Fundraising 2011

• What: Fundraising bug triage
• When: Wendesday, October 5, 2011, 17:00UTC (10:00am PDT)
• Where: #wikimedia-dev on irc.freenode.net

Wednesday, October 5, we will be conducting a bug triage for open engineering issues relating to the Wikimedia Fundraiser. We will be tackling some low-hanging fruit focusing mostly on some of the fundraiser metric reporting tools, as well as some non-Mediawiki related tools.

This is a great opportunity for new devs to get involved with MediaWiki development as well as for veterans to help out with one of the most critical projects for the Wikimedia Foundation. We also have some CiviCRM related bugs - so if you're a CiviCRM hacker or interested in exploring another awesome open source project, this is for you!

Please take a look at some of the open bugs we'll be addressing during the triage.

Looking forward to seeing you there!

Very exciting and fascinating test day! Throughout the week, we've been working on appeals from Maryana Pinchuk from the Wikimedia Foundation Community Department. We tested three different versions of her appeal in the US. Of the three appeals, the one that came out ahead of the others really reinforces our theory that the personal authenticity factor is critical in writing effective messages. This best performing message actually sounds like a real person talking to you. Maryana does say wonderful things about why Wikipedia is a remarkable tool, but it seems as though her personal message about why she is working here rather than anywhere else really makes her appeal. We received quite a few messages from donors with this same feedback. As one donor put it, "It was personal. If you had included marketing/PR lingo, I would have been turned off."

The next best performing appeal from the day wasn’t too far behind. It also used Maryana’s own language but didn’t seem quite as personal as the best performer. This message included a bulleted list comparing the number of servers and employees among the top five websites. One donor wrote that he donated because of the "comparison to the ‘big guys’ – and the idea that there will be no ads – no undue influence."

The third appeal was the least personal of the three and brought in significantly fewer donations than the other two messages. This appeal included quite a few general statements about Wikipedia, and it was lacking the personal factor. It was also the longest of the three appeals, which could have negatively impacted donations as well.

Maryana Banner Test

We also tested two different photos of Maryana, one photo of her smiling and one with a more serious facial expression.

• The particular photo didn’t affect donations nearly as much as the content of the appeal.
• The smiling photo had about a 5% increase in click rate.

Landing Page Report:Media:C_3009_Maryana_US_LP.pdf

Banner Report:Media:C_3009_Maryana_US_B.pdf

Maryana vs. Brandon Test

In the past several weeks, we have seen the message from Wikipedia programmer Brandon Harris perform consistently well. Once we discovered which of the three Maryana appeals performed the best, we ran this best performer against Brandon’s appeal for a short test.

Something pretty interesting happened when we ran the best Maryana appeal against Brandon:

• Brandon's banner had a much higher click rate (about 43% higher)
• BUT Maryana's appeal had a much higher conversion rate
• Overall they brought in just about the same amount of money.
• More people who actually read Maryana's appeal ended up donating, and with a 5ドル higher average donation.

We have plenty of tweaks we can make to her banner to try to bring up her click through rate including a variety of photos, titles, and wording. It is really encouraging to see Maryana’s own message resonate with readers and to find yet another story we will be able to use to appeal to readers during the rapidly approaching fundraiser.

We’re not done yet though! Yes, we’ve found a few new voices for the fundraiser but there are so many more we have not found yet. The Storytellers are still interviewing people from all over the world, so please send a message to wikistory@wikimedia.org if you’d like to help raise money for Wikimedia with a compelling message about your personal Wikipedia experience.

Report:Media:C_3009_Brandon_Vs_Maryana_US.pdf

Outside of the US, we tested an appeal from programmer Ryan Kaldari against an appeal from Wikipedia founder Jimmy Wales. We also tested two banner images of Ryan, one with him wearing his classic Kaldari hat (which is his standard accessory around the office) and one without his hat.

The good news: We've found another Wikimedian's story to to share with our readers during the fundraiser!

• Ryan's banner and appeal combination had about 15% increase in donations over Jimmy's banner and donation page combo.

Kaldari's message hit on what makes Wikipedia different from the rest of the Internet and described his personal involvement and commitment over the years. Once again, as in our past several successful appeal tests, this message used nearly all of Kaldari's own words from his interview. The authenticity and real language are proving to be extremely important in presenting an honest, compelling message to our readers.

An interesting finding:

• Ryan's donations were significantly higher coming from the banner without his hat.
• Both banners used the exact same appeal message, but the banner with Kaldari hat-less had about 30% increase in donations.
• The click rates for both Kaldari banners were very similar, the big difference was in the donations.

