Wikimedia monthly activities meetings/Quarterly reviews/Mobile contributions/July 2013

The following are notes from the Quarterly Review meeting with the Wikimedia Foundation's Mobile Contributions team, July 1, 2013

Present: Tomasz Finc, Vibha Bamba, Monte Hurd, Adam Baso, Jon Robson, Juliusz Gonera, Kenan Wang, Maryana Pinchuk, Erik Möller, Howie Fung, Tilman Bayer (taking minutes), Ryan Kaldari

Participating remotely: Arthur Richards, Brion Vibber, Kul Wadhwa, Max Semenik, Yuri Astrakhan, Yuvaraj Pandian

Please keep in mind that these minutes are mostly a rough transcript of what was said at the meeting, rather than a source of authoritative information. Consider referring to the presentation slides, blog posts, press releases and other official material

Proposed Agenda:

  • Q4 features released
  • What we learned about our users & mobile
  • Looking forward: Q1 and next year
  • Strategic topics: tablets
Presentation slides from the meeting

Tomasz: Welcome
Everyone: Introductions

Tomasz:
Q4 themes:
1000 mobile uploaders annual goal
continue experiments about mobile contributions
foundational features

Maryana:
Recall we build things in two tiers: beta / stable
This q we built apps
"Nearby" feature - upload oriented
Met uploaders goal early

Arthur:
Q3-Q4 features (alpha/beta/production)
meeting uploaders goal early meant we had room for experiments, UX improvements, refinements
selfie-pocalypse: in-article uploading call to action was exposed to anonymous users. People uploaded low quality images, e.g lots of self-portraits.
this upset Commons community - high deletion rate, ~75% for mobile uploads
therefore we removed the feature for anons, deletion rate dropped
introduced upload tutorial for 1st time uploaders
design team helped keep unified UX, also introduced in a consistent way to upload apps (Tomasz: both to Android and iOS)
in June, added education to upload workflow, e.g. nag uploader to confirm "this is not a personal image", alongside copyright reminder
deletion rate dropped from 72% to 28% (this was after already deactivating CTA for anons)
additional upload funnels in Q3:

  • invitation in article with missing lede image
  • "Uploads" link in left navbar inviting uploads
  • watchlist
  • in "Nearby" feature, marked nearby articles without images

Nearby feature generated a lot of press coverage, lots of positive feedback. but didn't turn out to be huge driver of uploads
still figuring out exact stats, but it's clear that left nav much more effective than Nearby, article lede invitation or watchlist
Erik: overall volume of Nearby usage known?
Maryana: yes, somewhere
Tomasz: but also need to figure in how many of those Nearby view actually included upload CTAs
Erik: everytime a new feature goes viral, there's an initial usage spike, but long term?
(Arthur:) quality (= deletion rate) seems roughly the same in left nav uploads vs other channels
editing improvements:
1581 registered users edited mobile enWP since January 2012, making 5327 edits
in May, section edit links "leaked out" accidentally on mobile, huge effect (>2000 edits on that day)
integrated Echo notifications
talk pages: can now add new section, but not yet contribute to existing one
navigation UX ipmprovements
"eye candy": support PageImages extension, displays thumbnails for articles (e.g. in Nearby and search results lists)

now to the apps:
Maryana:
ran CentralNotice banner campaign for apps, for English-language wp users, and on Commons
dramatically increased number of installs
banners drew in more 1st time users (most of them Wikipedians as banners were logged-in only)
conversion rates: 1st time UX was a problem to solve
upload quality: mobile web uploads deletions still very high as compared to apps uploads
even during the May campaign
usage of uploaded files: with exception of in-article mobile web uploads, most mobile uploads don't get used often on WP (i.e. probably less often than the average image on Commons)

