On Thu, Jul 17, 2014 at 7:05 PM, Maryana Pinchuk <mpinchuk(a)wikimedia.org>
wrote:
On Jul 17, 2014, at 6:33 PM, Howie Fung <hfung(a)wikimedia.org> wrote:
Thanks for putting this together -- this is really interesting data. It's
interesting that the 5+ activation rate for mobile web > desktop now --
just a few months ago, mobile was about half of desktop.
Yeah, the jump happened in May, and we still don't know why. My theory is
lastmodified, but it's a bit of a stretch for 5+.
Yeah, lastmodified is a stretch at 5+. . .I can't think of anything else.
Let's keep an eye on this.
It could be the case that mobile web editors make a smaller number of
small edits/spread a big edit out into smaller ones -- would be interesting
at some point to see whether these initial edits happen in rapid
succession.
I can play around with some visualizations for that :)
Cool -- look forward to seeing the results. A metric that comes to mind is
velocity (e.g., distributions of time-to-5th edit). Lots of ways you can
look at this.
It'll also be interesting to compare the (short-term) retention rates of
mobile vs. desktop.
The Activation rate for apps is weird. We know on desktop that about 70%
of users create an account, but never edit -- this would imply that on
mobile apps a far higher percentage of users create accounts for reasons
unrelated to editing. We should look a little closer to the data .
Something doesn't seem right.
If it's weird, it's at least consistently weird -- I saw the same thing (a
~1% 1+ conversion rate) in the early post-release data. This is coming from
Wikimetrics, which is pretty reliable. But yeah, let's get an independent
third party to verify...
In the meantime, some wild speculation: maybe people really are used to
signing up on all apps as part of first time ux without necessarily getting
anything out of it, more so than on other platforms.
If that's the case, we should be able to see if in the data, right? If
we're tagging account creations as first time ux vs. left nav, we should
see tons of account creations coming in from first time ux.
Howie
On Thu, Jul 17, 2014 at 5:40 PM, Maryana Pinchuk <mpinchuk(a)wikimedia.org>
wrote:
Greetings all, and happy 1 month anniversary of
the tablet redirect :)
I've got some data to share on new tablet, phone, and app editors and how
they stack up against new desktop editors. This is enwiki only for now; I
await the glorious return of Dario to help us out with global data :)
*New active mobile editors*
Our metric for new active mobile editors is 5+ edits within 30 days of
registration, so we don't quite have the full numbers on the first month of
newly registered users yet (because people who signed up yesterday haven't
had time to reach active status). But here's today's snapshot of the
percent of newly registered users we've converted to active editors in the
last 30 days across all out different platforms:
<image (21).png>
Note that the trend Dario observed still holds – both newly registered
tablet and mobile site users are being converted to 5+ editors at a higher
rate than desktop.
Slightly concerning is the conversion rate on the Android app – I'd have
expected this to be higher, more in line with mobile site editors on
phones. We'll need to figure out if this is due to users editing
anonymously on the app instead of signing in and counting toward our new
active user count.
*New user productivity*
I was also interested in the volume of contribution we're seeing across
each of the platforms, to get a sense of how much new folks are
contributing. So I ran the numbers for bytes changed (added, removed, and
total changed) by newly registered users across all our different platforms:
<image (20).png>
Though mobile site editors (both tablet and phone) are reaching 5+ edits
at a higher rate than desktop users, it looks like desktop editors make
significantly bigger edits in terms of bytes. Not super surprising, given
the smaller screen real-estate and greater difficulty in contributing
longform text, but it's good to validate that with real data. Note again
that Android app editors are on the lower end of bytes contributed.
To me, these two graphs raise the question of whether mobile site editors
(on both tablets & phones) make a larger number of small edits as opposed
to one big edit like they might on desktop. I don't know if we'll be able
to answer that definitively anytime soon, but it's definitely something to
think about and dig into more in the future...
--
Maryana Pinchuk
Product Manager, Wikimedia Foundation
wikimediafoundation.org
--
_______________________
Howie Fung
Director of Product Development
Wikimedia Foundation
hfung(a)wikimedia.org
--
_______________________
Howie Fung
Director of Product Development
Wikimedia Foundation
hfung(a)wikimedia.org