It's the shift nobody planned for.
For as long as most of us can remember, search strategy meant one thing: rank for the query, win the click. Keywords. Backlinks. Ten blue links. The entire discipline of SEO was built on a simple contract: if you structured your content the right way, the search engine would send people to your website, and your website would do the persuading.
That contract is changing.
A growing share of searches now get answered before anyone clicks anything. An AI-generated synthesis, assembled from the sources the system can find, delivered on the results page or inside a chat.
The websites are still there. And the clicks haven't vanished. But for more and more content and buying journeys, the first impression of your organisation is no longer your homepage. It's what AI says about you.
Which raises a question most organisations haven't asked: what does AI say about you? And where does it get its answer?
AI systems don't invent your organisation. They synthesise it.
When someone asks about your company or your category, the system pulls from everything available: your website, your help documentation, your old blog posts, your reviews, forum threads, analyst commentary, comparison sites. It weighs all of it and assembles a response.
That response reflects your collective signal. Not your best page. Not your latest messaging. The aggregate of everything you've ever published, alongside everything others have published about you.
And that second part matters more than it first appears. A journalist's description, an analyst's summary, a customer's review — the answers you don't control often carry more weight in the synthesis than your own homepage. That's a whole problem of its own, and one I'll come back to.
Every answer your organisation has ever given is now potential source material.
There's a name emerging for optimising in this environment: answer engine optimisation. AEO.
And it's already heading toward the same mistake SEO spent two decades making.
The early advice is tactical: structure content in Q&A format, use schema markup, get cited in trusted sources, publish more.
Here's the precise claim: tactics applied to answers your organisation hasn't agreed on are wasted effort, and may even make the problem worse. Because if your organisation gives different answers in different places, structuring them nicely just makes the inconsistency easier to synthesise.
Your homepage says one thing. Your help docs say another. Your two-year-old blog post says a third. Mark it all up with perfect schema and the AI now has three well-structured, machine-readable, conflicting answers.
You've just optimised the noise.
Here's something we can't fully prove yet. The research on how AI weighs conflicting sources is thin, and the systems keep changing.
But consider what a synthesis engine has to do. It's looking for the most confident synthesis it can make. Clear, consistent, corroborated signals are easy to synthesise. They produce specific answers.
Conflicting signals produce hedged ones. Blurred ones. Or worse: an answer built on your outdated positioning, because the old blog post was clearer than the new homepage.
If that's right, the organisations that show up well in AI answers won't be the ones producing the most content. They'll be the ones whose answers agree with each other. Everywhere. Over time.
Consistency becomes something like a ranking factor. Not algorithmically. Practically ... the coherent organisation is the one the machine can describe.
The hypothesis is worth acting on early The cost of acting on it is clarity, which pays for itself even if the hypothesis is wrong.
So the real work of answer engine optimisation isn't markup. It happens before any of that.
It's deciding the answers. What does your organisation do? Who is it for? Why should someone choose it? What does it cost, and what does it stand for?
One answer per question. Agreed at the organisational level. Owned by someone. Expressed everywhere.
Then, and only then, do the tactics matter. They amplify whatever exists, which is precisely why the deciding has to come first.
AI amplifies what it finds. Make sure what it finds is decided.
Yes. For a growing share of queries, AI now synthesises an answer directly rather than sending people to a list of links. That means the first impression of your organisation is increasingly what an AI says about you, assembled from across the web, rather than your own homepage. Being found is shifting from ranking for keywords to being represented accurately in a synthesised answer.
Answer engine optimisation is the emerging practice of shaping how AI systems represent your organisation when they answer questions. Early advice focuses on tactics like Q&A formatting and schema markup. But those tactics only help if your organisation has first agreed on consistent answers. Applied to inconsistent content, they simply make the inconsistency easier for AI to find and reproduce.
AI systems don't read your best page and repeat it. They pull from everything available — your website, help docs, old blog posts, reviews, third-party coverage — and synthesise it into a single response. That response reflects the aggregate of every signal you've sent, plus what others have said about you, not the message you most want to lead with.
A synthesis engine produces its most confident answer when the signals it finds agree with each other. When your content says different things in different places, the AI has to reconcile conflicting inputs. The result is often a hedged, blurred, or outdated answer, because the machine can only be as clear about you as your content is.
Not by publishing more content. Start upstream, by deciding your core answers — what you do, who you're for, why you matter — and making every signal you control say the same thing. Consistency is what lets an AI describe you clearly. The tactics come after the decisions, not instead of them.
Ask an AI system about your organisation — what it does, who it's for, how it compares. Then ask a different one and compare. Write down what comes back and hold it against what you'd want a prospect to hear. The gap between the two is your real search strategy problem, and no amount of extra content will close it until your answers are decided.