Wikimedia Meta-Wiki

Wikimedia Foundation Report, October-December 2014


See also the original PDF version of this report,
and (for online presentation) the Google Slides version


2008


2009
2010
2011
2012
2013
2014
2015

+ Search


Wikimedia Foundation

Quarterly Report:

October - December 2014

(Q2 of the fiscal year)


What we said and what we did

What we learned

What’s next


Objectives & design principles for this report

edit
  • Accountability: Help our movement and our supporters understand how we spend our effort, and what we accomplish.
  • Learning together: Highlight important internal & external data, trends and lessons.
  • Presentable: Anyone, from volunteer to ED, should be able to present the work of the WMF using this report.
  • Reasonable effort: Pull as much as possible from existing sources, e.g., quarterly review slide decks & minutes.

Agenda

edit
  • Summary: Scorecard, key insights and trends (5 minutes)
  • Top objectives for the previous quarter, results (70 minutes)
  • Top objectives for the next quarter (5 minutes)
  • Discussion (10 minutes)

Quarterly metrics scorecard (beta)

edit
         Participation
Sign-ups 1,527k -5.7% from Q1
+52.4% y-o-y
New editors TBD  
Active editors
(5+ edits/month, est.) 		77.5k/mo -0.6% from Q1
+2.6% y-o-y

 

          Site reliability
Read uptime
(Eng. WP main page) 		100% 100% in Q1
100% in Q2 2013/14
Read latency
50th percentile 		to come in Q3  
Write latency
50th percentile 		to come in Q3  

 

          Readership
Page Views
Crawlers excluded 		16.7B/mo +6.1% from Q1
+0.2% y-o-y
Visitors to come in Q3 comScore desktop UVs (deprecated):
455MM/mo in Q2

 

           Content
New articles 7.4k/day -44.8% from Q1
-7.2% y-o-y
Edits (on WP, est.) 9.70 million/mo -5.7% from Q1
+7.8% y-o-y

 

          Fundraising
Amount raised $44m
(exceeded $24M target) 		Year-end campaigns:
$33M in 2014
$19M in 2013
On mobile 16.1% 2013: 1.7%

Key insights and trends: October - December

edit
  • Readership: Globally, pageviews are flat. Mobile is growing, desktop is shrinking. Given a growing global potential audience, this means we need to invest in the readership experience, with focus on mobile.
    We have learned that we can move at highest velocity on mobile apps due to their self-contained nature.
  • Beyond editing: Inviting readers to perform classification tasks on their smartphone is showing promise; response quality is exceeding expectations.
  • Performance: The implementation of HHVM across Wikimedia sites is an engineering success story and demonstrates that dedicated focus in the area of site performance can pay off relatively quickly.
  • Fundraising: Mobile matters -- thanks to focused effort, we were able to increase the mobile revenue share from 1.7% to 16.1% (2013 vs. 2014 year-end campaign).

Top objectives for Q2 (selection, abridged)

edit
Department Objective / Key Results Status
Engineering Mobile App: Improved search & browsing, validated through qualitative testing

Mobile Web: Launch A/B test for microcontributions to Wikidata

Done
Engineering Editing performance: Instrument all edit funnels (after finishing HHVM rollout; halving save latency) Partial
(HHVM done)
Engineering Front-end libraries: Create server-side version of the standard front-end library; document it Done
Grantmaking Global South: India community consultation Done
Grantmaking Annual Plan Grants: Shift focus from money and process to impact and non-monetary support Done
Finance/Admin Facilities: 5th Floor Project Launch and SOW identified, approved and communicated to org Done
Communications Annual communications deliverables: Year in review video (new), annual report (web [new] & print) Done
Fundraising Revenue: Raise $24M in the quarter ($20M of which in online campaign) Done
(Exceeded: $44M)
Legal/Community Legal and Community Advocacy: Strong core support, e.g., 92% contract response time w/i 3 days Done
Talent/Culture HR: CA & CL Communications Training Rescoped

Mobile Apps

edit

What we said: Increase reader engagement

  1. Improved search & browse
  2. Pilot at least one new reader-focused feature

What we did:

1. Added Wikidata descriptions to search results, supplemented prefix search with fulltext search, switched backend to CirrusSearch

2. Added lead images and Wikidata description to top of articles, move first paragraph up so it’s visible without scrolling

Mobile Apps: Wikidata descriptions in search results

Before
Now
 
 


Mobile Apps: Top-of-page improvements

Before
Now
 
 

Screenshots including text from en:Barack Obama, by Wikipedia contributors, under CC BY-SA 3.0

Mobile Apps

What we learned

  • Readers love the lead image feature - user comments after the launch were very positive.
  • We were able to reduce the rate of searches without results from 18% to 7%.

