What AI Says About Your Company (and How to Check in 5 Minutes)

OHM Editorial Team

You know what your website says about your company; you paid for it. You roughly know what your reviews say. But there is now a third version of your company’s story, and it is being told thousands of times a month to people you never see: the version the AI assistants tell when a homeowner asks about companies like yours.

That version was assembled without your input, from everything public about your business, and it may be wrong in ways that cost you jobs. The old number from two offices ago. The service you stopped offering. A competitor described in glowing terms while you get a shrug. Most owners have never once checked, which is strange, because checking takes five minutes and costs nothing.

Key takeaways

• AI assistants such as ChatGPT, Gemini and Google’s AI mode now tell a third version of a company’s story, assembled from public data without the owner’s input, and it may be wrong in costly ways.
• The five-minute audit is asking four homeowner-style questions in two different assistants, saving the transcripts, and asking the assistant where a surprising impression comes from.
• Results fall into four buckets: described accurately, described wrongly, absent entirely, or saddled with old baggage, and each bucket comes with its own to-do list.
• Assistants draw on a website’s literal words, review platforms, directories and the Google Business Profile, and keeping that profile accurate to Google’s profile guidelines is the single highest-leverage habit.
• No ad buy, keyword stunt or fake review durably changes an assistant’s opinion, and fake reviews now carry real legal risk, so the only lever is feeding the machines better evidence and waiting.
• Run the audit quarterly on a calendar reminder, keep the transcripts to see movement, audit the two biggest competitors alongside, and ask whoever runs the marketing why AI reputation never came up.

The Five-Minute Audit

Open ChatGPT, Gemini, or the AI mode in Google. Ask the questions your customers actually ask, phrased the way a homeowner phrases them.

• “Who are the best [your trade] companies in [your city]?”
• “Tell me about [your company name]. Are they reputable?”
• “I have [a common emergency in your trade] in [your city]. Who should I call?”
• “What do reviews say about [your company name]?”

Run the same questions in two different assistants, because their pictures of you differ. Save what you get. That transcript is your company’s AI reputation as of today, and it is what a real homeowner heard this morning.

One refinement worth the extra minute: when an answer surprises you, ask the assistant where that impression comes from. The reply often points straight at the culprit, a stale directory, an old review cluster, a directory profile you forgot existed. You are essentially asking the machine to hand you its bibliography, and it usually will.

How to Read What You Find

The results sort into four buckets. Every company we have ever run this audit for landed in one of them, and each comes with its own to-do list.

• You appear, described accurately. The machines can read you. Your job is maintenance: keep the reviews flowing and the facts current, because this picture refreshes continuously.
• You appear, described wrongly. Wrong number, wrong services, wrong coverage area. The model is faithfully repeating bad data it found somewhere: an abandoned directory listing, a stale profile, an old site. The fix is hunting down the sources and correcting them.
• You do not appear at all. Your competitors get named and you do not exist. This is the expensive bucket, and it almost always traces to thin public presence: few reviews, vague website copy, missing structured data, inconsistent listings.
• You appear with old baggage. A cluster of bad reviews from a rough stretch three years ago, framed as the current story. Models weigh volume and recency, so the cure is a steady stream of fresh evidence that outvotes the past.

Where the Machines Get Their Facts

Understanding the sources makes the fixes obvious. The models draw on your website’s literal words, your review platforms, business directories, your Google Business Profile, and anywhere else your company leaves a public trace. They cross-check, and consistency is credibility: when your hours, address and services match everywhere, the machine speaks about you with confidence. When sources disagree, it hedges, or skips you for a competitor it can verify. Keeping your Business Profile accurate is the single highest-leverage habit, and Google’s own profile guidelines spell out what complete looks like.

This is why AI visibility work is mostly disciplined fundamentals rather than magic. The website rewritten to say plainly what you do and where. The structured data layer that labels it for machines. The citation cleanup across directories. The review engine that never stops. Run together, they are the core of our AI search optimization practice, and they double as classic local SEO, because the machines and the search engines drink from the same well.

What You Cannot Do

Clarity about the limits saves money. You cannot pay a model to like you; there is no ad buy that edits its opinion of your company. You cannot trick it durably with keyword stunts or fake reviews, and the fake-review route now carries real legal risk on top of platform penalties. And you cannot fix the picture overnight, because models refresh their views of the web on their own schedules.

What you can do is feed the machines better evidence and wait for them to digest it. That asymmetry favors honest companies for once: the shop with real reviews, real expertise and consistent facts cannot be outspent here by a competitor with none of those things and a bigger budget. It is the rare corner of marketing where being legitimate is the whole strategy.

While you are at it, run the same audit on your two biggest competitors. Their transcript next to yours shows exactly where the gap is, and sometimes reveals that the market leader everyone fears is invisible to the machines, which is an opening someone in your market will notice first.

Make It a Habit, Not a Scare

Run the audit quarterly, on a calendar reminder, and treat it like checking your rankings: a vital sign, not an emergency drill. Keep the transcripts so you can see movement. When a fix ships, correcting a directory, publishing better pages, hitting a review milestone, watch for it to surface in the answers over the following months. The feedback loop is slower than ads and faster than you fear.

And loop in whoever runs your marketing. If they have never mentioned your AI reputation, ask why. The share of customers who meet you first through an assistant grows every quarter; we walked through that shift in your customers are asking ChatGPT who to call, and it has only accelerated since.

Owners sometimes ask whether all this matters yet in their town, and the honest answer is that the audit itself answers the question. If the assistants already name companies in your market, the shift has arrived there, and someone is collecting those referrals today. If the answers are still vague, you are early, which is the best position anyone has ever been in at the start of a channel.

Final Thoughts

Somewhere right now, a homeowner in your service area is asking a machine about companies like yours, and the machine is answering with total confidence, accurate or not. You cannot opt out of that conversation. You can only find out what is being said and start shaping it.

Five minutes, four questions, two assistants. Do it this week. Whatever you find, you will know more about your own company’s public life than most of your competitors know about theirs, and in a shifting market, that is exactly where you want to stand.

Frequently asked questions