We suspected the hat might have an effect, but really did not expect it to impact donations this much. A possible hypothesis we're stirring with is that age could be a factor. Kaldari's banner with the hat received a higher percentage of PayPal donations than we usually see compared to credit card donations. It may be that younger readers (who tend to have PayPal accounts more than an older readers) are more engaged by his unique style than older readers. The reasoning behind this result is not too clear yet. We want to come back to this finding again to confirm the results and use a survey to find out more about the donor group from each banner.

The most exciting result from this test, is that people are responding to genuine messages from Wikimedians other than Jimmy.

We’re not done yet though! Yes, we’ve found a few new voices for the fundraiser but there are so many more we have not found yet. The Storytellers are still interviewing people from all over the world, so please send a message to wikistory@wikimedia.org if you’d like to help raise money for Wikimedia with a compelling message about your personal Wikipedia experience.

To Kaldari and all the other programmers, here's a quick comment from a donor: "My heart goes out to programmers."

Report:Media:C_3009_JimmyVsKaldari2_EN.pdf

Mhernandez 00:51, 2 October 2011 (UTC) [reply ]

We ran a one hour campaign (from 1900-2000 UTC) with Wikimedia UK to test their new setup for accepting donations by Direct Debit (a popular method for recurring payments in the UK). The banner and appeal used were the successful Brandon Harris one from our weekly fundraising tests.

Our landing page gave donors the option to either make a one time donation by credit card/PayPal, or to set up a recurring donation with a Direct Debit. We made some changes to our code and WMUK set up a backend database, to allow them to record tracking data (such as campaign and banner name) for Direct Debits.

The test ran smoothly. Single donations unsurprisingly proved more popular, but we did have a number of people successfully set up Direct Debits. Now the system is known to work, WMUK have already begun identifying opportunities for future form tests.

Pcoombe (WMF) 20:41, 29 September 2011 (UTC) [reply ]

This week we will be running banners to recruit translators in many different languages. The banners will, like the previous times, only run for logged-in users; they will target (a) wikis in specific languages and (b) English wikis geo-targetted to specific countries. The schedule is as follows:

Tuesday 27th

• Languages
  • 15:00–21:00 UTC: Afrikaans, Catalan, Danish, Finnish, Norwegian Bokmål, Norwegian Nynorsk, Swedish, Welsh
  • 17:00–21:00 UTC: German, Dutch, Italian
• Countries
  • 17:00–21:00 UTC: Andorra, Aruba, Austria, Belgium, Denmark, the Faroes, Finland, Greenland, Iceland, Italy, the Netherlands, Netherlands Antilles, Norway, San Marino, South Africa, Suriname, Svalbard and Jan Mayen, Sweden, Switzerland, Vatican City

Wednesday 28th

• Languages
  • 17:00–22:00 UTC: Portuguese, Spanish
• Countries
  • 17:00–22:00 UTC: Angola, Argentina, Bolivia, Brazil, Cape Verde, Chile, Colombia, Costa Rica, Cuba, Dominican Republic, East Timor, Ecuador, El Salvador, Equatorial Guinea, Guatemala, Guinea-Bissau, Honduras, Mexico, Mozambique, Nicaragua, Panama, Paraguay, Peru, Portugal, Puerto Rico, São Tomé and Príncipe, Spain, Uruguay, Venezuela

Thursday 29th

• Languages
  • 06:00–20:00 UTC: Cantonese, Chinese, Korean, Malay, Min Nan, Indonesian, Japanese, Tagalog, Thai, Vietnamese
  • 12:00–23:59 UTC: Arabic, Estonian, Persian, Greek, Hebrew, Hungarian, Polish, Romanian, Turkish
• Countries
  • 06:00–20:00 UTC: China, Christmas Island, Cocos (Keeling) Islands, Hong Kong, Indonesia, Macau, Malaysia, North Korea, Philippines, Singapore, South Korea, Taiwan, Thailand, Vietnam
  • 12:00–23:59 UTC: Afghanistan, Algeria, Bahrain, Cyprus, Egypt, Estonia, Greece, Hungary, Iran, Iraq, Israel, Jordan, Kuwait, Lebanon, Libya, Moldova, Morocco, Oman, Poland, Palestine, Romania, Qatar, Saudi Arabia, Sudan, Syria, Tunisia, Turkey, United Arab Emirates, Western Sahara, Yemen

Friday 30th

• Languages
  • 12:00–23:59 UTC: Bulgarian, Croatian, Czech, Lithuanian, Macedonian, Serbian, Slovak, Slovene, Ukrainian
• Countries
  • 12:00–23:59 UTC: Belarus, Bulgaria, Croatia, Czech Republic, Lithuania, Macedonia, Montenegro, Serbia, Slovakia, Slovenia, Ukraine

