Fundraising 2010/Updates

From Meta, a Wikimedia project coordination wiki
In other languages
  Updates       Testing       Translation       Messaging    

Updates from the Fundraising Team


Information about testing, our findings & plans for the upcoming weeks

Post-fundraising thoughts and reports[edit]

January 7: Contrib Campaign & 10th Anniversary[edit]

We just switched the “Thank You” banner to ones for the contribution campaign and Wikipedia's 10th anniversary. Traffic for anonymous users will be split between the two banners leading up to January 15th. On the English Wikipedia, the contribution team has been working hard to ensure an easy edit experience for new users, and to get current editors involved in the the “Great Backlog Drive.”

For the Contribution campaign, we're starting with Ragesoss's graphic banner submission of “your contribution is a piece of the puzzle” and are in the process of building more community submitted banners. If you have a good idea, it isn't too late to submit it!

The 10 year banner links to the party planning page, inviting people to attend volunteer-organized parties and events, or plan their own.

We're looking to build quite a bit of excitement for the two projects, help us spread the word however you can. - Deniz (WMF) 19:33, 7 January 2011 (UTC)[reply]

January 3: Fundraising Style Guide[edit]

As we end our 2010 fundraiser, we are focusing on putting together a style guide, or “Fundraisers Guide to the Galaxy” with everything we've learned this year. The guide will focus on testing, and how not to reinvent the wheel, so future fundraising teams can take off running.

We'll be covering outreach and community involvement, tech lessons, testing, design and form optimization, messaging, translations and feedback from the community, donors and readers.

Have suggestions for next years? Feedback about what worked well, and what didn't? I'd love to hear them all so we can build a comprehensive guide for next year's fundraiser. - Deniz (WMF) 01:38, 4 January 2011 (UTC)[reply]

January 1: 2010-2011 fundraiser draws to a close[edit]

I’m delighted to report that the Wikimedia Foundation can ring in the New Year with the close of our seventh annual fundraiser, having exceeded our goal of $16 million. More than half a million people pitched in an average of around $22 each to support Wikipedia and its sister projects, in our shortest (and most successful) fundraiser to date.

Our community of volunteers is deeply honored that, in only 50 days, 500,000 people from 140 countries came together to support the only non-profit, user supported top-10 website in the world. In addition to this humbling support for the Wikimedia Foundation, our chapters around the world have raised millions of dollars from hundreds of thousands of donors of their own.

We want to thank every one of our donors for making this year a success, and on behalf of the fundraising team, I personally wish to thank the one-thousand community members who helped us create and test messages, who wrote appeals, and translated banners and letters into over 80 languages.

In 15 days, Wikipedia will turn 10 years old. Since the beginning, Wikipedia’s community of readers and editors have remained dedicated to keeping the site ad-free, and free for use for its 400 million monthly visitors. This year’s success demonstrates a continued commitment to those principles.

With the close of our annual fundraiser, we are transitioning into the contribution phase of the campaign. We will be running banners for the next few days to thank everyone who came together in the spirit of creating and “effectively disseminating the sum of human knowledge available for all.” We will also begin to celebrate Wikipedia’s tenth birthday, with banner ads encouraging readers to join us in a local celebration.

We want to invite every one of the readers of Wikipedia and its sister sites to make their first edit, or upload their first photograph, and join our community of volunteer contributors to continue the growth of Wikipedia for the next 10 years.

Thank you again, and happy New Year! Here’s to 2011, and to the next 10 years!
Regards,
Philippe Beaudette
Head of Reader Relations

December 30: Banner Testathon[edit]

Since Monday, the fundraising team has been conducting a big banner "testathon." The team is in the process of testing over 20 text banners with Jimmy's picture. So far we've run:

The two banners performing the best from Monday's tests:

What's most exciting about these tests is that the $5 banner has actually outperformed the "personal appeal from Jimmy Wales," that we had been unable to beat before.

