Monitoring: engagement

[eljhkrr]

% of users actively using the CHT platform represented by form submissions

[derickl]

How's this progressing @eljhkrr? It's also not tracked in sprint. Is that intentional?

[mrjones-plip]

I'm curious, when I look at the monitoring API there's both couchdb.users.doc_count and connected_users.count . Using those can you calculate engagement?

If not, can you quantify how you plan to calculate engagement with forms submitted? Something like "if user submitted a report in past 7 days, they're active"?

[eljhkrr]

• doc_count is useful in the context of replication and for now we can only get this information from the client perspective i.e. how many docs are on the device which can be a good indicator for purging rules implementation. It can be a rough indicator of engagement but would need to be supported by other data points.
• Connected user count would need to be shown against the total count of expected active users but isn't actionable.
• We discussed this today and resolved to have a table with username and last sync time to make the data actionable i.e. specific users can be located and investigated why they're not using CHT which can inform interventions to get them back online.

[mrjones-plip]

how many docs are on the device which can be a good indicator for purging rules implementation [SNIP] Connected user count would need to be shown against the total count of expected active users but isn't actionable.

I think you're thinking of couchdb.**medic**.doc_count which would indeed be tied to docs on devices. I'm referencing couchdb.**users**.doc_count which will be a count of users on the system (but may include things like deleted users so might be off the older a deployment gets).

There's clearly more than one way to do this, and I don't know which is best, but am trying to see if there's ways of doing things with the cheaper monitoring API calls vs more expensive authenticated calls.

[eljhkrr]

Right, I'm agreeing with you that connected users is indeed an engagement indicator but from a programs perspective we need to make this information actionable. The main question being "why are the CHWs not engaged". To investigate, there's a need to identify who they are and follow up on them directly which is why a table ranking has value. Connected users would be useful for measuring impact and there needs to be a mechanism to enable intervention where necessary.

[mrjones-plip]

Right, I'm agreeing with you that connected users is indeed an engagement indicator but from a programs perspective we need to make this information actionable.

Gotcha - thanks!

[kennsippell]

@freddieptf This is the last P1 metric for App Monitoring. Are you planning to work on this in April? If not - do you mind if I take it?

[eljhkrr]

This data point is returning 0's for almost all projects. Reopening to consider options to make it more usable.

[eljhkrr]

Gather related metrics and present as different widgets for interpretation by dashboard users