Monday 3rd

• Languages
  • 00:00–23:59 UTC: Bengali, Esperanto, Gujarati, Hindi, Kannada, Malayalam, Marathi, Punjabi, Russian, Tamil, Telugu, Urdu + languages we didn't get enough translators from before
    • Additional languages: Afrikaans, Danish, Estonian, Finnish, Hungarian, Malay, Min Nan, Norwegian Bokmål, Norwegian Nynorsk, Portugues, Swedish, Tagalog, Thai, Welsh
• Countries
  • 00:00–23:59 UTC: Bangladesh, India, Pakistan, Russia, Singapore?, Sri Lanka + countries of languages we didn't get enough translators from before
    • Additional countries: Brazil, Denmark, Estonian, Faroes, Finland, Greenland, Hungary, Malaysia, Philippines, Portugal, South Africa, Sweden, Taiwan, Thailand

Thursday 6th–Friday 7th

• Languages
  • 03:00–03:00 UTC: Esperanto, Estonian, Lithuanian, Malay, Norwegian (Nynorsk), Slovak, Slovene, Serbian, Thai
• Countries:
  • 03:00–03:00 UTC: Estonia, Lithuania, Malaysia, Montenegro, Slovakia, Slovenia, Serbia, Thailand

On Wednesday September 21 we will run two-hour banner campaigns to recruit translators. The banners will be run in two different campaigns: One targetting wikis in Danish, Finnish, Norwegian Bokmål, Norwegian Nynorsk and Swedish; and the other targetting users of English-language projects located in Denmark, the Faroes, Finland, Greenland, Iceland, Norway and Sweden. The banners will run for two hours from 17:00 to 19:00 UTC.

Statistics from the banners will be available at Fundraising 2011/Translation/Recruitment. Fundraising 2011/Translation Banners/September/16

On Thursday September 15 we will run one-hour banner campaigns to recruit translators. This will be a test to see what banners we should use, but also a real call for action. The banners will be run on French- and Romanian-language projects, and also for users of English-language projects who are located in France, Belgium, Romania and Moldova. The banners will only be shown to logged-in users, for one hour some time between 15:00 and 17:00 UTC.

Some statistics from the banner tests are available at Fundraising 2011/Translation/Recruitment. Fundraising 2011/Translation Banners/September/09

(Cross posted from the Wikimedia Foundation tech blog)

The last three weeks have flown by as fundraiser engineering starts to heat up. Aside from the usual bug fixes and cool new features, we added a new member to the team, made some modifications to our development process and began tackling one of our biggest challenges this year: integrating with a new payment service provider.

• Jeremy Postlethwaite joined the engineering team and is quickly getting up to speed.
• We had our first "tech showcase", where we demoed all of the functionality we’ve developed to date *. The showcase provides an opportunity for project stakeholders to see progress in near real time, which allows for better decision making as well as more effective change/risk management. This will be a regular part of our sprint wrap-ups.
• First production-level test of the RapidHtml system, which is very light-weight solution that allows for quick html-based credit card form development and provides template tokens for dynamic form elements. This was tested during last week’s weekly fundraising test, when we tested the efficacy of collecting the donor’s billing information from the landing page rather than on the credit card form.
• The Mingle engineering team over at ThoughtWorks Studios invited us to visit and see what the their development cycle is like. Seeing what their processes are like proved valuable. It hammered home that agile development is more about the mindset and values found in the Agile Manifesto than about any specific development practices. We will continue to collaborate with their engineering team to share informatoin/ideas and hope to have the opportunity to do so with other engineering teams in the future.

• Increased logging of changes that happen in mw:CentralNotice, including interfaces and filters to search and review those changes.
• Added an API to ContributionTracking which allows us to bypass the interstitial page that a donor gets sent to prior to donating when they choose to donate via PayPal.
• Began abstracting and refactoring DonationInterface (links to current development branch) in preparation for adding an additional payment provider.
• Bug fixes to the RapidHtml form delivery system in the DonationInterface extension.
• Bug fixes to our contribution auditing framework (which ensures our contribution records in CiviCRM align with accounting, etc.).

You can view sprint 5 and sprint 6 in Mingle *, and view our notes from the retrospectives.

* For access to Mingle, log in with username/password of guest/guest.

For Sprint 7, we are going all-in on integrating with Global Collect, a new payment processor which will allow us to take donations in more currencies and with more region-specific payment methods. Work will continue abstracting/refactoring the DonationInterface extension, as well as building a payment notification listener compatible with Global Collect’s "Payment Status Communicator".

If you are not interested in getting involved, help smash our hopes and dreams.<3 open bugs and/or visit us on IRC in #wikimedia-fundraising.

• Fundraising 2011