Getting your business mentioned by AI assistants

Ask an assistant for a good electrician in your suburb and it will name three. Nobody has explained how it chose them, and the honest position is that nobody outside the companies building these things fully knows.

What can be observed is where the answer comes from, because most assistants show their sources. Read enough of those citations and a fairly clear picture forms.

What they are actually doing

An assistant answering a local question is not consulting a secret ranking of businesses. It is running searches, reading what comes back, and writing a summary.

That has a consequence worth sitting with: it is summarising the web's existing opinion of you. If the web has no opinion, you are not in the answer, and no amount of optimising your own website changes that, because your own website is the one source an assistant discounts as self-interested.

You cannot describe yourself into an assistant's recommendation. You can only be described.

The three things that show up in the citations

1. Being mentioned somewhere that is not yours

Directory entries, local news, association member lists, supplier pages, community pages, forum threads. Unglamorous, and it is the single strongest pattern in the citations we read.

This is the same work as building citations and earning local links, which is a slightly boring answer to a question everyone hopes has a new answer. It is nevertheless the answer.

2. Reviews, read as text rather than counted

An assistant summarising your reviews is reading them, not averaging them. So a hundred reviews saying "great service" contribute almost nothing to a summary, while fifteen that describe what happened, what was fixed, and how long it took give it something to say.

This changes how it is worth asking. A request that prompts specifics produces reviews that are more useful to a machine and more persuasive to a human at the same time.

3. Facts that are stated plainly and agree everywhere

Hours, service area, what you do and do not do, whether you are licensed, whether you cover emergencies. If these are consistent across every source, a summary can state them confidently. If they contradict, a summary hedges or leaves you out in favour of a business it can describe cleanly.

The part that is genuinely new

One thing here does not map onto ordinary SEO, and it is worth doing deliberately.

Assistants answer questions rather than matching keywords, and the questions people ask them are longer and more conditional than the ones they type into a search box. "Who can fix a leaking shower in a second floor apartment where the body corporate needs to approve the work" is a real question and nobody types it into Google.

Pages that answer those conditional questions directly, in the customer's words, are being cited. Not keyword pages. Answer pages. The practical version of this is to write down the ten questions you are asked on the phone every week and answer each one properly somewhere on your site.

What to do this month

1. 01Ask three assistants for a business like yours in your area. Write down who they name and which sources they cite. That is your competitive set and your target list in one exercise.
2. 02Get onto the sources they cited. If a local directory or an association page keeps appearing, that page is doing more for its members than most link building does.
3. 03Change your review request so it prompts a specific answer. "What did we fix and how did it go" produces text worth summarising.
4. 04Make your facts agree everywhere: hours, service area, what you do not do.
5. 05Write answers to the ten questions you actually get asked, in the words people ask them.

We wrote more broadly about what changed and what did not in AI search and local SEO. The unglamorous groundwork underneath all of this is still citations and off-page work.