What we learned about mobile:
1) pageviews:
mobile still looks small compared to overall. but it grew rapidly this FY: ca 11% to ca. 16%. follows industry trends
For 2014, we project that mobile pageviews might reach 30% on WMF projects
majority from mobile web, but (reading) apps are not trivial
2) uploads:
on Commons, 4% of uploads are mobile - i.e. lower ratio than in reader pageviews
Commons app draws in much more experienced users (in terms of Commons edit count) than mobile web uploads
also, retention rate is higher for apps
Sue: so, apps users are more "our people", already experienced?
Maryana: yes
Erik: also, app does not have account creation feature
Howie: and mobile web generated the selfiepocalypse
3) login/account creation
(Maryana:)
mobile ratio is about the same in login and account creation
most users who log in on mobile are not yet editors
power users log in more often on mobile
but the ones that are *editing* on mobile tend to be newer
interviewed 5 mobile editors (1 newbie, 1 intermediate, 3 power users)
3 of the 5 used tablets
some liked mobile editing, some not
all agree simplicity is important
some power users switch back to desktop mode to do mobile edits
form factor (tiny screen...) still concern for another power user
Erik: participation in review tasks/discussions on mobile seems very doable for power users
Howie: so this means there's no data we suggest we should focus efforts on enabling long-form (whole article) editing, right? yes
Maryana: percentages now vs. 2014 projections (pageviews, uploads, logins...)
trend to mobile continues, displacing a good portion of desktop (don't know when it will exceed desktop)
most existing editors will edit as before
but new users will have entirely new experience
upside: we have a lot of room to experiment on mobile, until changes start to impact existing editors

Arthur on improvements this quarter:
1. Max and Michelle spent lots of time to get mobile frontend working in Beta labs
it's very difficult for us to test on our local machines
until recently had to do a lot of last minute testing before deployments
now can do this on Beta labs
caveat: it's not very stable yet
2. worked hard to increase visibility of bugs. created new tool "Bingle" to integrate Bugzilla and Mingle, Bugello to integrate Bugzilla and Trello (earlier, quite a few bugs got dropped)
3rd improvement: quarterly/annual planning, goals

Maryana on features in progress (Q1), e.g. diff, integrate GettingStarted, ... ...
next year's focus areas:
1. Mobile Onboarding. on desktop it's easier for newbies to poke around, learn from examples...
2. feature parity (between desktop and mobile):

  • e.g. watchlists

3. features that take advantage of the device:

  • microcontributions (e.g. dabsolver)
  • upload campaigns

--break--

Tomasz: for second half: 1. questions 2. tablet study
Questions:
Sue: good to hear we are keeping pace. Biggest threat to WP: desktop usage goes down. Is mobile usage additional or cannibalizing? Tomasz: will get to that later.
Sue: when thinking about which kind of mobile contribution form to encourage: It should not only add opportunities for additional work (new kinds of work) on mobile, but also take work off the existing editor base.
Maryana: yes. Mobile contributions still young, like a toddler which still tears down edifices when toying with building blocks, but learns quickly ;)
Erik: uploads still low, on downward trajectory as of last month at least. what's the thinking at the moment for retaining e.g. app users?
Tomasz: at the moment see web acquisition feature, app retention feature, but no strong connection between the two. used centralnotice
Erik: perhaps nudging feature after desktop uploads, to highlight existence of apps, other integration in upload workflows
Low usage would be a concern; we put a lot of resources into building these apps, now make sure they get used
Erik: performance, JavaScript...?
Jon: Improving site performance would require working closer with MediaWiki community. Upstreaming changes and ResourceLoader for example.
Erik: followup communication with Google after QuickView, on them helping with optimizations? Tomasz: not yet followed up.
Erik: consumer feedback about this has been very positive, might get help from them to implement this on our side
Sue: yes, was good for readers, bad for us
Erik: Google only optimized *perceived* end user peformance (sending along cached article snippet rather than e.g. speeding up javascript)
concern that we add lots of mobile features degrading performance in the long run
we can draw in support on metrics and measures
Arthur: agrees a lot of work to be done. all features now for logged in only. no caching for logged in users. this is a general architecture concern, but affects mobile disproportionally
Erik: agree
Erik: % of logged out traffic on mobile? ~99%? Tomasz: would need to look this up, but sounds plausible
Tomasz: any other questions? no

