Data-driven Personas: Constructing Archetypal Users with Clickstreams and User Telemetry
summary
Presents a reproducible pipeline that turns telemetry into five personas by aggregating 3.5 million clicks from 2,400 users into ~39,000 clickstreams, extracting 10 workflows via hierarchical clustering, and fitting a mixed models mixture to assign users fractional persona membership. Validates clusters against product experts, surfaces unexpected cross-role behavior (BI users doing IT tasks), and leads to concrete product changes (new data-prep features and event-series analytics) that demonstrably align with evolving user cohorts.
AI-written summary · about
chapters (9)
- 0:00Introduction and project
- 0:38Context and user personas
- 4:48Telemetry data collection
- 6:21Clustering user workflows
- 9:28Modeling user personas
- 13:28Company behavior analysis
- 15:06User behavior trends
- 16:00Product improvements
- 16:58Q&A and methodology
related talks
42mData-Driven Personas
20mIterative Persona
21mRemixing Design with Data
01h 24mHow to bring your personas to life without an exorcist
19mZero-party data giving AI the power of "why"
05mA Multisensor Person-Centered Approach to Understand the Role of Daily Activities in Job ...
24mHow Spotify built an audience framework that lasts
45mLeveraging Power of Analytics for Martech