Reviews and citations shape what AI says about you.

To describe a local business, an AI assistant draws on reviews and third-party mentions: what customers wrote about you, and where else on the web your company is named. Your own website can say you’re good at heat pump installs. An assistant asked who homeowners recommend needs someone other than you to have said it.

Why assistants lean on what others say

A recommendation needs evidence, and for a local contractor the available evidence is mostly reviews and independent mentions. When an assistant searches the web to answer a question, the pages it finds for a company are often review platforms and directories. As far as anyone outside these companies can observe, the short descriptions assistants give of a business frequently read like a summary of its reviews. A company with thin or stale reviews gives them little to summarize.

What makes a review useful

“Great service” states nothing, to a person or to software. A review saying a crew replaced a heat pump in Encinitas, explained the thermostat and came back to adjust it states a service, a place and how the job went.

You can’t script that, and shouldn’t try. The way to get it, as part of review management, is to ask at the right moment, soon after the job, with an open question such as what the work was and how it went.

Why one platform isn’t enough

Reviews in one place leave you thin everywhere else. Google reviews come first because they also affect the map pack. An assistant that searches the wider web may be reading other sites entirely, so I look at which ones show up as sources for your trade before suggesting where to ask. Each platform has its own rules. Some don’t want businesses soliciting reviews at all, and I follow those rules.

Two meanings of the word citation

In local SEO, a citation is a listing of your business details on a directory. In AI answers, a citation is a source the assistant points to. They overlap: a directory listing can become the page an assistant cites. That’s why accurate listings matter to GEO (Generative Engine Optimization), along with the mentions that are harder to get, such as a local news story, a manufacturer’s dealer page, a trade association directory or a youth league you sponsor. Those come from real involvement in the community, which is also how honest local link building works.

What I won’t do

I won’t write, buy or arrange fake reviews, and I won’t set up review gating, where unhappy customers are steered away from public sites and happy ones are sent to them. Both break Google’s review policies, and fake reviews are also against federal consumer protection rules. Every customer gets the same request, whatever their experience. Critical reviews get a calm reply.

How I measure it

I start by recording which review sites and third-party pages the assistants cite when asked your customers’ questions, and what they say about your company and your competitors. That shows which platforms deserve effort. Then I change one thing, usually how reviews are requested, run the same prompts again about two weeks later, and report what happened either way. The assistants are never guaranteed to respond, and reviews accumulate slowly, so expect this to take time.