Use cases in Agent Analytics helps you understand how visitors are using your AI agent by grouping similar prompts or conversations into use cases. To find them, go to Product > Agent Analytics, select an AI agent or saved report, then select the Use cases tab. This tab includes two types:
- Emergent use cases are generated automatically and update dynamically based on recent interactions.
- Tracked use cases are persistent and allow you to monitor specific topics over time.
Use cases are particularly helpful when:
- You want to understand the most frequent types of questions your visitors are asking.
- You notice high rage prompt rates and want to investigate what’s frustrating users.
- You want to identify opportunities to improve your agent’s prompt, guardrails, or coverage.
The Use cases tab has two views: Emergent and Tracked use cases. The Emergent tab opens by default. Select the Tracked tab to view use cases you've manually created to monitor over time.
Understand emergent use cases
Emergent use cases are automatically generated by grouping prompts or conversations that share a similar meaning. These groupings update based on the filters you apply at the top of the page and change as new data comes in.
Note: Data only appears if there are at least 25 visitor-submitted prompts for the selected agent and filters. If more than 10,000 prompts or conversations are included, the data may be sampled for performance. A tooltip appears next to the Prompts or Conversations column if sampling is active.
The Emergent use cases table contains the following columns:
- Emergent use case. A title and description summarizing the topic of the grouped prompts or conversations. If Leo is turned on for your subscription, you can select the Ask Leo icon next to the use case name to investigate the use case further. The Ask Leo option becomes available when a use case has one or more conversations or prompts.
- Conversations or Prompts. The number of matched interactions, depending on how your agent is configured to collect data.
- Visitors. The number of unique visitors who submitted those interactions.
- Accounts. The number of unique accounts associated with those visitors during the interactions.
- Retention. The percentage of returning visitors who submitted prompts or conversations for this use case in at least two different weeks within the selected date range. Calculated as returning visitors divided by all visitors who matched the use case during the same period.
- Rage prompt rate. The percentage of interactions that include frustration signals, such as uppercase text, profanity, or repeated attempts. This metric is more commonly available for conversations.
Adjust the level of detail for emergent use cases
You can control how broadly or narrowly emergent use cases are grouped using the Level of detail slider. This allows you to tailor clustering to match your investigative intent, from high-level themes to highly specific, actionable clusters.
All users can access this control at the top of the table.
Use the slider to choose one of the following levels of detail:
- Broad. Groups conversations by general themes or goals. Best for identifying high-level trends.
- General. Organizes conversations into high-level categories. Best for understanding user needs across key areas.
- Balanced. Shows clear, distinct use cases. Best for identifying patterns without too much detail.
- Focused. Highlights specific questions or tasks. Useful for refining known use cases or identifying gaps.
- Granular. Breaks conversations into very detailed use cases. Ideal for deep analysis or reviewing edge cases.
The emergent use cases refresh automatically based on your selection.
When you save or update an Agent Analytics report, your selected level of detail is preserved with the report configuration.
Track a use case
Tracking a use case lets you monitor how a specific topic performs over time. While emergent use cases change based on filters and recent activity, tracked use cases persist as named topics in your agent’s analytics, making it easier to measure ongoing impact.
You can create a tracked use case if you want to:
- Measure the long-term performance of an important use case or user intent.
- Monitor the effects of changes to your prompt, agent behavior, or product experience.
- Share consistent insights with other teams using saved reports and dashboards.
You have three ways to create a tracked use case:
- From the Emergent tab, select + Track use case at the far-right of any row.
- From the Tracked tab, select + Create tracked use case.
- From the full conversation or prompt side panel, hover over a visitor prompt and select the Add to use case icon.
Then:
- Enter a name for the use case. If you started from an emergent use case, the field is prefilled and editable.
- Describe the requests this use case covers. Include specific keywords, related topics, and common variations people might use. More specific descriptions improve semantic matching.
- Add up to 10 example prompts that represent the kinds of requests you want this use case to include:
- Select + Add example.
- Choose Enter manually to type a prompt, then select Add. Repeat for each prompt you want to manually add.
- Choose Select prompts to pick from existing prompts. Select the checkboxes next to each prompt you want to add, then select Add selected prompts.
- To remove all examples and start over, select Clear all examples.
- Select Create use case.