On Tuesday we tested removing the "only" from Only _ days left, but there was no significant difference between the two. We also tested $10 against $5 in "if everyone reading this donated $__, our fundraiser would end today. Again there was little difference between the two.

Yesterday, the team introduced an animated fading banner, which opens up a lot of great testing potential. We can now test more messages, and how different messages interact together. So far, we have seen a huge boost in donations from the new banners. We're currently running variations of the messages that have performed best over the past 2 days with the animation.

I've updated the testing page with results from the banner tests on Monday and Tuesday, take a look and tell us what you think! - Deniz (WMF) 18:20, 30 December 2010 (UTC)[reply]

December 24: Ask Strings & Banners[edit]

We let the December 21st ask string test run overnight; there wasn't a significant difference between the two donation strings ($15-$500 and $20-$250). On Tuesday, we ran two "non-profit" letters with the same ask strings, and tested two new banners in the afternoon.

Banners

Test 1

  • Control - Please read: an urgent appeal from Wikipedia founder Jimmy Wales
  • Imagine - Imagine the Internet without Wikipedia. Wikipedia needs your help right now.

Test 2

  • Control - Please read: an urgent appeal from Wikipedia founder Jimmy Wales
  • Learned - You learned a lot from Wikipedia this year. Please help Wikipedia pay the bills for 2010.
Letter & Ask Strings

There is a one-line difference between the two letters:

Comments, questions & feedback always appreciated!- Deniz (WMF) 01:56, 25 December 2010 (UTC)[reply]

December 21: More Ask Strings[edit]

We continued to test radio button ask string variations overnight. There was no significant difference between the $5, $20, $35, $100, $250, $500, $1,000 string and the one we've been running ($20, $35, $50, $75, $100, $150, $250), so we removed the $1,000 string. Currently, we are running two of Jimmy's best letters from Sunday's test, the “I'm a volunteer” and “Temple for the mind” appeals. We're still trying to find the “sweet spot” for the ask strings, and are now testing another set alongside our control. The new ask string starts slightly lower, at $15, and goes up to $500. We'll likely leave this test up for another day as the two ask strings being tested are so similar, it may take longer to get statistically significant data. - Deniz (WMF) 20:09, 21 December 2010 (UTC)[reply]

December 20: Removing Banners for Logged-In Users[edit]

Today, as Philippe announced on Foundation-l, we have temporarily taken down banners for all logged-in users. We expect to have them down through the end of the year, and will turn them back on sometime after the first of January for the final end-of-campaign push.

Our taking down the banners has no affect on what local chapters are running, our actions will only show up in areas where the Foundation controls messaging.

At this point, almost everyone from the community or with an account who was going to donate has likely done so. We're seeing almost no money coming in from logged in users anymore - so we decided that it was better to drop the banners entirely. With that said, if you want to donate.... you know how!

As Philippe wrote, “enjoy the respite,” and enjoy it until the final push, in January! - Deniz (WMF) 02:14, 21 December 2010 (UTC)[reply]

December 20: Ask Strings[edit]

Earlier this morning we tested two different suggested donations 'asks' in the text of Jimmy's letter -- ($10, $20, $35 vs. $20, $35, $50). The lower ask amount tested better, which is contrary to what our testing found last month. The change may be indicative of a different time in the campaign.

Currently, we're running a similar test using radio buttons with a higher series of suggested donations on the green side of the payment form. We're testing the suggested donation string we've been running ($20-$250) and a new ask string which begins at $5 and goes up to $1,000. We hope to catch both the people with just a few dollars to contribute, and those making larger end-of-year tax deductible giving. --Deniz (WMF)

December 19: Sunday Testing[edit]

We did some arranging and rearranging of the Jimmy letter and testing of a few new ways of explaining our campaign. Some things did poorly compared to the current Jimmy letter, and somethings did about as well.

For example, explaining our need for more donors in this way seemed to be effective: "Last year about one in 1,000 people who use Wikipedia donated. To reach our goal this year, we need two in 1,000."

