Is AI Recommending Your Business? A 20-Minute Self-Audit

Most business owners have never asked an AI engine about their own category. They assume the answer is either "of course we show up" or "nobody uses that yet." Both are guesses. Twenty minutes and a notepad will replace the guess with a baseline you can measure against later.

Set Up the Test

Open four engines in separate tabs: ChatGPT, Perplexity, Gemini, and Claude. If you want a fifth, use Google's AI Mode. Log out or open a private window so your own history doesn't color the results. Then write down five questions a real prospect would ask. Not your keywords. Questions, in plain language, the way someone types at 11pm:

• "Who is a good [what you do] in [your city]?"
• "What should I look for when hiring a [your category]?"
• "How much does [your core service] cost in [your area]?"
• "Is [your business name] legit?" or "What does [your business name] do?"
• "[Your business name] vs [your closest competitor]"

Run Every Question in Every Engine

That's twenty answers. For each one, record three things: did your business get named, was the description of you accurate, and who else got named instead. The third column is the most useful thing you'll write down all week, because it's the list of businesses the engines already trust in your category.

Ask a follow-up on the ones that miss you: "What about [your business name]?" If the engine says it has never heard of you, that's a different problem from the engine knowing you but choosing someone else.

How to Read the Results

Never named, and the follow-up draws a blank: the engines can't find or can't parse you. Start with the legibility diagnostic: does your site say what you do in plain words, is your content in the HTML rather than assembled by JavaScript, and do your name, address, and phone match across the web.

Named on the direct question but not the category question: they know you exist but don't consider you an authority. This is a depth and evidence problem. You need more specific content on the topics you want to be recommended for, and more third-party sources saying the same thing about you.

Named, but described wrong: an old service line, a previous city, a description scraped from a stale directory. This is an entity consistency problem, and it's usually the fastest fix on the list, because you're correcting facts rather than earning trust.

Named accurately across most engines: you're ahead of nearly every competitor. Now the job is keeping it that way and expanding the set of questions you win.

Make It a Habit

Put the twenty answers in a spreadsheet with the date. Run the same twenty questions on the first of every month. AI answers shift as models update and as your competitors publish, and a monthly log is the only way you'll notice a slide before it costs you a quarter of leads. This is exactly what we do inside our AI Search Optimization retainers, and it's the first thing we show a new client, because nothing motivates a fix like watching a competitor get recommended by name.