After creation, the tracked use case shows as Processing in the Tracked use cases table. Processing starts at the beginning of every hour and can take up to 15 minutes.
Example prompts only apply to new data after you add them. They don’t change how past prompts are grouped.
View and manage tracked use cases
Tracked use cases are always listed in the Tracked use cases table, but the data shown for each use case reflects your current filters at the top of the page. These use cases are useful for measuring long-term agent performance on a specific topic or initiative.
You can:
- Favorite a use case using the star icon in the first column. You can then filter by favorites using the star filter at the top of the table.
- Filter by favorites using the filter above the table, or filter by creator using the dropdown in the table header.
- Edit or delete a use case you created (or any use case if you're a subscription admin).
- Use tracked use cases as filters in the Overview, Use cases, and Conversations tabs in Agent Analytics.
The Tracked use cases table contains the following columns:
- Star icon. Indicates favorited use cases. Use the filter above the table to view only favorites.
- Tracked use case. The title and description entered at the time of creation to summarize the topic of the grouped prompts or conversations. If Leo is turned on for your subscription, you can select the Ask Leo icon next to the use case name to investigate the use case further. The Ask Leo option becomes available when a use case has one or more conversations or prompts.
- Conversations or Prompts. The number of matched interactions, depending on how your agent is configured to collect data.
- Visitors. The number of unique visitors who submitted those interactions.
- Accounts. The number of unique accounts associated with those visitors during the interactions.
- Retention. The percentage of returning visitors who submitted prompts or conversations for this use case in at least two different weeks within the selected date range. Calculated as returning visitors divided by all visitors who matched the use case during the same period.
- Rage prompt rate. The percentage of interactions that include frustration signals, such as uppercase text, profanity, or repeated attempts. This metric is more commonly available for conversations.
- Created by. The Pendo user who created the tracked use case.
- Created time. The date and time the tracked use case was initially created.
- Last updated by. The Pendo user who most recently updated the name or description of the tracked use case.
- Last updated time. The date and time that the name or description were most recently updated for the tracked use case.
Interact with the use case table
You can investigate each emergent and tracked use case by selecting any of the following in a row:
- Ask Leo icon. If Leo is turned on for your subscription, select the Ask Leo icon next to the use case name to launch a guided prompt and ask questions about the use case.
- Conversations or Prompts. Opens a side panel with a list of matching interactions. You can select any quantity to open the full conversation or prompt, view rage prompt labels (if applicable), and watch the related replay if you have Session Replay activated.
- Visitors or Accounts. Opens a side panel listing the related IDs. You can search, create a segment, download a CSV, or select an ID to view its details page.
- Rage prompt rate. Opens a side panel showing all rage prompts within the matched conversations, including a count per conversation and the reason. For conversations with rage prompts, the preview shows how many occurred. When you open the conversation, you're directed to the first rage prompt.
If you have Leo turned on in your subscription settings, you can select the Ask Leo button next to the use case name to ask questions and identify patterns.
You can also use the icons in the top-right corner of the table to add the table to a dashboard and adjust the table columns.
Customize your table view
You can adjust either table layout with the following options:
- Manage columns. Select the Manage columns icon in the top-right corner of the table to show or hide specific columns.
- Resize and sort. Hover between column headers to resize columns, or select a header to sort the table.
- View more data. Use the pagination controls below the table to move between pages. You can adjust the Items per page (10, 20, 50, or 100), see your current position (for example, 1 to 10 of 48), and use navigation arrows to go to the first, previous, next, or last page.
If you’re a subscription admin or assigned the AI Agent Admin role, select any of the prompt quantities to open a side panel that lists the associated prompts, visitors, or accounts. If you have access to these prompts, you can also search by keyword.
Note: Topic-level metrics might not add up to the total prompt count for the agent. The table includes only the most common use cases that Pendo can label and describe based on prompt similarity. These values are intended to show directional trends, not exact totals. To explore specific use cases, use the search bar to run your own topic query.
Save and share a report
If you're viewing use cases for a specific segment, timeframe, or question, you can save the view as a report:
- In the top-right corner of the page, select Save as a report.
- Name your report and add a description.
- Choose Everyone under Visibility if you want others to access or edit it.
- Select Save report.
Saved reports appear in the Reports list under Product > Agent Analytics and can be added to a dashboard.