Can I just check by hand, or do I need a tool?

Start by hand. Ask the questions your real customers ask across ChatGPT, Gemini, and Claude, and see whether you get named at each stage, from the first category search to the final comparison. That's a real first read. It falls apart once you need dozens of questions, competitors scored the same way, and every buying stage covered. By hand, that's slow to repeat and easy to skew. That's when a tool earns its keep.

Which AI assistants should I check, and what about Perplexity?

Check ChatGPT, Gemini, and Claude. Those are the assistants buyers actually use for recommendations, and they disagree enough that any single one is a partial read. Adacity covers those three. It doesn't cover Perplexity yet, so if Perplexity matters in your category, check it separately by hand.

How is this different from checking my Google rank?

A rank check tells you where your page sits in a list of links. This tells you whether an assistant names your brand at all when it answers a buyer, and where it got that from. You can sit at the top of Google and never get named by ChatGPT, because the assistant is reading review sites and press coverage, not your homepage.

How often should I re-check?

Treat any check as a snapshot. AI answers move as models update and as the sources behind them change. Re-check after a positioning change, after you earn a notable piece of coverage, or roughly quarterly to catch drift. Adacity gives you a point-in-time report, not live monitoring, so plan to run it again rather than wait for alerts.

How to actually check whether AI assistants recommend your brand

The short version

Checking your AI visibility is easy to fake. It goes wrong three ways:

Counting mentions. A mention count looks like data and feels like data, and by itself tells you very little. It can't say whether AI pointed a customer to your site, or which pages it read to form the answer.
Grading your own homework. You write the questions and you score the answers. The hard part is sounding like your buyer, not like keywords or hypothetical questions that miss the countless ways people actually ask. Get the questions wrong and every number after them is fiction.
Asking one assistant. They frequently disagree. We ran Quince, and Gemini named it about five times as often as ChatGPT did, same brand, same category. Look at either alone and you'd draw the opposite conclusion.

What does "is AI citing my brand" actually mean?

"Is AI citing my brand" isn't one question. It's a series of questions, one at each stage of the marketing funnel you already run:

Does AI mention you while someone's still figuring out the category?
Does it shortlist you once they're comparing names?
Does it pick you when they're ready to buy?

You can win the top of that funnel and lose the bottom, and a single yes-or-no hides which.

Why can't you just ask ChatGPT and count?

When you write the questions yourself and check only those, you bias what you find. You're grading your own homework. It's easy to type in search keywords, or the questions you assume customers ask, but that's an unreliable way to catch how people actually search. Real buyers don't talk like your marketing. They ask "a good cashmere sweater that won't pill," or "something like Everlane but cheaper," in words you'd never guess. Ask the wrong questions and the count is meaningless: you're measuring visibility for searches nobody runs. Finding that real customer language is the whole job.

What does an AI visibility (AEO) report you can trust actually include?

The questions your real customers ask, run across ChatGPT, Gemini, and Claude, and scored at every stage from first search to final purchase.

Start with real buyer questions, not the handful you'd think to type. A report you can trust builds them from your customer personas and every way those personas phrase the search. When we ran the questions buyers actually ask in Quince's category (the full read is here), it got named in 3 of 78, and sat seventh out of 150 brands at 7.3%. Seventh out of 150 is nothing to celebrate, but it's a real position you can plan around, unlike the flattering self-check.

Then check more than one assistant, because they rarely agree. Gemini named Quince far more often than ChatGPT did, roughly five mentions to one, same brand and same category. Check only ChatGPT and you'd think Quince barely shows up; check only Gemini and you'd think it's everywhere. Both are true, but only together do they mean anything.

Buyers move through stages before they buy. They search the category, then products like yours, then you by name, then how you compare head to head. Each stage gets its own questions, so you see where AI stops recommending you, not just whether it mentions you at all.

Free tools, cheap tools, and real checks

Price won't tell you if a check is any good. Free tools aren't automatically junk, and expensive ones aren't automatically worth it. What matters is what the tool measures, and whether that's what you need. Ask yourself, your team, and your provider:

Where do the questions come from: real customers, or you grading your own homework?
How many assistants does it check: all of them, or just ChatGPT?
How much of the buying journey does it cover: every stage, or just whether you got mentioned?

Skip any of those and the price is the least of your problems.

Adacity gives you options to meet you where you are. The free check builds real questions from your own site and sizes you up against your competitors. The Deep report widens the net, pulling the questions real buyers ask from across the whole internet, not just your site, for a truer read of your market. Deep and Broad runs that same wide set across ChatGPT, Gemini, and Claude, so no single assistant decides your score.

See where you actually stand

You can run the honest version yourself this week. Ask the questions your real customers ask, put them to all three assistants, and follow each stage from the first category search to the final head-to-head. You'll learn more than most AEO tools on the market give you.

Or skip a week of work and let Adacity run it for you. Free to start, $39 for the full run across all three assistants, no account, no sales call, in your inbox within the hour. See how it works, or check your brand.