Mobile Web Contributions

edit

What we said:

Launch A/B test for microcontributions to Wikidata (“WikiGrok”)

What we did:

Delivered and tested WikiGrok UX

Mobile Web Contributions: WikiGrok UX

  • Yes/no question vs. more complex version
  • Answer options for each pre-generated by automated analysis
Version A
Version B
 
 
Screenshots including text from en:Brad Pitt, by Wikipedia contributors, under CC BY-SA 3.0

Mobile Web Contributions: WikiGrok UX

What we learned

  • Engagement is high
  • Quality is high (80-90% correct), even from readers (compared to logged-in users)
  • Stickiness is encouraging - 26% answered again, even without any gamification
  • Need to dig more into question types. Which ones yield the most useful contributions?

For more detail and other goals, results and insights, see the Mobile Web and Apps teams' quarterly review documentation

Editing performance

edit

What we said:

Instrumentation of all edit funnels in wikitext editor and VisualEditor on all platforms, after completion of HHVM rollout

What we did:

HHVM rolled out successfully, halving save latency

Instrumented edit funnels in VisualEditor

NOT DONE: Instrumentation of funnels in wikitext editor

Editing is slow. We want to make it faster.

Instrumenting edit funnels means measuring how many people start an edit, how many people hit errors, how many people complete it, etc.

Editing performance

 
Mean save time during HHVM rollout
  • Finished HHVM rollout
    • Reduced mean save time from ~6s to ~3s
    • Reduced median save time from ~7.5s to ~2.5s

Also in Q2:

  • Implemented optimistic saving for wikitext editor, reducing median save time by 300ms.

For more detail and other goals, results and insights, see the MediaWiki Core team’s quarterly review documentation

Front-end libraries

edit

What we said:

  1. Create server-side version of the standard front-end library to enable use in all parts of MediaWiki
  2. Document it

What we did:

Completed and shipped the port to PHP

Developers can now find an extensive documentation of OOjs UI on mediawiki.org

OOjs UI (Object-Oriented JavaScript – User Interface) is a library that allows MediaWiki developers to rapidly create front-end web applications that operate consistently across a multitude of browsers. It was first developed for VisualEditor. It contains easily reusable interactive elements such as buttons, switches, popups and progress bars. The new server-side version enables generating compatible output in PHP in cases where JavaScript is not supported.

Global South: India Community Consultation

edit

What we said:

Convene a strategic community consultation about the future of Wikimedia work in India

What we did:

  • Gathered Wikimedians from 15 different language communities
  • Compiled proposed roadmap of possible activities
  • Formerly WMF-run catalyst program continues with local grantee CIS, gaining credibility

Global South

What we learned:

  • Main lesson from Catalyst programs (in Brazil and India): Endorsement and active interest by the community is essential for effective deployment of paid staff.
  • We need to continue to maintain trust and effectiveness with our communities through well-designed and high-mandate community consultations.
  • In the Global South, most groups are in need of (and would welcome) proactive support in strategy and non-financial resources.

For more detail and other goals, results and insights, see the Grantmaking team's quarterly review documentation

Annual Plan Grants

edit

What we said: Shift focus from money and process to impact and non-monetary support

What we did:

Streamlined forms and simplified reporting requirements

Emphasis on impact in the inputs to the FDC’s decisions, including impact analyses and staff assessments

Annual Plan Grants (APG) are Wikimedia movement funds allocated to support an organization’s overall annual plan to achieve mission objectives. The all-volunteer Funds Dissemination Committee (FDC) makes recommendations to the WMF Board about APG funding proposals, supported by input from WMF staff.

Annual Plan Grants: FDC deliberation outcome

  • $3.8 million to 11 Wikimedia orgs vs. $4.4 million to 11 Wikimedia orgs last year.
  • Total amount decreased by 14%.
  • The FDC reduced 5 largest grants by $750k, increased 6 smallest grants by $190k.

While grant requests and overall budgets have increased year after year for almost all the organizations, impact has not grown at the same rate… This is of significant concern to the FDC, particularly for the largest organizations that have considerable financial and staff resources.

— FDC funding recommendations, 2014-2015 Round 1

Annual Plan Grants: Outcomes reported by grantees

 

Q2: Fiscal Year’s Second Quarter, or Quarter Two
YTD: Year-to-date
Includes data reported by 12 organizations.

