More sophisticated portal analytics

[ACharbonneau]

There are a lot of non-web analytics we want to track, and I think that Carl said there was already some tracking in place. I'd like to know what we already have, and who has access to it, and how mutable it is.

For reporting, in addition to webtraffic, the kinds of things I'm looking for are:

• who is using the portal (from their profile info, and so not until after the "personalized dashboard" epic)
• downtime/uptime
• if possible, things like 'what gets searched'

[ACharbonneau]

A partially annotated list of the analytics NIH would like. Don't assume that all of it necessarily has to happen in the portal per se, I think at least some will be meta-statistics of more basic info.

Things I think we should be able to get from standard analytics

• Where did the user come from/what led them to the website?
• How many users do we have?
• From where in the country/world?
• Do we get repeat users?

Things I think we ask for in the user profile rather than a thing we try to guess

• Affiliation: what university or org are they from?
• Position: PI, grad student, postdoc, staff researcher, etc

Things we'd have to get/surmise from portal logging:

• That seem possible:

  • Number of returning users vs registered users vs one time users
  • Most common searches (so in the search columns bar, what do they type?
    • Here they are interested in using the data to prioritize features, and therefore want both successful and unsuccessful searches. So we'd want to know that people keep trying to search gene names, for example.

• That I don't have a good idea how it would physically work:

  • What do they do with the results? Are people saving searches? Are they pushing manifests to tools? Which tools?
  • Which datasets are popular? Which DCCs are popular?