How to Tell If AI Engines Are Citing Your Brand

Most founders are still asking "are we ranking?" That was the right question when buyers typed a query, scanned ten blue links, and clicked. It's the wrong question now. A growing share of your buyers never see the links. They ask ChatGPT or Gemini to compare the options, name the best vendor, or summarize who does what — and they act on the answer. You can rank first on Google and be completely absent from that answer, and nothing in your analytics will warn you.

The question that matters now is narrower and more useful: when a buyer asks an AI engine about your category, does it name you? If so, how does it describe you, and what did it read to decide? Those three facts are measurable. You just have to look in the right place, because the place isn't your dashboard — it's the answer box itself.

Why can't you see AI citations in Google Search Console?

Search Console, your analytics, and most SEO tools were built for a world of links and clicks. They tell you what you ranked for and who clicked through. But when an AI engine composes an answer and names you inside it, often the buyer never clicks anything — they read the summary and move on. That interaction leaves little or no trace in your traffic reports. So the single most important impression you make on an AI-assisted buyer is the one your existing tools are blindest to.

There's a second problem: AI answers are not fixed. The same prompt can return a slightly different answer to two people, or to the same person a week apart, because the engines rephrase and re-rank as they go. There is no single "ranking" to look up. The only reliable way to know what the engines say about you is to ask them, repeatedly, and write down what comes back. That sounds crude. It is also exactly how you get real data instead of a hunch.

What does it actually mean to be "cited" by an AI engine?

"Cited" can mean two related things, and it's worth separating them. The first is being named in the answer text — the engine mentions your company by name as an option, an example, or the recommendation. The second is being listed as a source — the engine links the page it drew the claim from. The best outcome is both: you're named in the answer and the link points to a page you own. The common outcome is being named because of a page you don't own — a directory, a review, a podcast, someone else's roundup. Knowing which one is happening tells you where your presence is actually coming from.

The major engines expose this differently, and all of them expose something. Perplexity cites sources inline and lists them openly. ChatGPT, when it searches the web, shows the sources behind an answer. Google's AI Overviews link the pages they summarize. Gemini surfaces supporting sources as well. That visibility is the lever: because the engines show their work, you can see not just whether you were named but what they read to name you — which is the thread you pull to get named more.

How do you check whether AI engines are citing you?

The method is a short, repeatable audit. You're building a fixed set of buyer questions, running them through each engine, and scoring the results the same way every time so the numbers are comparable week to week. Here's the order that works:

1. Write down the prompts your buyers actually use. Not keywords — questions. "What's the best [your category] for [your buyer]?", "[Your company] vs the alternatives", "Who should I hire to [the job you do]?", "Is [your company] any good?", and the plain category question "What companies do [the thing you do]?" Aim for ten to twenty real prompts. If you don't know what your buyers ask, that gap is its own finding — go ask three of them.
2. Run every prompt through each major engine. Put the same prompt to ChatGPT, Gemini, Perplexity, and a normal Google search that triggers an AI Overview. Use a clean session (logged out or incognito) so your own history doesn't skew the answer toward you. Do each engine for every prompt — coverage varies a lot between them.
3. Score three things for each result. Presence: were you named, yes or no? Framing: how did it describe you — accurately, vaguely, or wrongly? Source: which page did it cite, and do you own it? Keep it to those three columns so the audit stays fast enough to repeat.
4. Note who got named instead of you. When the answer names other companies and not you, that's your real competitive set in the AI's eyes — often different from the rivals you track. Write down who shows up so you can see the pattern.
5. Put it in a simple sheet and date it. One row per prompt-and-engine, one column each for presence, framing, and source, and the date you ran it. The sheet is the whole instrument. You don't need anything fancier to start.
6. Repeat on a schedule and watch the trend. Re-run the same prompt set every two to four weeks. The single run tells you where you stand today; the repeated run tells you whether your presence is growing, which is the only number that proves the work is paying off.

The first time you do this, the result is usually sobering — founders discover they're invisible for prompts they assumed they owned, or that they're being named off a page they'd never think to optimize. That's the point. You can't fix a citation gap you can't see, and one honest afternoon with this audit shows you exactly where the gaps are.

What should you measure? The three things that matter

It's tempting to reduce all of this to a single score, but three separate signals are far more useful than one blended one, because each points to a different fix.

Presence: are you in the answer at all?

The first and most important measure is simply how often you're named across your prompt set — the AI-era version of share of voice. If a buyer asks the category question ten different ways and you show up twice, you have a presence problem, and no amount of on-page SEO fixes it. Presence is the headline number to watch over time: the share of your buyers' real questions where an engine names you.

Framing: how are you described?

Being named isn't enough if the description is wrong or weak. The engines summarize you from whatever the web says, so if your positioning is muddy online, the AI's one-sentence version of you will be muddy too — or flatly inaccurate. Reading the framing tells you whether the market's understanding of you matches the one you intend, which is a content and positioning problem you can actually act on.

Source: what did the engine read to name you?

The cited source is the most actionable signal of the three. If you're named off your own pages, keep feeding that surface. If you're named off third-party pages — a podcast, a guest post, a founder's LinkedIn, someone's "best of" list — that tells you your presence lives off-site, and the way to grow it is to widen that footprint, not to rewrite your homepage again. The source column turns a vague goal into a concrete next move.