Tomasz: mobile strategic topic
general development: desktop -> laptop -> smartphone -> tablet
tablets most of concern
tablets grow more rapidly than smartphones did
industry-wide: tablets cannibalize desktop, laptops aggressively
Sue: in terms of purchase numbers, not time spent, right? yes
(Tomasz:) stats.wikimedia.org numbers show dramatic tablet rise (http://stats.wikimedia.org/wikimedia/squids/SquidReportUserAgents.htm )
Sue: but is desktop/laptop usage going down?
Howie: difficult to answer
if mobile growth rate would be much higher (than rest), that would suggest displacement
Sue: if overall energy for article writing stays roughly the same (maybe small decline), we could mitigate by pushing some to mobile
but other scenario would be that the pool really shrinks significantly, this would be worrying
Erik: desktop still growing, but slower
Howie: industry-wide data interesting, but really need to look at our own data
what are our editors doing? could work with Analytics and (Tomasz:) Legal on this
Kul: editor behavior similar to desktop so far?
Maryana: in my interviews, yes
Kul: at a mobile conference few weeks ago, they asked ppl if they had brought their laptop to the conference (or relied on mobile devices alone), that number dropped 30% compared to last year. Might be professional bias, but still interesting.
Sue: people may send the same kind of emails, but shorter and more typos... could be that for WP article writing, 1h tablet = 5min on laptop
Tomasz: already have more edits from tablets than from Linux [non-Android] devices
on enWP: 2.2% from tables, 2.9% from smartphones

2013 tablet goals:
1H: redirect tablet users to mobile site
new skin: Minerva, scaling up mobile design to tablet size
2H: build tablet workflows (editing/curation/media)
hires
Erik: what does redirection to mobile entail; how to make sure we don't take away features that people want?
Jon: It would be a good way to bridge gap between desktop and mobile in a 'safer' environment. For example enable VisualEditor on tablet
Erik: large differences in size even within tablet bracket, might need to adjust that
Jon: 'safer' in that we can e.g. assume tablet users more likely to be on Wifi instead of 3G
Erik: eventually could extend back to desktop version
looks ambitious with current resourcing. Tomasz agrees

Tomasz: Asks:
1. for remote team members, continue working on video/audio equipment in WMF office
2. assistance in upstreaming mobile frontend features into core
Jon: e.g. stripping TOC, sections, ...
Erik: have you talked to Parsoid team about their next generation HTML?
Jon: not yet
Erik: not spend lot of time on legacy, focus on upcoming architecture. Parsoid content precached etc. Set up this conversation soon after VE deployment rush subsides for them
Jon: work on skins...
Erik: other big needs that are not met? from Platform team?
Jon: lot of our fixmes are related to API
Sue: as with many asks, someone needs to convey them to the people who can meet them (Erik did this mostly so far in our quarterly reviews)
Yuri: is working on revamp of API features, collaborate with Parsoid. previous functionality superseded but remaining. perhaps not publicize API 2.0 until stable
3. community liaison
Yuvi: "market" the apps, especially the campaigns feature
Erik: currently 4 part-time contractors (in addition to Oliver); right now preoccupied with VE, but it's planned to redeploy them later. Could probably assign one of them; Howie might be able to help
also, Fabrice might be able to drive research on this. Think beyond WLM - team has good institutional memory about this, but also consider other cases (e.g. Wiki Loves Earth, Polish train expedition..)
Howie, Erik: keep focused/limited at first
Howie: consider the potential of the feature, along with community liaison support required to generate interest. If worth pursuing, then community liaison can be allocated
Erik: those two are related
Tomasz: server-side work
4. Beta labs stability
Erik: a year ago, weren't sure investment was necessary, with other deployments doing OK without this step
could get one or two Ops folks to help with that
Juliusz: wants to use production database with bleeding edge software version
Erik: yes, if security/legal concerns can be met
Tomasz:
5. additional capacity for apps team
continue support for dual development apps/web
Erik: app PVs only 600million
Howie: doesn't make sense to double down on readers apps separately
Erik: appreciated the level of detail in the presentation. look forward to see new contribution experiments take off
Tomasz: thanks everyone