It is an inexact science, testing landing pages like this, because we don't have large enough sample sizes for 95% or higher confidence. But it still makes sense to try many different things out, looking for something that might do massively better than the control. In other words, if we test 2 letters against each other, and A does 5% better than B, we might only be 50% confident that A actually did beat B. But we can be more confident that neither A or B will do 100% or 200% better than the other if we tested it with a proper sample size.

A few of the new arrangements did within 5% of the control. A couple were 20% or so lower. None quite beat the control. We decided to test last year's letter, just for fun. It was about 50% lower! So that made us feel good that we have actually been accomplishing *something* this year.

At the end of the day, we left the control and the next two best letters up over night. One did 4% better than the control in donations per readers. It did 37% better than the control in total money earned. So we're leaving that one up today while we do a bunch of banner and ask string testing that Megan has planned today. Zackexley

December 18: Chapter Update & Recap of this Week[edit]

A short recap of what was tested and new changes:

  • Ran a new letter from Jimmy
  • Tested the language on the banners, and found that “please read: an urgent appeal from Wikipedia founder Jimmy Wales” performed best
  • Tested border frames on the banners with different weights, which improved click-through rates. The 1pt bordercame out ahead.
  • Sue's update banner and Jimmy's new letter went up on multilingual projects
  • Recurring donations are now live!
  • We sent out a series of emails to past donors informing them of the new sustaining donor option
  • We are currently running a banner with urgent bolded in the text. Looking at the initial numbers, it looks like the original, uniformly bolded banner text is better. We're running the all-bold banner now

Friday, we wrapped up testing of the Jimmy banner with two kinds of progress meters, one that has only the dollar amount, and one that wrote out how much was left to go. The latter meter performed better and is currently running. We had previously tested progress meters on Sue's banner, but they hurt the click-through rate. This may be because the bar is more filled in now as we're closer to our goal, or that the "please read" text is now centered over the meter.

Chapters Update[edit]

Wikimedia Hungary

The Hungarian chapter is now hosting their fundraising operation entirely on their servers. Previously we had hosted the first step, the appeal page, for them. They are currently running abanner and letter from Tamas, their chapter's former Vice President at 95% and the control Jimmy banner at 5%.

Wikimedia Italia
Is running an active offline campaign. They had a half page spread in the Italian publication Internazionale last Saturday, and plan to start a campaign on Milan tramways soon using a fundraising banner they designed. They may use the same banner on it.wikipedia once they create a landing page.

The Italian Chapter also has an offline campaign using the URL http://www.sostienilacultura.it, which redirects to their chapter donation page.

Wikimedia De

The German chapter is currently running an appeal from WMDE director, Pavel; testing a banner with, and without a "read now" button.

Wikimedia Israel

The Israeli chapter has begun running Lilaroja's editor appeal, and a banner by one of their own local editors, Tamar. They are running the editor appeals alongside the control banner.

Wikimedia France

The French Chapter has begun to run Lilaroja's banner, with text which reads “through the work of Lilaroja and thousands of volunteers, Wikipedia is always there when you need it. Today Wikipedia needs you. Make a Donation.” The banner leads to a localized landing page.

Wikimedia Australia

The Australian Chapter is currently running thank you banners for the State Library of Queensland, who recently donated 50,000 images from their collection to WMAu. They are using some of the donated public domain images on these banners, they look really great!

Wikimedia Austria

The Austrian Chapter will begin running Lilaroja's editor appeal at 90%, keeping the control banner at 10%.

Feedback & comments are always welcome - Deniz (WMF) 23:38, 18 December 2010 (UTC)[reply]

December 16: Bordered Jimmy & Emails[edit]

Today we continued our upward trend, but unfortunately the database that hosts our live contribution statistics came down again, leaving me with nothing to refresh madly with excitement It's back! That particular disappointment aside, we've continued to test banners and polish language today, and work on our first ever email campaign, inviting people to make a recurring monthly gift to the Foundation.

