As a way of understanding the behavior of readership of our users, the page view metric has been limiting to what it allows us to understand. Instead, we want to have the capability of measuring and viewing how long users are interacting with our products. This will be provided through:
- Making instrumentation available that tracks the session length of users
- Enabling data consumers to query and view session length across different dimensions
- Given there will be a unique identifier to associate a user's journey through using our applications, there are opportunities to use this as a unifying metric to piece together different actions and steps users take in achieving a goal we want to understand better.
- e.g. As we want to better understand what types of users complete an edit, a session ID can be the link between what a user does first and last before making a complete edit.
- As a data consumer, I want the ability to query the aggregation of how long users are active in a given period of time (session lengths).
- As a Product Manager and Data Consumer, I want Session Length to be representative for web activity across all devices, on all public wikis, across all features [x Final Instrument is deployed to all wikis
- As a stakeholder who does not run queries, I want access to reports and summaries (visualizations) about SessionLength in superset
- Superset queries run within 60 seconds
- Data QA checks must pass for both raw and aggregated data
Out of Scope
- Querying SessionLength for specific features
- SessionLength specific to iOS and Android native apps
- Providing a unique ID for a given users' session
- SessionLength Data Cube available in Turnilo
- SessionTick instrument
- Visualization Dashboard
- v0.0.1 Static Report
- v0.0.2 Data Cube
- Q: What aggregations are possible in using session length?
- Q: What are the differences between Session ID and Session Length?
- A: Session length metric is defined by the aggregations of many users' session lengths. Session ID is a unique identifier which could be used to identify a specific users' session. Due to privacy reasons, Session Length metric was designed to not allow for unique identifiers like session IDs to be associated to users' session lengths.
- Q: Who will be maintaining sessionTick?
- A: The PDE Team will be owning the maintenance of sessionTick instrument as it is a "platform" instrument that covers the broad spectrum of client teams.