Annual Plan Grants

What we learned:

  • Organizations want non-monetary support
  • Organizations are recognizing their potential and focusing on institutional partnerships

For more detail and other goals, results and insights, see the Grantmaking team's quarterly review documentation

Communications

edit
 
Photo by Sydellewillowsmith, under CC BY-SA 3.0
“Wikipedia Edit 2014″ video by Victor Grigas and others, licensed under CC BY-SA 3.0.

What we said:

  • Delivery of soft and hard copies of annual report (“Knowledge is a foundation”)
  • Delivery of first ever YIR (year in review) video

What we did:

Published annual report online on January 21, print copies to be mailed in February

YIR video published on December 17
(>225k views, 130 pieces of press coverage from 19 countries)

Communications

What we learned:

  • A carefully created product like the Year in Review video can reach a worldwide audience -- there is demand.
  • Although the Wikimedia trademarks are highly valued, the WMF has very little baseline data on brand perception and audiences. In order to tell a complete, compelling story about our movement, we require more data around our global brand.

For more detail and other goals, results and insights, see the Communications team’s quarterly review slides

Facilities

edit

What we said:

5th Floor Project Launch and Scope of Work identified, approved and communicated to org internally

What we did:

Done (with one month delay due to dependencies)

Since early 2011, the Foundation’s office has occupied the 3rd and 6th floor of 149 New Montgomery in San Francisco. After rearrangements of the 3rd floor (in 2013) and the 6th floor (in 2014) to accommodate growth, the office space is now being extended to the 5th floor.

Revenue (Fundraising)

edit

What we said:

Raise $24M in the quarter ($20M of which in December online campaign)

What we did:

Done and exceeded: Raised $44M

  • Reached annual goal of $58M six months ahead of schedule
  • This puts us in strong position in case overall traffic in key fundraising countries declines again next year

Revenue (Fundraising)

While we are now fundraising year round internationally, the bulk of our revenue comes in December annually from the 5 biggest English speaking countries.

 

Revenue (Fundraising)

What we learned:

...but we are fundraising successfully in non-English countries too: Raised $2.8M in 2-week campaign in France.

Last campaign (2012)  
WMF 2014 campaign
 


Revenue (Fundraising)

What we learned:

Bulk is still raised via banners on desktop, but massive gains on email and mobile

   
Year-end campaign donation breakdown

For more detail and other goals, results and insights, see the quarterly review slides of the Fundraising/Fundraising Tech team

edit

What we said:

Top notch, quick legal advice and support on wide host of issues (e.g. 7/day turnaround of 85% contracts)

What we did:

  • E.g., 90 contracts processed in Q2
    • 8% ↓ from Q1, but 34%↑ from 2013
  • 92% response time w/i 72 hours

Contracts drafted related, for example, to venue rentals, preparations for Wikimania, hiring international contractors, tech services, non-disclosure agreements, grants, and chapter and user group agreements

Human Resources: CA & CL Communications Training

edit

What we said:

Design training for Q3 that ensures all community facing employees (Community Advocates - CA and Community Liaisons - CI) have increased clarity and skill in dealing with conflicts and situations that arise.

What we did:

Rescoped to fleshing out a map of community interaction points before designing targeted training.

Human Resources: CA & CL Communications Training

What we learned

Preparing the training needed to involve the identification and mapping of key pain points, and the development of cross-departmental engagement guidelines.

We now plan for training to occur on two levels:

  • generalized out-of-the-box training that is widely applicable
  • customized training for specific audiences that is leveraging in-house expertise.

What’s next: Top objectives for Q3 (selection)

edit
Department Objective / Expected results
Engineering Prepare to provide VisualEditor to new users on all wikis
Engineering Release and test WikiGrok and Collections
Engineering Full instrumentation of Editing User Experience to support VisualEditor rollout
Grantmaking “Inspire” campaign to source and support 20 new gender-focused projects
Grantmaking Impact evaluation overview of 10 movement-wide programs, and Edu program toolkit
Finance/Admin Update and prepare 5th floor for occupancy
Communications Deliver “State of the Wiki” report, parts I & II of 3
Fundraising Raise $3 million
LCA Ongoing core legal support, like monthly board meetings. Initial high-level strategic consultation with community.
HR Provide a better system for international contractors
Retrieved from "https://meta.wikimedia.org/w/index.php?title=Wikimedia_Foundation_Report,_October-December_2014&oldid=11361629"