If you maintain one source this month, make it your Google Business Profile. When AI assistants answer local questions — who does this near me, is that place open Saturdays, what’s their number — the profile is the single highest-leverage source they read: it’s structured, it’s maintained on a platform machines already trust, and the searches assistants run on your customers’ behalf surface it constantly. It is not the only source that matters. It is the one where an hour of upkeep changes the most answers.
Here’s why it carries that weight, which fields actually do the work, and the two failure modes — bad reviews data and duplicate profiles — that quietly poison it.
Why one profile carries so much weight
Three properties stack up, and no other single source has all three.
It’s structured. A Business Profile is a form, not prose. Hours go in the hours field, services in the services field, categories in the category field. Machines read labeled facts with far less guesswork than they read paragraphs — the same reason schema markup exists. Your profile is effectively a schema block Google maintains the format of, so every fact you put in it arrives machine-readable by construction.
Google trusts it. The profile feeds Google’s own local results and Maps — and Google’s AI Overviews compose local answers on top of that same local data. For the assistant surface most of your customers see first, your profile isn’t a source; it’s upstream of the answer layer itself.
Retrieval keeps landing on it. The other assistants — ChatGPT, Perplexity, Claude, Meta AI — answer local questions by running web searches and reading what comes back. Search for any local business and profile-derived results are hard to miss. You don’t have to take that on faith: run the check yourself and read the citation trails on the answers. Profile and maps-style sources show up over and over, market after market.
Put together: the profile is where your facts are most legible, most trusted, and most retrieved. That’s the leverage.
The fields that do the work
Treat this as a checklist. None of it is exotic — the point is completeness and exactness, because an empty field is a question the assistant answers from somebody else’s source, or not at all.
- Name — your exact legal-and-signage name, nothing appended. Stuffing keywords into the name field (“Example Tint Co. | Best Ceramic Tint Sacramento”) violates Google’s rules and, worse for our purposes, creates a name that matches no other source about you — a consistency problem you’re paying for voluntarily.
- Categories — the primary category tells machines what you are; secondary categories cover what else you do. Wrong or vague categories mean the assistant may never shortlist you for the question you most want to win.
- Hours — including holiday hours and temporary changes. “Is anywhere open right now?” is exactly the kind of question assistants field, and they answer it from this field.
- Services and attributes — list what you actually offer, and set the attributes that apply (mobile service, walk-ins, accessibility, payment types). When a customer asks a qualified question — “who does mobile tint?” — a source has to state that fact about you or the assistant won’t assume it.
- Phone, address, website — matching every other source character for character. The profile is one voice in the choir; it has to sing the same notes as your site and your listings.
- Photos and description — real photos of real work, and a description in plain language. This is material an assistant can say something with — a bare profile gives it a name and nothing to recommend.
Google’s own documentation for all of it lives at support.google.com/business.
Your reviews are sources too
Here’s the part owners consistently underestimate: reviews aren’t just a star number. They’re retrieved text. When an assistant reads your profile, the review text comes along — and facts stated in reviews can surface in answers exactly like facts stated anywhere else.
Usually that helps you: reviewers naming specific services, specific staff, specific results give the assistant concrete, corroborating material. But it cuts both ways, and we’ve published a documented example of the other edge: in a recorded test for a healthcare client, an assistant listed a practitioner who does not exist — and the phantom name traced back to a single patient review that had misspelled the real practitioner’s name. One wrong source, one wrong answer. The full case study is on our results page.
The practical discipline: read your reviews for factual content, not just sentiment. Respond to correct the record where a review states something wrong — your response is retrieved text too. And make the true version of every fact easy to corroborate, so one stray misspelling isn’t the only source a lightweight model has on the subject.
The failure mode that outranks all the others
Everything above assumes one thing: that there is exactly one profile. When there are two — an old address that never got closed, a listing a third-party service created, a forgotten profile from a rebrand — every signal above splits. Reviews divide across the pair, facts drift apart, and a machine reading both has two versions of your business to reconcile and no way to know which is real.
We’ve seen this from the inside, documented month by month, and it’s the subject of its own article: how duplicate listings poison AI answers. If you do nothing else after reading this page, search Google Maps for your business name and your old addresses, and make sure exactly one of you exists.
Keep it current — a rhythm, not a project
A profile decays quietly: hours change, a service gets added, a phone system migrates, holiday weeks come and go. Nothing announces that the profile is now wrong — assistants just start repeating the stale version, and nobody sends you a notice when they do. So put upkeep on a rhythm:
- When anything changes — hours, phone, services, address — the profile updates the same day, along with your website and its schema block.
- Monthly, five minutes — skim new reviews for factual errors worth correcting on the record; add a recent photo.
- Quarterly, fifteen minutes — full pass over every field; set upcoming holiday hours; re-search for duplicates.
That’s the whole discipline. It isn’t clever, which is the point — the profile rewards accuracy and completeness, not tricks.
If you want to see what assistants currently say about your business — and whether your profile is what they’re citing when they say it — the free check shows you exactly that: what three assistants answer, which sources they cited, and the first gap we’d fix. Keeping those sources right, month after month, with the before-and-after receipts to prove it moved, is the service itself.