Discovery/Maps and KPIs 2015-09-14


This meeting was in response to concerns raised during the Discovery departmental retrospective, regarding what we are tracking for maps, and how that information relates to the KPIs.OK

Existing ambiguities we might need to discuss:Edit

  • Session tracking
  • Sampling rates: Mix of per-user and per-action? ** All per-action, so resolved.
  • KPIs have been cut down to just 3. One easy to track, one sessions, and one [didn't catch this]


Session trackingEdit

  • KPIs were originally written with sessions in mind. Implementing sessions was difficult, so fell back to actions.
  • Logging by actions is a reasonable starting point. It's not ideal because we can't really understand usage as well.
    • e.g. Can't differentiate between one heavy user and a lot of light users
  • Don't have to block on this.
  • What is the technical hurdle to tracking sessions?
    • We would have to do it through EL.
      • We could keep existing schema, but add a schema with sampling to track sessions.
      • Works because we would only session-track mapping events, not all events.
  • Session = continuous series of browsing. A gap of one hour empirically marks the end of a session.
    • We pick a sample of users, and track page views and maps events with a short-expiry cookie
    • Can show "proportion of sessions that include a map session"
  • What are the benefits of tracking sessions vs. just tracking users. Why do we need it?
    • "Reasons".
    • As an organization, for a variety of reasons, we standardized on session tracking, so we'll go with it.
    • prerequisites for session tracking is the same as user tracking, so could switch later
  • Whatever we pick as our initial design/implementation, it can change over time.

Blockers to announcing mapsEdit

  • T: When can we see early data from what has already been deployed?
  • O: "By end of tomorrow"
    • Dashboards are already up, showing real data. Tracking usage by platform.
    • Several loose ends need to be tidied before we can call it done
  • T: When could I send announcement?
  • O: Would suggest Tuesday afternoon, or Wednesday morning
  • T: Max and Yuri should be available for the announcement, to handle any incoming questions
  • M: Either should be fine (<--paraphrasing)

Action ItemsEdit

  • Oliver will document his proposal in phab
  • Oliver will create a phab to document pros/cons of session tracking and user tracking (and action tracking?)