, fast page with its content present in the server-rendered HTML was always the base requirement for being indexed; it is now also the base requirement for being read by the crawlers that build AI search indexes and the fetchers that pull pages live for assistants.

A page these systems cannot reach or cannot read is not merely ranked lower in the answer layer. It is absent from it, because you cannot be a source for an answer that never saw your content.

The same is true of the less technical fundamentals. Clear site structure and internal linking help an AI system understand what your pages are and how they relate, just as they helped classic crawlers.

Genuine authority, earned through real recognition rather than manufactured signals, is what makes a system willing to trust and draw on your page rather than a competitor's. An answer that stakes a claim on a source is more conservative about which sources it trusts than a ranking that just lists them.

None of this is new work. It is the same work, now doing double duty, which is why the correct response to AI search is to reinforce the fundamentals rather than to treat them as legacy, a point argued in full in is SEO dead.

The one thing that genuinely changed

If the fundamentals hold, what actually moved is the target.

In classic search the goal was to rank a page: to occupy a position in the list for a query. In the age of AI there is an additional, often more important goal: to be a source the answer is assembled from. Those are related but not identical, and the difference is where the new work lives.

A page can rank respectably and still be passed over by the AI answer, because ranking gets your page considered and something else determines whether the answer actually draws on it.

That something else is extraction. An AI system building an answer reaches for passages it can lift cleanly: a direct response stated near the top of a section, self-contained enough to make sense out of context, under a heading that says what the section answers.

A page that buries its answer across several paragraphs the reader has to assemble is a poor extraction candidate even when it ranks well, because the system reaching for a clean answer finds a cleaner one elsewhere.

This is the genuine addition the AI era makes: on top of being rankable, your answers have to be liftable. It is much of what people mean by generative engine optimisation, laid out step by step in a GEO checklist.

Why SEO and GEO are one job, not two

Because the fundamentals feed both surfaces and the extraction layer sits on top of them, treating SEO and generative engine optimisation as separate programmes is a mistake that wastes money and creates turf wars.

They share a foundation. The crawlability, structure, and authority that classic SEO builds are exactly what make a page eligible to be an AI source; the extraction and distinctiveness that AI answers reward are a layer added to that same foundation, not a parallel structure built elsewhere.

Run them as two teams with two budgets and you get duplicated foundational work, contradictory priorities, and an argument about which one owns the page.

Run them as one job on one foundation and the economics are far better: the hard, slow work of technical health and authority is done once and serves both surfaces, and the extraction-and-originality layer is added on top for the answer surface specifically.

This is also why the "do I need a separate AI agency" question usually resolves toward integration, a question worth thinking through in GEO vs SEO. Keep one foundation, add the answer-layer work on top, and measure both what you rank for and what you are cited in.

What to stop, and what to start

If the fundamentals hold and the target moved, the adjustments are specific.

Stop treating a ranking as proof of visibility, because on many queries an AI answer intercepts the user before your ranked page is reached; measure citation and mentions in AI answers alongside classic rankings. Stop publishing generic content at volume, which the answer layer routes around, and redirect that effort into fewer, more distinctive pieces. Stop rebuilding foundations under a new name.

Start making your answers extractable as a default habit, so that whichever sub-question an answer is built from, you have a clean passage for it. Start treating being a source, rather than holding a position, as the goal for the queries that matter most. And start measuring the answer surface deliberately, because the whole point of the age of AI is that a second surface now decides much of your visibility.

The takeaway

SEO in the age of AI is mostly the SEO you already know, now doing double duty. The fundamentals did not become obsolete; they became the precondition for being retrievable by AI systems as well as classic search, which makes them matter more rather than less.

The single genuine change is the target: alongside ranking a page, you are now trying to be the source an answer is assembled from, which adds an extraction-and-originality layer on top of the old base.

Keep one foundation, add the answer-layer work on top, and measure both surfaces. The base holds; the target moved.

If you want to know whether your existing SEO foundation is actually feeding the AI answer layer or quietly failing to, that is part of what an AI search visibility audit checks.