Hey all,
I'm very pleased to announce that the new pageviews definition is (1)
complete and (2) implemented. Prominent features include:
1. A removal of the per-project double-counting due to banners;
2. The removal of meta over-over-OVER-counting due to EventLogging;
3. The inclusion of Mobile App traffic;
4. The inclusion of projects with non-standard URL schemes.
What this means in practice is that when the data begins coming out
through
stats.wikimedia.org and elsewhere, you can expect to see a
substantial drop in traffic. This is not a drop in traffic; it is a
correction for the massive inaccuracies in the existing definition,
which are causing an artificial /rise/.
So, what's next? Well, the Analytics Engineering team has to implement
the functionality on a regularly running job to get the data released
on a consistent basis. We also need to split out per-article pageviews
and do some tagging to provide granular reports - see
https://meta.wikimedia.org/wiki/Research:Page_view#Future_work . But
the core definition is complete.
Huge thanks to Andrew Otto, Christian, Nuria, Aaron and Bob West for
their contributions to this project.
--
Oliver Keyes
Research Analyst
Wikimedia Foundation