A tool to monitor how often your brand shows up in AI search answers: what we use and why
What to look for in a tool that monitors how often your brand shows up in AI search answers, and the setup we use for client programs, from prompts to crawler logs.
Clients ask for "a tool that tells us how often we show up in AI answers". The request is reasonable, but the first version of the answer is usually a dashboard number nobody trusts. Here is how we set monitoring up so the number holds up in a client review, and which tool we run it on.
Start with the prompt panel, not the tool
A monitoring tool is only as good as the questions you give it. We build the panel with the client before we look at any dashboard. Four groups cover most of it:
- Category questions: "best [category] for [persona]".
- Comparison questions: "[brand] vs [competitor]".
- Alternatives questions: "alternatives to [competitor]".
- Brand questions: "is [brand] good for [use case]".
We tag each prompt by topic and intent, and we set the country and language that match where the client sells. Personas help when different buyers ask differently, and persona stacking in Promptwatch lets us test that without duplicating the whole panel.
We keep the panel stable. Changing prompts mid-quarter ruins trend lines, so additions are logged as a separate cohort.
What "how often" should mean
A raw count of mentions is a weak metric. We report visibility as the share of tracked responses in which the brand appears, per engine and per topic, next to the same figure for two or three named competitors. That is Share of Voice. We also report Position in Answer and Sentiment, because being mentioned last with a lukewarm description is a different outcome from being named first.
Why we run clients on Promptwatch
We run client programs on Promptwatch. The reasoning is practical. Monitoring answers "how often", but the client's next question is always "why, and what do we do". Promptwatch keeps those answers in one place:
- Daily prompt panels per model, country and language, with Visibility Score, Share of Voice, Sentiment and Position in Answer. There are dashboards with templates, a heatmap that compares two dimensions such as brands against models, and A/B period comparison for before-and-after checks.
- Citation analytics. When a competitor is named and the client is not, we look at which pages and domains the engine cited, including Reddit, YouTube and LinkedIn.
- Agent Analytics. IP-verified AI crawler logs with error tracking. If visibility drops, we check whether a bot started getting errors before we blame content.
- Visitor analytics. AI-referred visits and conversions, so the client sees what the visibility is worth.
- Action Board. A prioritized list, with a Discovery agent that investigates the data weekly and suggests actions.
We also use scheduled PDF reports, so the client gets the same view every week without us rebuilding it.
Plans, briefly
Promptwatch has a free Explore plan with 10 prompts on ChatGPT only, which is enough to try the idea. Paid brand plans start at Essential ($95 a month) and include all active models. Crawler logs start on Professional ($245 a month), so if a client needs bot data, we budget for that tier. Agencies have separate self-serve plans (Kick-off $199, Growth $399, Scale $799) with unlimited projects and unlimited prompts. We mention prices here because they affect which tier a client lands on, and they are what Promptwatch publishes.
How we read the output
We look at trends over several weeks, never a single day. We separate engines, because ChatGPT and Google AI Overviews often behave differently for the same brand. We compare against competitors so a flat line is read in context: if everyone rises, the topic is growing, and if only the competitor rises, something specific happened.
Mistakes we see
- Tracking only branded prompts. They make a brand look visible when category prompts show the real gap.
- Changing the panel too often.
- Reading one engine as "AI search".
- Ignoring crawler access.
- Treating a mention as success without checking sentiment and position.
What we tell clients
Monitoring does not guarantee a recommendation. It gives you an honest, repeatable picture and a list of causes to test. Some fixes will move the number quickly and others will do nothing, and the data tells you which is which. If you want one tool to monitor how often your brand shows up in AI search answers and also tell you what to do about it, that is the reason we picked Promptwatch for client work.