AI prompt tracking: see which chats suggest your app
AI prompt tracking in gaintage runs the prompts your buyers type into ChatGPT and Claude and records which app or connector the assistant suggests, how often, and in what position. We call these chat keywords. Answers change between runs, so each prompt runs many times and you see a rate, not a single result.
A chat keyword is the full prompt that triggers an app suggestion in ChatGPT or Claude.
A chat keyword is a full prompt a person types into an assistant, such as "summarize my 3pm meeting". When the assistant suggests an app for that prompt, the prompt is a chat keyword for that app. A directory keyword is shorter, like "meeting notes", and only matters in directory search.
See your suggestion rate, topic breakdown and which competitor wins when you don't. Each prompt runs 10 to 20 times so the rate reflects real behavior.
Suggestion rate and position
How often the assistant suggests your app for each prompt, and where it sits in the ChatGPT plugin picker.
Topics and intents
Prompts grouped by the job behind them, like "meetings" or "CRM updates".
Competitor suggested instead
Which app the assistant picked when it did not pick yours.
Set up prompt tracking and see your first rates in three steps. You can add your own prompts or use the ones gaintage suggests.
- 01
Pick or import prompts. We suggest prompts from your listing and your category.
- 02
We run each prompt 10 to 20 times in each assistant.
- 03
Read the rates by topic, and watch how they change after you edit your listing.
gaintage measures how often assistants suggest your app inside the chat where users can install it.
Most LLM visibility tools track how often a brand is named in AI answers about websites. gaintage's AI prompt tracking measures whether the assistant suggests your app inside the chat, where a user can install it. Measure AI visibility for apps across ChatGPT, Claude, Muse and Meta AI.
ChatGPT picks plugins based on your tool metadata and the conversation context. Your listing text feeds this choice directly.
ChatGPT looks at the conversation, at your tool metadata (names, descriptions and parameter docs), at brand mentions and at whether the user already linked your plugin. Since September 29, 2026, ChatGPT can search its plugins during a chat and show a picker of plugins that could help, with an install button. Your listing text feeds this choice, so a low rate on a topic often points to the listing.
Low rate on a topic? Fix the listing that feeds it: name, description, starter prompts.
Frequently asked questions.
How many times do you run each prompt?
10 to 20 times in each assistant. Answers change from run to run, so one run proves nothing.
Which assistants do you track?
ChatGPT, Claude, Muse and Meta AI.
What is the difference between a chat keyword and a keyword?
A keyword is a short term people search in a directory. A chat keyword is a full prompt in a chat that leads the assistant to suggest an app.
Can I add my own prompts?
Yes. Add them by hand or import a list, next to the prompts we suggest.
Which plan includes chat keywords?
Growth and Enterprise. Get chat keywords on Growth.
Check one prompt now with the free AI visibility checker.
Related features. Use them with prompt tracking.
- App ranking tracker
Daily position for each keyword in ChatGPT, Claude and MCP registries.
- App category rankings
Your rank inside Productivity, Developer Tools and every other category.
- App competitor analysis
Competitor rankings, share of voice in chat, and listing changes over time.
Back to all features.