Wikimedia monthly activities meetings/Quarterly reviews/Wikipedia Zero/July 2014

The following are notes from the Quarterly Review meeting with the Wikimedia Foundation's Wikipedia Zero (Mobile Partnerships) team, July 8, 2014, 1PM - 3:30PM PDT

Present: Tomasz Finc, Manprit Brar, Yana Welinder, Carolynne Schloeder, Kevin Leduc, Oliver Keyes, Dan Foy, Adele Vrana, Smriti Gupta, Tilman Bayer (taking minutes), Erik Möller, Jorge Vargas, Lila Tretikov

Participating remotely: Yuri Akstrakhan, Adam Baso, Mark Bergsma, Brandon Black

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

Agenda:

  • Overview
  • Wikipedia Zero
    • Q4 Results
    • Carrier marketing
    • Local communities
    • Pipeline
    • Q1 goals
  • Other partners
    • Pipeline
    • Q1 goals
  • Product development
    • Q4 Results Portal
    • Work in progress
    • Portal
    • Q1 goals
  • Dependencies/Ask
Update on Wikipedia Zero and other partnership activities (slides)

Welcome and overviewEdit

Carolynne:
Welcome
Agenda: Q4 results, Q1 pipeline for Wikipedia Zero, other partnerships and our product/tech work.
Start with overview of what we're doing
Team is growing
e.g welcome Smriti - manager for Asia
Smriti and Adele do account management with partners
Dan works with Adam and Yuri in engineering
Yana, Manprit and Jorge from Legal are critical for contract work
going to hire for frontend dev, another partner manager and a communications manager (on Katherine's team)
mission: "Our team leverages partnerships to accelerate readership growth in the Global South."
Global South opportunity: (data)
growing percentage of mobile users have internet capable handsets but aren't yet using them
"our next billion users" (not an actual goal yet, but like to use that perspective)
Lila: this is humans or handsets?
Carolynne: good question - it's a multi-SIM world. These are mobile subscriptions.
Lila: fine with ignoring that as long as we focus on actual WP usage, or can correlate subscriptions with actual humans
4B people at the bottom of the pyramid (<$1.5k/year)
and also 4B people not yet using internet
also looked at penetration rates (e.g. Africa <20%)
upshot: growth story is the important one
growth is going to come from mobile, not desktop
Right now, (data) purchase decisions dominate usage in GS
That's why we do WP0
"We serve ¾ page views to ¼ global users" (mobile pageviews vs. mobile subscribers in those countries)
Lila: reason we don't use uniques?
Oliver: we don't have uniques. could use fingerprinting (reasonably accurate within 24h, becoming exponentially more unreliable after that) and extrapolate pageviews/unique ratio from there
(Carolynne:)
Next step: drill down on usage stats (slide)
Lila: why is iOS so high?
Oliver: might be an artifact
Erik: Opera Mini difficult to support technically (JS capabilites), but we need to support it if we want to reach people in these countries
Oliver: Opera says their unique numbers for WP keep increasing
Erik: (Safari UA data suggests usage on iOS devices is higher, too)
(Carolynne:)
Besides merely offering WP0, we learned we also need to raise awareness
also want to work more with Analytics on unique user data
Lila: how does WP0 look like on a device?
Carolynne: working to get WP featured on e.g. Opera Mini speed dial
separately, starting to work with handset vendors (these are open handset markets)
Lila: need to tell people about WP0 when they get a new device
(demonstration of WP0 banners)
Carolynne: just making it free, without further promotion, doesn't work
Lila: yes
(Carolynne:)
also, need to deliver great UX
country segmentation: basic/smartphone, small/large WP
initiatives to drive usage: speed dial, app preloads, find other partners (besides carriers) for marketing
priority countries: large markets
but also have a lot of obligations for smaller markets under contracts with multinational carriers and inbound requests
time allocation targets for team (slide)
building a carrier self-service portal to scale support

WP0 Q4 resultsEdit

Adele:
reach is 520M subscribers
don't have number of internet-enabled handsets, so estimate internet-enabled subscriber reach at 350M
did well on exceeding monthly PV goal
learned to prioritize partner deals: focus on big players first, others will follow
aim to have more than one partner per country
Lila: what stopped us from reaching country goal?
(Adele:)
Many reasons. The number of countries was arbitrary and prioritizing that didn't prove to be the best way to increase pageviews.
Besides a lot of the work depends on partner's responsiveness. Team needs to do a lot of followup.
Carolynne: We also made the decision not to prioritize just for the purpose of reaching that goal
don't have # country goal for next year - rather, users
Lila: makes sense
(Adele:)
Q4 launches: biggest market was Nigeria
Nepal and Rwanda were also highlighted as countries where the Wikipedia Zero launch represented a significative increase in the country pageviews.
(Adele:)
program total traffic:
from April, artificially high pageviews due to keeprefreshing.com issue
Carolynne, Dan: need to substract those
Carolynne: It's been flat despite new launches because others drop - game of whack-a-mole
one partner removed speed dial, so temporarily we weren't featured there
also, some offer general "free zone" where we are listed with other content
Erik: in some sense this is good news for zero-rating in general...
Carolynne: we are competing for promotional attention with revenue-generating plans
that's one reason why we need include other partners (besides carriers) for promotion
Adele:
Also, seasonality because of students (vacation)
Total WP pageviews were doubling in Nepal.
Carolynne: so far, we have never lost a partner
a lot of current discussion are about their CSR efforts, we want to support that