What not to do when measuring AI citations

• Don't check once and declare victory or defeat. A single answer on a single day isn't a trend — the engines vary. One datapoint is a screenshot, not a measurement.
• Don't run the prompts while logged into your own site and tools. Your personalized history nudges the engines toward naming you. Use a clean, logged-out session so you see roughly what a stranger sees.
• Don't test vanity prompts. "Is [your company] the best?" will flatter you and teach you nothing. Test the neutral category and comparison questions a skeptical buyer actually asks.
• Don't obsess over one engine. Buyers are split across ChatGPT, Gemini, Perplexity, and Google's AI answers, and your presence can be strong in one and absent in another. Measure the spread, not your favorite.
• Don't treat the number as the goal. The citation audit is a mirror, not a scoreboard. The goal is to be genuinely worth citing — the measurement just shows you whether you're getting there.

The Citation Mirror: the one habit that makes AI visibility manageable

Here's the framework that ties it together. The Citation Mirror is the practice of asking the AI engines the exact questions your buyers ask, and reading back whether, where, and how they name you. It's a mirror because you're not guessing what the market thinks of you from the inside — you're holding up the buyer's actual view and looking at your own reflection in it.

The reason this works is that it converts an anxious, abstract worry — "are we showing up in AI?" — into a small, repeatable act with a number attached. You can't manage a presence you can't see, and until you run the Mirror, your AI presence is invisible to you precisely when it's most visible to buyers. Once you run it on a schedule, you have a dial. You can tell whether a new wave of content, a podcast run, or a founder posting consistently actually moved your presence, because you can watch the presence number climb. That feedback loop is the difference between doing AEO on faith and doing it on evidence.

The Mirror also quietly tells you what to do next. A presence gap says make more of the web mention you credibly. A framing problem says sharpen how you're described everywhere buyers and engines look. A source pattern that's all third-party says invest where that presence already lives. The measurement and the strategy are the same instrument read two ways.

The shorter version

Your analytics can't see whether AI engines cite you, and a growing share of your buyers decide inside those answers. So measure it directly: list the real questions your buyers ask, run each through ChatGPT, Gemini, Perplexity, and Google's AI Overviews from a clean session, and score three things — are you named, how are you described, and what page did the engine cite. Put it in a dated sheet, re-run it every few weeks, and watch the presence number. That's the Citation Mirror, and it turns "I hope we're showing up in AI" into something you can actually see and grow.

Frequently asked questions

How do I know if ChatGPT or Gemini is citing my company?

Ask them. Put the real questions your buyers ask — the category question, the comparison question, "who should I hire to do this" — to ChatGPT, Gemini, Perplexity, and a Google search that triggers an AI Overview, using a clean, logged-out session so your history doesn't bias the answer. Record whether each engine names you, how it describes you, and which source it cites. One pass tells you where you stand; repeating it on a schedule tells you whether your presence is growing.

Why doesn't Google Search Console show AI citations?

Search Console and most analytics tools were built to track rankings and clicks. When an AI engine names you inside an answer, the buyer often reads the summary without clicking through, so the impression leaves little or no trace in your traffic reports. The most important impression you make on an AI-assisted buyer is the one your existing tools are blindest to, which is why you have to check the answers directly.

Do AI engines actually show their sources?

Yes, and that visibility is what makes measurement possible. Perplexity cites sources inline, ChatGPT shows the sources behind a web-searched answer, Google's AI Overviews link the pages they summarize, and Gemini surfaces supporting sources too. Because the engines show what they read to compose an answer, you can see not just whether you were named but which page earned you the mention — the thread you pull to get named more often.

What's the difference between being named and being cited as a source?

Being named means the engine mentions your company in the answer text as an option, example, or recommendation. Being cited as a source means it links the specific page it drew a claim from. The best case is both — named in the answer, with the link pointing to a page you own. Often you're named because of a page you don't own, like a directory or a roundup, which tells you your presence is living off-site.

How often should I measure my AI citations?

Run the full prompt set once to establish a baseline, then repeat it every two to four weeks. The engines vary day to day, so a single check is a snapshot, not a trend. The value is in the repeated measurement: it shows whether a new content push, a run of podcasts, or consistent founder posting is actually moving your presence, rather than leaving you to guess.

What do I do once I know I'm not being cited?

Use the source column of your audit to decide. If you're absent entirely, the fix is to make more of the credible web mention you — AI engines summarize what the web says about you, not just your own pages. If you're named weakly or inaccurately, the fix is to sharpen your positioning everywhere buyers and engines look. If you're named off third-party pages, widen that off-site footprint rather than rewriting your homepage again. Measurement points straight at the next move.

Do I need a special tool to track AI citations?

No — a dated spreadsheet and a fixed list of prompts are enough to start, and starting manually is better than waiting, because it forces you to read the actual answers your buyers see. A tooling category has grown up to automate this at scale, but the discipline is the same with or without it: a consistent prompt set, scored the same way every time, tracked over weeks. The habit matters more than the software.

Measuring your AI presence is the easy half. Growing it is the work — and it's the work of being genuinely, credibly present across the web your buyers (and the engines) actually read: consistent founder content, substantive pages, and mentions you earn in the places that get cited. That's what Invisible Keyboard builds for founders — a presence worth citing, made for you — so the next time a buyer asks an AI engine about your category, your name is already in the answer.See how Invisible Keyboard works

Related reading

More on getting named in the AI answer, not just the search result: How to get your B2B content cited by ChatGPT · What is Answer Engine Optimization (AEO)? · GEO vs SEO for B2B: what changes in the AI-answer era