Explore personas and behavior | Brief Documentation

User Intelligence has three views: Personas, Research Signals, and Behavior. Personas describe the people you build for; Behavior shows groups based on product usage. Use them together to check whether a customer need is reflected in what people do. Start in Personas, then switch views using the tabs at the top of the page.

You can create a persona yourself without connecting a source. To have Brief find personas in customer material, connect a readable source first. Behavior groups require usable product usage data from PostHog or Mixpanel. Langfuse, Arize, and PostHog LLM can supply metrics for the separate AI Chat Engagement card; connecting an AI chat source alone does not create behavior groups. See Connecting your tools for the connection flow.

Find and read a persona

Open User Intelligence → Personas. The list shows each Persona, its Priority, the number of Pain Points, linked Signals and Features, and when it was Updated. A Suggested badge marks a persona waiting for review.
Use Search personas... to search the name, description, needs, and pain points. Select Filters → Priority to show High, Medium, or Low priority personas. You can also sort by a table heading. If you see No personas match your filters, change the search or priority filter.
Select a persona to open its details. Read its description, Priority, Needs, Pain Points, and Quotes where available. A quote opens its original source when that source has a usable link; some quotes are text only.
Follow the related Features, Research Signals, or Decisions shown below the persona to see the connected work and evidence. A persona may have no links in a section.

For the individual customer observations behind a persona, see Research Signals.

Create or change a persona

In Personas, select New Persona. Enter a required Segment Name, an optional Description, and a Priority of High, Medium, or Low. Medium is selected initially. Select Create. The form shows Segment name is required if you leave the name blank.
To revise a saved persona, open it and select Edit. Change the same fields, then select Save. You can also open the row's actions menu and select Edit.
To remove one, open it and select Delete Persona, or choose Delete from the row's actions menu. Confirm the prompt before it is removed.

The create and edit form changes the name, description, and priority. Needs, pain points, and quotes appear in the detail view when the persona has them; those fields are not in this form.

Review personas Brief suggests

If the list is empty, select Have Brief Find Personas to look for personas in connected material. Brief first attempts to refresh research signals, then synthesizes personas. If the signal step fails, the persona step can still continue. Stop stops waiting for the scan; the work may finish in the background. A completed scan can add Suggested rows to the persona list. It does not guarantee a suggestion when the available material is insufficient.

Open a Suggested row. Read its description, needs, and pain points. Suggestions do not yet have the saved persona's linked records.
If needed, change Priority on the suggestion page. Select Chat with Brief there if you want to discuss it in the workspace chat.
Select Accept Persona to save it with the selected priority, or Reject to dismiss it. The row's actions menu also offers accept and reject; it offers a chat action when workspace chat is available.

Explore behavior groups

Open User Intelligence → Behavior. The table lists each User Segment, its Users, % of Total, Engagement, and Updated date. Select a row to inspect that group. The detail page shows its description, size, usage measures, Data source, Last updated, and Confidence when available. It may also link to related personas and features.

For an active group, review Usage frequency, Events/day, and Avg last active when shown. What makes them different compares feature use with the average user. Opportunity lists features the group has not adopted as much, and Feature adoption shows use within the group. If the group is too small for its own feature rates, that section says the percentages are for all users instead.
For a dormant group, review Avg last active, Historical activity, and Who went dormant when a breakdown is available. Last features used before leaving and What active users do that they didn't appear when the source provides those comparisons.

Sections that need a particular measure or comparison appear only when that data exists. If Usage frequency says Not enough data or the table says Engagement data pending, check the source and analyze usage again after it has more data.

Read AI chat engagement

The AI Chat Engagement card appears above the behavior table. When metrics are available, it summarizes users, return rate, satisfaction, and messages per session. Select the card to open details: sessions for the displayed period, unique users, repeat and one-time users, feedback counts, and, when measured, trends, user distribution, top AI uses, error rate, average latency, and total messages. The Source section identifies the provider, project, and source context when available. Satisfaction shows a dash when there are no thumbs-up or thumbs-down responses. Error rate and latency say not measured when the provider did not supply them.

If the card says Unavailable, read its explanation before retrying. It distinguishes an unconfigured source, a failed source run, missing identified users or events, and metrics Brief could not read. A Source error message gives additional detail when present. If a Langfuse result set reached its record limit, the card says the metrics are approximate and trends are unavailable.

Refresh or troubleshoot missing results

If there are no behavior groups and the empty state offers Connect a Source, no AI chat source is configured. Select it to open integrations. Once a product usage source has data, return to Behavior and select Have Brief Analyze Usage. This rebuilds the usage groups. Stop stops waiting; the analysis may still finish in the background.
If the table says No behavior segments yet, an AI chat source is configured, but no behavior groups are available for this window. Read the AI Chat Engagement card above the table for the reason. When source data is available, select Have Brief Analyze Usage again.
If personas or behavior seem stale, select Sync at the top of User Intelligence. This refreshes sources and runs the research signal, behavior, and persona analyses. Brief reports whether the run completed or some sources did not refresh. Stop stops waiting for that run; work may continue in the background.
If a saved persona or behavior link says it cannot be found, return to its list. The record may have been removed or may be inaccessible to your account.

For connection problems, see Troubleshooting connections. For the research evidence these views can link to, see Research Signals.