Carriers promoting Wikipedia ZeroEdit

Nepal and Rwanda did an excellent job promoting WP Zero launch locally. Banners, SMS, social media promotion.
We have learned that global press release without local promotion doesn't create the same impact in awareness and pageviews:(
other example: carriers have involved the government (Ministry of Education) to support the program and promote it in schools
Carriers in Asia were supporting the community with meeting space and general event sponsorships.
That's not the case in Africa yet: currently we don't have many active communities and/or chapters (organized Wikimedia groups) in Africa.
Carolynne: With small language communities in Asia, carriers are actually concerned about lack of WP content in that language
some supporting local WP communities, e.g. contests, providing rooms for meetups. kind of complementing our grantmaking efforts ;) (talking with Anaysuya about this)
Lila: editing on Opera Mini?
Dan: not really possible to support rich editing experience
Erik: comes down to supporting non-JS editing. WIll never be high on priority list, but a WP0 developer could work on this

Deep DivesEdit

(Adele:)
trying to spend more time on bigger, strategic countries, leaders in the region
Deep Dive in South Africa: continue to further expand what started with a great grassroots initiative
Currently working with nonprofit organization called SchoolNet which provides training on technology for teachers in SA.
Trying to address the fact that some teachers discourage students to use WP.
They have posted 3 blogposts on their newsletter talking about Wikipedia Zero and the Education Program.
Zero team is working with Education Program (Anasuya's team) to share with South-African teachers materials and overall information about WP.
Additionally, work at Sinenjongo high school is continuing, Wikimedia ZA supporting this
Lila: can we measure/monitor that?
Adele: that's actually one ask, talked about that with the chapter at the recent Wiki Indaba conference
Erik: Wikimetrics could be used for that
overall data indicates editathons (good for outreach, relation building but) don't have long-term impact. Education program has more sustained impact
Adele: we're working to include Wikipedia content in Unicef's ureport , e.g., referencing WP health content
Erik: using our existing SMS gateway?
no, they are working with Praekelt (i.e. same partner), but different system, really low tech could just be a link to the mobile web to start out.
came through Katherine's contacts

Q1 GoalsEdit

Adele: (shows launched/pipeline/target/deep dive map)
Lila: average time to close a deal?
Carolynne, Yana: several months; longest ones can take up to a year
Adele: on our side, we can launch a new partner in 4 weeks if the partners are responsive.
Carolynne: we try to reply immediately, to be responsive, but it's not worth chasing down every operator
(Adele:)
in India and Brasil, general Internet penetration is high, but low WP penetration: respectively 20% and 12%
Lila: how do we measure that? (...relation to subscriber numbers)
really need to shift to uniques
Carolynne: glass is half full - I see some traction and lots of room for growth
Lila: indeed, without a lot of investment...
(Adele:)
Expected launches in Q1 (slide)
(Carolynne:)
Challenges:

  • bundles: e.g. can't agree to be part of limited service bundle
  • Data plans

Lila: what about pre-installed bookmarks? yes, aiming for that

  • Opera Mini web pass

Lila: "get traction" isn't a very concrete goal ;)
can there be a goal to track uniques?
Carolynne: hoping it will be one of Analytics' goals
Kevin: developing capability for that, installing Hadoop cluster. right now it's a stretch goal. Also dependent on resolving on privacy issues
Lila: 75M pageview goals is for September?
Carolynne: from annual goal (125M) with some seasonality. e.g. hard to finalize deals over summer
building pipeline, hiring partner manager
besides PVs also need to watch user/market growth
Legal team helped avoid problematic deals

Other PartnersEdit

Carolynne:
OEMs:
We're preloaded on Mozilla/Firefox OS
Android handset manufacturers
Qualcomm: we'd like to get WP approved in QRD
Brazil: goal is to get a preload with at least one local OEM
want to get in contact with Google for Android OS level preload if possible
Lila: already on Windows phones?
(Adam, post-meeting note: http://lists.wikimedia.org/pipermail/mobile-l/2014-June/007382.html )
Yana: Microsoft requested trademark use
Opera Mini and Nokia Xpress
Carrier has to have an existing (commercial) partnership with them for us to zero-rate through the proxy browser
Have good relationship with Opera
They have great analytics, Oliver working with them
Wifi programs? Most wifi services in GS are metered, so could zero-rate. If it's free, don't need to offer WP0 of course, but would stilll like to promote WP
Q1 goals for other partners: (slide)
Lila: would like to understand what is happening with zero-rated Wifi, monitor that


