Open KastanDay opened 4 months ago
New metric: Admins vs non-admins.
User
or Admin
?User
or Admin
?Average number of "turns" per conversation. Example 4 turn conversation:
New Metrics:
Active admin vs active users
Bounce rate = (Single-page visits / Total visits) × 100%
Cumulative graphs for everything
Accuracy of response, user satisfaction --- add satisfaction
Admin side of uiuc chat – ask for access
Usage patterns= what are the peak hours or peak times
Customer segmentation 2- admins and users
Market the product ? survey of professors/ student who already use it
⭐️ == WORK FOR YOU ALL 👉 == WORK FOR KASTAN
⭐️ Retention per URL? Per project, number of days since last interaction?
✅ Per user email, number of days since last interaction? -- good enough with retention
chart. Retention per user.
✅ Active user
- track every time a user logs in to a specific course/project.
e.g. Find users
who viewed course
within 7 days.
✅ bounce rate
- session duration after going directly to a page...
✅ Number of unique users per project. - unique user who has pageview on pagename...
👉 /chat feedback from user
- thumbs up / report problem
👉 Project created (Properites: Project_id, person_id)
👉 Admins updated (Properties: number of admins, list of admin person_ids, and project_id)
👉 Users updated (Properties: Number of users, list of user IDs, project_id, is_public_project)
A/B test the new landing page:
@KastanDay to do: New metric I want to track: e2e latency of time to first token. From sending a "Chat" to "first token response".