Tools that track your company's visibility in ChatGPT and Perplexity: what we use

"Can you tell me if we show up in ChatGPT and Perplexity?" is the first question in most of our client kickoffs. The honest answer is yes, but not by typing a few queries yourself. Generated answers shift between runs, and the wording of a prompt changes the result. You need a system that asks the same set of questions daily and records the outcome.

What the two engines reward

We have seen the two behave differently in practice, so we track them separately.

ChatGPT with search runs live searches behind the scenes and cites pages. Whether a page is fetchable matters, and the bots involved are GPTBot, OAI-SearchBot and ChatGPT-User. Perplexity leans heavily on cited sources, with PerplexityBot and Perplexity-User doing the fetching. In both cases a mention without a citation and a citation without a mention are different outcomes, and we report them separately.

Setting up the tracking

Build the prompt set from buyer language: the questions sales hears, the comparisons prospects ask for, and a few that name the company. Tag by topic and intent. If the audience varies by country or role, add those dimensions from day one. Aim for 30 to 50 prompts per brand to begin.

Then pick the metrics: share of voice versus named competitors, sentiment, position in the answer, and citation rate. A single "visibility score" is useful for a headline but not for decisions.

The tool we run it on

We use Promptwatch for client programs. It monitors ChatGPT (Search) and Perplexity in their real interfaces, plus Google AI Overviews, AI Mode, Copilot, Gemini and others, running prompt panels daily per model, country and language. The dashboards give Visibility Score, Share of Voice, Sentiment and Position in Answer.

What makes it the platform we trust for this question is what sits beside the tracking. Agent Analytics shows IP-verified crawler logs, so when Perplexity visibility drops we can check whether PerplexityBot has been failing to fetch the page. Citation analytics shows which pages and domains each engine cites. Visitor analytics shows ChatGPT and Perplexity referrals and their conversions. That closes the loop between "we were mentioned" and "it brought business."

One practical detail: the free Explore plan covers ChatGPT only. To track Perplexity you need a paid plan. Essential is $95 a month, Professional $245 (with crawler logs), Business $579. Agency plans begin at $199 a month for self-serve multi-client work.

Other options we have looked at

AthenaHQ tracks many models and adds a GEO score, with credit-based pricing. Scrunch AI covers four LLMs on its Core plan, including ChatGPT and Perplexity, with bot traffic reports. Profound is an enterprise platform. Semrush, Ahrefs and HubSpot offer tracking inside their suites. Those are all valid for particular needs. For us, the missing piece in most is crawler logs paired with conversions.

A weekly check that works

Compare share of voice for ChatGPT and Perplexity separately.
Open citations for the prompts that dropped.
Look at the matching crawler logs for errors.
Choose a fix: repair access, refresh a page, or add content for a missing question.
Re-run and compare in a few weeks.

Setting expectations

We tell clients that neither engine can be controlled. Tracking shows how often you appear and why, which lets you improve the odds. It does not buy a placement, and results will wobble from day to day.

If you want a quick start, track your top 30 questions in both engines for a month before changing anything. The baseline will tell you where to look.

What we tell clients before they start

Expect the first two weeks to be mostly baseline. The numbers will move around, and that is normal. We resist changing content during that window, because we would not know whether a change or ordinary variation moved the line. Once the baseline is steady, we pick one or two prompts where the company should clearly appear and does not, and treat those as the first test cases.