The short answer is yes
Traditional SEO still matters, and for a reason that is easy to miss: AI answers are made from the web. An assistant that searches before replying depends on pages it can find and read. A model that answers from training learned from pages that were published, linked and discussed. In both cases the raw material is the same web that search work has always tried to improve.
The claim that search optimisation is finished usually comes from people selling its replacement. The opposite claim, that nothing has changed, usually comes from people selling the old thing. Neither holds. The foundations keep their value and some long-standing habits lose theirs.
Why the foundations carry over
Consider what an assistant needs in order to use your page. The page has to be reachable by a crawler, quick to load, readable without running complicated scripts and clear about its subject. Those are the requirements of technical SEO, unchanged.
Next, the page has to be found for the search the assistant performs on the buyer's behalf. Those searches resemble the ones people type, though they are often longer and more specific. Pages that rank well for them are the ones most likely to be read and cited. A ranking is now an input to the answer as well as a result in its own right.
Finally, the system needs grounds for relying on you. The signals search engines use to judge credibility (references from respected sites, consistent business information, evidence of real expertise) are close relatives of the ones described in how AI assistants decide which businesses to name.
- Crawlable, fast pages that state their subject clearly.
- Rankings for the specific searches made on a buyer's behalf.
- Credibility earned from other sites and from customers.
Buyers still use search results
People have not stopped using search results, and they are unlikely to for tasks where they want to choose for themselves: comparing suppliers, checking a price, finding a specific company, reading reviews, looking at a map. Many journeys now mix the two. A buyer asks an assistant for a shortlist, then searches each name to see whether it holds up.
That second step is classic search. If the assistant names you and a search for your name returns a thin site, an empty profile and no reviews, the recommendation is wasted. Branded search, local results and the pages that answer due-diligence questions matter more as a result, because they are where a buyer prompted by an AI answer arrives.
There is also a plain matter of evidence. Search data is the richest record most businesses have of what their buyers want, in their own words. Giving up the discipline of reading it would leave a company guessing at the questions that assistants are now being asked.
What changes
Three things change. The first is the goal. A click was once the only prize. Now a buyer can read your answer, or your name, without visiting. Being the source of the answer and being named in it are outcomes in their own right. They need content written to be quoted and a reputation that exists beyond your own domain.
The second is measurement. Positions and clicks describe only part of what is happening. Visibility in AI answers has to be sampled with repeated prompts and reported as a range, and some of its effect will arrive as direct visits and branded searches with no obvious origin.
The third is what content is worth making. Pages that restate common knowledge to catch informational searches lose value when an answer can be generated on the spot. Pages with first-hand detail, a clear position, real prices or a specific method gain value, because they supply what a generated summary cannot invent.
Where the old habits stop paying
Some practices were already weak and are now a poor use of money. The list below is where we would look first when reviewing an existing programme.
What these have in common is that they aimed at the mechanics of a ranking system and ignored the reader. Systems that read and summarise text are harder to satisfy with form alone.
Stopping them frees budget. The money that paid for the fortieth variation of a page is better spent on one piece of first-hand research, a properly maintained set of business profiles or the time of a specialist who can answer a journalist's question well.
- Publishing many near-identical pages to cover keyword variations.
- Long introductions that delay the answer.
- Reports that treat ranking positions as the outcome.
- Link building that produces links with no real mention or context.
How we would allocate effort
For most businesses the sensible plan is one programme with shared foundations. Keep the technical base sound. Keep building pages that answer commercial questions thoroughly, and structure them so they can be quoted. Add the off-site work that makes your company a known, consistently described entity. Measure classic results and AI visibility side by side.
That is what we mean by AI SEO: an extension of SEO services to cover the new places where buyers get answers. The vocabulary is set out in what AEO, GEO and AI SEO each mean.
The split between the parts depends on your category. Where buyers already ask assistants for recommendations, shift more towards reputation and quotable content. Where they still search, compare and call, the classic work remains the larger share. Check which applies to you before moving a budget on the strength of a headline.
The position, stated plainly
The lasting purpose of search optimisation is to be the answer a buyer finds when they look for help. The interface for looking is changing. The work of deserving to be found is recognisably the same, with a wider definition of where being found happens.
A business that abandons its search foundations to chase AI visibility will find it has removed the thing AI visibility is built on. A business that ignores AI answers will keep its rankings and gradually lose the buyers who no longer look at them.
Our view is that the two should be planned, funded and reported together. Splitting them into rival budgets invites each side to claim the other is obsolete, and the buyer, who moves between an assistant and a results page without noticing the boundary, is served by neither.
