Requires a schema, which requires a list of user actions that we want to track.

@phuedx --this didn't make it into this sprint because it was not yet scoped..and then slipped through the cracks!

I met with Leila about this. First ideal, then if that is not possible, the acceptable :)

1. Ideal: if a user hits one of the pages with a "tag" on it, then we somehow track the rest of their session--time or pageviews. How? Cookie, appended urls that somehow get passed to each subsequent page (I could then track through weblogs).

2. Acceptable: track clicks on tags, and track clicks on the fancy new collections

@phuedx and @KLans_WMF do you think we could push this into current sprint as was originally intended? It feels like we are going to be able to launch prototype, but without ability to measure, it is 95% unhelpful.

@phuedx for "acceptable" above, I think we would want to have the following actions:

land_on_page (if this is hard, I coooould do it from weblogs, but it would be a challenge to go through the data without sampling)
click_on_tag
click_on_article

@phuedx any sense of initial estimate for this? @JKatzWMF @phuedx I don't see any problem pulling this in to the sprint if we know how to move forward now. We may need to trade off another task to adjust our commitment.

Tracking session time, dwell time, and PVs is already being tackled by Analytics IIRC – not localised to the Browse experiment, obviously. I'd like us to defer to them for those measurements.

As for the suggested acceptable actions:

• land_on_page: how about tags_impression? We tracked the user seeing WikiGrok, why not track the user seeing tags? Especially as they are at the bottom of the page

The other two are fine.

Here's my initial design of the schema: https://meta.wikimedia.org/wiki/Schema:MobileWebBrowse

My initial estimate would be 3.

@phuedx I like that idea and the schema looks good.

Anyway, let's do this! Put T95300 into needs analysis, as I think that it can happen anytime whereas the prototype should launch asap.

Re: tracking session length analytics can be used for it, but we have no way of tagging the sessions that happened upon tags v. the sessions that did not. Regardless, that is a problem I don't think we should get into now.

I've added the provisional estimate.

@kaldari, @bmansurov: We'll need to estimate this during CATSOUP!!1 today.

@kaldari, @bmansurov: If you're happy with the estimate, then just move it through the process.

A 3 sounds good to me.

Change 209399 had a related patch set uploaded (by Bmansurov):
Implement Schema:MobileWebBrowse logging

https://gerrit.wikimedia.org/r/209399

Change 209399 merged by jenkins-bot:
Implement Schema:MobileWebBrowse logging

https://gerrit.wikimedia.org/r/209399