Investigate replacing usage ping with Snowplow tracking
Problem
Before we can deprecate usage ping, we need to replicate what's being tracked in usage ping in Snowplow (version check, counts, SMAU/MAU etc.).
Proposal
- Configure a version check ping using Snowplow for self-managed instances
- Replicate count tracking with Snowplow (do we still need this alongside what we're tracking with Snowplow (page pings, page views, forms and link clicks)?
- Collect data on SMAU/MAU with Snowplow
- Anything else?
Why is this important?
- We Consolidate our data collection into one robust, more up to date tool that is already well documented and maintained
- Less confusion within the org around how we collect data ourselves (i.e. not with a 3rd part tool like Pendo)
- Increased efficiency around how we can collect data and answer questions for our product stages
Considerations
- What're the problems we are currently solving with usage ping today?
- Do we still need to solve those problems?
- What value do we currently get from usage ping that we don't want to lose?
- What improvements do we want to make to this? We should consider the current problems that exist with usage ping and aim to resolve those with Snowplow.
- Should we collect this data daily rather than weekly? What performance impact would this have on self-managed instances? Should we allow them to choose?
This may make more sense as an epic but we should discuss and see what the scope is first.
Edited by Luca Kisielius