Your customers are asking AI who to hire. What decides the answer
Search is no longer only a list of links. A growing share of buyers ask an assistant for a recommendation and act on it. Here is what influences that answer.
The way people look for a supplier is changing, and it is changing faster than most marketing budgets are.
Someone who needs a commercial cleaner, a fractional CFO or a roofing contractor used to search, scan a page of links, click three, and compare. A growing number of them now ask an assistant instead. “Who are the best commercial cleaners in Tampa, and what should I expect to pay?” And instead of a list of ten links, they get a short answer naming two or three companies.
That is not a search result. That is a referral. And if your name is not in it, you were never considered.
Why this is different to normal search
With traditional search you are competing for attention on a page where you are visible even at position six. The buyer sees you, even briefly.
With an assistant’s answer there is no page. There are two or three names and a paragraph of reasoning. There is no position six. You are either in the answer or you do not exist for that question.
That makes the outcome much more binary than ranking ever was, and it is why this is worth paying attention to now rather than in two years.
What seems to influence the answer
An honest caveat up front: these systems are not transparent, they change constantly, and nobody outside the companies that build them knows the full picture. Anyone selling you a guaranteed method is selling you something. What follows is what the evidence currently points to, and it will shift.
Being described consistently across the web. Assistants assemble an answer from many sources. A business described the same way on its own site, its directory listings, its social profiles and anywhere it is mentioned is easier to summarise confidently than one described differently in each place. Inconsistency reads as uncertainty.
Being mentioned somewhere that is not your own website. Your own site says you are good. That carries limited weight. Industry publications, local press, podcasts, directories, review sites and forums carry more, because they are independent. This is closer to public relations than to traditional search optimisation.
Answering the actual question somewhere. A lot of buying questions are specific: what does this cost, how long does it take, what goes wrong, what should I ask a supplier. A page that answers one of those questions plainly and completely is a useful thing for an assistant to draw on. A page of marketing copy about your commitment to excellence is not.
Being structured and readable. Clear headings, plain sentences, direct answers near the top. The same things that make a page easy for a person to skim make it easy for a machine to extract.
Reviews and reputation. Where an assistant can see that a business is well regarded, that appears to matter, which is the same reason it always mattered.
What this does not mean
It does not mean abandoning search engine optimisation. The same signals largely overlap, and traditional search is still where most volume lives.
It does not mean pumping out AI written articles. Thin content produced at scale is exactly what these systems are getting better at discounting, and it damages the consistency you are trying to build.
And it does not mean this replaces advertising. Nobody is bidding for placement in an assistant’s recommendation yet in the way they bid on search ads, and when that becomes possible it will cost money like everything else.
The practical version
If you want to start somewhere useful, start here.
Write down the ten questions a buyer asks before they hire someone in your category. The real ones, including the awkward ones about price and what goes wrong.
Answer each one properly, in plain language, on a page of your own. Not a sales page. An answer.
Make sure your business is described identically everywhere it appears, and fix the places it is not.
Get mentioned somewhere independent. One genuine article, interview or listing is worth more than ten pages on your own site.
Then ask an assistant the question your customers would ask, and see what comes back. That is your baseline, and it costs nothing to check.