The old path was simple. Somebody needs a plumber, they search, they call whoever is at the top. That path still works.
It is no longer the only one. A customer can now open ChatGPT or Perplexity and ask who the best heating and air company in their town is, and get an answer with a name in it. Google does the same thing with the overviews that sit above the search results.
The businesses named in those answers are not buying the position. They are being chosen, based on what those systems can find out about them.
How the recommendation gets made
It reads what already exists about you. Your website, your Google Business Profile, your reviews, directory listings, anywhere your business is mentioned. If little exists, or what exists disagrees with itself, there is not much to work with.
It looks for signs you are real and relevant. Consistent details across the web, recent reviews that name actual services and places, a site that says plainly what you do and where, and evidence that somebody is still running the business.
It assembles an answer from that. Not the fanciest website. The business that most looks like an established local expert everywhere the system checks.
Then the customer calls. This is the part that matters: they often do not go back to a search results page afterwards. They take the recommendation. If the name in it was not yours, you did not slip down a ranking. You never entered the running.
What the signals actually are
- A Google Business Profile that is complete and active. Right categories, accurate service areas, current photos, reviews that get replies.
- Reviews that mention the work and the place. A review naming the job and the town tells a system far more than a star rating and a one-line compliment.
- A website that answers real questions. What you do, where you work, why you can be trusted. Thin pages get passed over.
- Details that match everywhere. Name, address, and phone identical across your site, your listings, and the directories. Disagreement reads as doubt.
- Content that shows you know the trade. Service pages and plain explanations do more than a homepage of adjectives.
- Signs of recent activity. A profile that has gone quiet reads as a business that may have gone quiet too.
What does not work
The instinct is to game it the way people once gamed search rankings. It does not transfer.
- Keyword stuffing. These systems read for meaning. Packing your trade and your town into every sentence does nothing.
- Bought reviews. Manufactured patterns are increasingly easy to spot, and the downside is worse than the upside.
- Being on every platform. Everywhere is not the same as obvious where it counts.
- Doing it once. Activity is itself a signal, so a profile tidied up and abandoned goes backwards.
- Paying for the slot. There is no paid placement in an AI recommendation. It is earned.
Why now
This is not a forecast. It is happening in your service area already, with your customers.
The work that earns it is the same work that has always earned local visibility, done consistently and kept current. That is the good news and the catch at once: there is no shortcut, and there is no version of this you can buy your way into.
Be obvious where it matters.