The answer
Survey respondents may not represent your whole user base. Compare reported satisfaction with observed behavior, and keep the coverage and limitations of each source visible.
Flowlines does not label inferred behavior as a survey response. It shows the direct signals it can support: negative feedback, the share of analyzed sessions without issues, repeat users, activation depth, average execution time, and cost per chat.
The population map plots activity against issue-free sessions. The action queue highlights explainable groups such as new users, power users to learn from, highest-spend users, and users with the slowest experiences.
Which signals can stand in for satisfaction?
- Whether the task was completed, not only whether the agent replied.
- Repeat contact about the same issue within a few days.
- Rephrasing the same request several times in one session.
- Requests to talk to a person, and handoffs to support.
- Explicit thanks or complaints in the user's own words.
- Whether the user comes back within a week. Check these signals against a small set of survey answers before relying on them.
Published · Updated