How to Track Your Brand's Mentions in ChatGPT

Why is ChatGPT visibility hard to track manually?

ChatGPT answers are non-deterministic — ask the same question twice and the sources can change. A single manual check is noise. Reliable tracking means asking many relevant questions and sampling each more than once, then averaging the result.

There's a second trap: the answer depends on the exact wording of the prompt. 'Best CRM for startups' and 'CRM tools for small teams' can surface entirely different sources. Tracking one phrasing tells you almost nothing about the dozens of ways real buyers actually ask. You need a spread of realistic questions, not one.

And ChatGPT only cites sources when web search is active. If you test against the model's training knowledge alone, you're measuring memorization, not citation — a different thing that you can't influence with on-site changes.

Mentions vs. citations — which are you tracking?

These are two different signals and it's worth being precise. A citation is your domain appearing as a linked source in the answer. A mention is your brand named in the text with no link. Absent is neither.

Citations matter most: they drive referral clicks and signal that the engine trusts your page as a source. Mentions still have value — they build brand familiarity and often precede citations — but a page full of mentions with zero citations means the engine knows you exist yet doesn't send readers to you. Track both, separately.

What should you actually track?

• ▪Citation rate: how often you're cited vs mentioned vs absent.
• ▪Share of voice: which competitors get named instead, and how often.
• ▪Per-query and per-intent breakdowns (commercial, comparison, problem, brand).
• ▪Trend over time as content and engines change.
• ▪Which of your pages get cited — so you can make more like them.
• ▪Whether a fix worked: re-probe the exact question after each change.

How do you track brand mentions in ChatGPT, step by step?

• ▪List 15–30 real buyer questions across intents — not just your brand name.
• ▪Ask each in ChatGPT with web search enabled, since that's what cites sources.
• ▪Record the outcome per question: cited, mentioned, or absent, plus who else was cited.
• ▪Sample each question 2–3 times and average — single runs are noisy.
• ▪Repeat weekly and after every content or technical fix to see the trend.

Should you track it manually or use a tool?

Manual tracking works for a handful of questions once, but it collapses at scale. Thirty questions, sampled three times, weekly, across even one engine is ninety checks a week to run and log by hand — and ChatGPT's variance means occasional checks mislead you.

AEO tools automate the probing, sampling, and reporting so the trend builds itself. If you're comparing options, see the best AEO tools and agency-focused tracking tools. The mechanics of why an answer cites one source over another are covered in how AI engines decide what to cite.

Free way to start tracking today

WebCited generates your buyer questions, probes live answer engines, and reports your citation rate and the competitors cited instead — a free baseline before you adopt a paid monitor. If ChatGPT is your priority engine, the ChatGPT mention tracker focuses the scan there.

Once you have a baseline and see a competitor cited where you aren't, the next step is fixing it — start with why your competitor is cited and you're not.