Does AI Know Your Business? The Three States and What They Mean

The thing most often confused when measuring is the difference between "it mentioned you" and "it recommended you". They're not the same, and there's a third state in between: not mentioning you at all. protan.ai reports show one of these three states for every engine. Here's what they mean.

1. Recommends

When you ask the model for "a reliable X company in İzmir", the answer includes your name, what you do and why you're good. Sometimes first, sometimes as one of three options.

This is the best state you can reach. The model recognizes you as an entity, knows what you do and matches you to the need being asked about.

The job: protect it. Models change; the engine that recommends you today may drop you in its next version. Keep your information current, ask the same questions again at regular intervals, and watch whether competitors are entering the same answer as you.

2. Recognizes

The model knows your name. Ask "What is Protan?" and it gives a correct description. But ask "Can you recommend an X company in İzmir?" and its list has other names on it.

This is the most common and most misleading state. A business owner asks about their own name, sees the correct answer and says "AI knows us". It does, yes; but the customer isn't asking for your name, they're asking about their need. If you're absent when the need is asked, being recognized alone doesn't bring customers.

The cause is usually this: the model knows you but can't match you to a specific service, area or customer type. If your site says "we do everything", or your service pages don't say which areas you serve, this state is inevitable.

The job: build the match. Open your service pages with the sentence the customer asks ("glass balcony installation in Bornova, İzmir"). Fill in the areaServed and serviceType fields in Schema.org. On your FAQ page, write real questions with real answers.

3. Doesn't mention

The model neither knows your name nor lists you when the service is asked about. Ask about your name and it either says "I couldn't find a business by that name" or confuses you with something else.

There are two sub-states here, and telling them apart matters:

• The door is closed. robots.txt blocks AI crawlers, the site is empty without JavaScript, or the pages are noindex. The model would like to learn about you but can't. This is good news; the fix is technical and quick.
• The door is open but there's nothing inside. The site is accessible but there's no clear text describing the business: a one-page site, a portfolio made of images, three sentences on the "about us" page. The model got in but found nothing to learn.

The job: check the door first, then fill the inside. You can test crawler access in a minute; llms.txt and an Organization schema can be added in a day. The real work is writing about your business as if explaining it to a stranger: who you are, what you do, where, for whom, since when.

Why three states, not a ranking

In classic SEO, being fifth is slightly worse than fourth and slightly better than sixth. Generative engines have no such gradient. If you're in the answer, the customer sees you; if you're not, they don't. That's why a protan.ai report doesn't say "what position you're in"; it tells you which state you're in for each engine and shows how many different questions produced that result.

Don't decide based on a single question. An engine can answer the same question differently twice; that's why a state is only assigned when at least three different questions give a consistent result. If you're doing the measurement yourself, apply the same rule: one question is not evidence.