-- break--

ProductEdit

Q4 ResultsEdit

Dan:
HTTPS support improved
got pagesize reduced
kicked off portal design, did some infrastructure work
support Nokia Xpress browser (most widely used after Opera)

HTTPS support:
for editing, privacy, app support
in early phases, were whitelisting by URL
these are obscured in HTTPS, so need to whitelist via IP instead
going back to existing partners to have them change that over too
Lila: can we tunnel HTTP via HTTPS for editing?
Dan: for Opera, also lack JS support
Carolynne: it's a priority for us, also with respect to upcoming switch of search engine traffic to HTTPS
Erik: (explains) some WP0 traffic comes from HTTPS
Lila: yes, need graceful failover
Erik: so all partners who don't support HTTPS yet are URL whitelists? yes
so we could redirect users only for those operators from non-free HTTPS to free HTTP
Mark: search redirection is goal for Q3

(Dan:)
Page size reduction
bandwith cost is a concern for some partners
Current options:

  • Opera achieves ~50% reduction
  • zero.wikipedia.org: no images, little to no discoverability

restructured code to introduce variable image quality
no noticeable difference with ~30% reduction
Yuri (on chat:) on zero without images, note that the image has been replaced with the text "click to view ...". example: low size 131KB vs 52KB
to try it out - add "qlow-" in front of any thumb url - in front of the px size of the image
(Dan:)
looking to test with partners
could become part of WP in general

Android App support
launched in June with WP0 banner support
app checks network that it is on, for free/nonfree warning
often partner tells us their IP ranges, but might not tell us about updates. app might remedy that
requires HTTPS
Other apps:
iOS app will launch soon with WP0 banner support
Firefox OS legacy app (based on our last multiplatform app)
still doing fixes. low investment thing for us
Erik: are we phasing that out?
Dan: depends on usage
Erik: might want to define cutoff
Carolynne: they are actually expanding

(Dan:)
Portal - to automate and scale partner acquisition
Erik: so this will be at zero.wikimedia.org?
(Dan:)
yes
while we still need to invest a lot of time to get large partners onboard, have a lot of smaller partners coming to us without outreach
have wireframes, will give them to frontend engineer who will implement them

Q1 goalsEdit

(Dan:) Q1 goals: (slide)
Erik: image compression rollout means pilot, or default usage?
Dan: need to do test with partners to measure actual bandwidth; our analytics team currently doesn't have the capability to do that
from then, will offer it to partners as option during account reviews
Carolynne: important for partners who were grandfathered in on zero.wikipedia.org
Erik: let's follow up on that
Lila: mixed HTTPS configuration - that means these are going to be done at end of Q1?
Dan: hope to convert half of existing backlog to support HTTPS
Carolynne: aim all for end of year
Lila: be wary of holidays at end of year...
setting a concrete number would be good (not telling you which)
Dan: need to work a lot on finding out where partners are on this
partners tend to understimate task
(goals:)
SMS pilot 2:
Carolynne: ...
Lila: what is the latest incarnation of the project?
Dan: after search, multiple choice menus, they get a SMS with article text. new: with link to online article
Erik: still with same partner?
Dan: same carrier in Kenya; all the hosting and tech work is done in ZA with Praekelt
Lila: this is a timeboxed project, right?
yes
Carolynne: we won't endorse it if it's not free
Lila: we need statistics on usage
Dan: we have them
Carolynne: there are successful commercial WP via text services in India, some abusing trademark
(Dan:)
working with mobile Web team on non-JS support
Analytics migration to our new mixed configuration

Carolynne:
Summary of Q1 goals (slide)

Dependencies/asksEdit

(Carolynne:)

  • Communications manager (working with Katherine on hiring) - hurts me personally because I have to do some of this and worry about what we're not doing

Lila: shouldn't be a budget issue
...

  • Ops support for CRM

Erik: weren't we going for third-party hosting?
Dan: yes, eventually
Mark: what specifically is needed?
Dan: e.g. change firewall rules
Carolynne: not a lot of asks, feel a little overloaded still, but requests /hires progressing
Erik: but recall the UVs
Kevin: depending on new infrastructure; Analytics is right now still in firefighting mode with regard to WP0 issues
Dan: Analytics still using old configuration pages on Meta, need to switch over to new API
Erik: need to set up Zero/Analytics meeting soon
Carolynne: also have support from Oliver, need to make use of data
Erik: looks good, always amazed about how many balls you are juggling