We ran the same banner with an urgent appeal from Jimmy, only slightly tweaking it to test adding borders around the edges. We tried:

We can tentatively say that so far it appears that the new bordered banners are doing better than the non-bordered versions we've been running.

The email campaign is doing well; we have had about a thousand recurring donors sign up, and in the past few days the emails have brought in nearly half a million US dollars.

More to come, but till then -- tell us what you think. - Deniz (WMF) 03:19, 17 December 2010 (UTC)[reply]

December 14: New Letter from Jimmy[edit]

See the stats for today's test

Monday, Jimmy wrote a new appeal while on a plane to Moscow. His high-altitude composition has had an incredible effect on donations. Since we put it up yesterday afternoon our numbers are swinging back upwards from the downward trend that we had seen before. That trend wasn't at all unexpected: it's a natural fatigue factor in a long campaign such as this.

Jimmy's letter is emblematic of the mission we support, and reminds readers that, like most of us, he is a volunteer.

In addition to putting up the new letter, we completed two banner tests last night and today. We ran the control text of “Please read: a personal appeal from Wikipedia founder Jimmy Wales” against “Please read: an urgent personal appeal.” The urgent appeal language tested better.

Yesterday we experimented with replacing “please read” with “urgent” – there was no difference between the two.

Sue's update banner went up on multilingual projects yesterday evening; and we have an "urgent" translation request for Jimmy's new letter. If you can, please help us localize the appeal. His newest letter is quite idiomatic, so we're emphasizing localizing, not translating. We imagine “temple for the mind” might not make sense in many languages :)

In addition, we're doing some email testing, primarily around a new method of giving that we're featuring: automatic monthly donations!

I've been updating the banner testing data as it becomes available, check it out, and of course, tell us what you think. - Deniz (WMF) 15:52, 15 December 2010 (UTC)[reply]

December 10: Weekly Recap & Recurring Donations[edit]

Editor Appeals

On Tuesday of this week we tested Liam's appeal. We ran two banners, one with the traditional banner layout with the image on the left, and one with his picture on the right. The test showed that the original layout performed better. In addition to the banner test, we ran two variations of his letter. The two were similar enough that there was no significant difference in donations.

Wednesday we ran the most successful editor appeal, Lilaroja, at 95%.

Last night we launched the multilingual versions of Lilaroja and Kartika, they are currently running on most projects.

Progress Meters

We introduced progress meters on Thursday, adding two types of meters to the landing pages: one counts the number of donors who have contributed so far, and the other the amount raised so far. Out of the two progress bars, the one counting $ raised has the highest donation rate, and also performed better than the landing pages with no meters.

Sue Gardner Appeal

On Thursday we started testing an appeal from our Executive Director, Sue Gardner. People seem to be responding well, we brought in $40,000 more than the previous day.

Landing Page Test:
We ran two letters from Sue testing the length of the appeal, the shorter version did better. We dropped the longer letter with the bolded phrases.

Banner Test :
We tried several versions of Sue's banner; testing white vs. gray banners, and the gray banners with a progress meter. Our preliminary numbers show that the white banner had a higher click-through rate, and that banners with the progress meters did not do as well as the plain ones.

We have taken down all of Sue's banners except for the winning white appeal, and tested changing the text on the banner. We kept “please read: an update from Wikimedia Executive Director” and ran it alongside “please read: an appeal from Wikimedia Executive Director Sue Gardner;” the results were consistent with our previous tests, the 'appeal' banner performed best.

We also experimented with rearranging the layout of the banner; moving the button to the far right and centering the text like the Jimmy banner. The newer banner performed better.

Friday we ran two more Sue banners. One was zoomed out, and the other included some of the original photo's background. The zoomed out banner with the white background performed the best, it is the only Appeal from Sue banner currently running.

Recurring Donations

Thursday night, we started a trial offering to test recurring monthly donations. We've been working on supporting sustained giving for a while now, and donors have frequently requested the option. Year-round support makes all of us in fundraising smile, and we're sure OTRS volunteers (who have to keep saying "no, we don't have that") feel the same way :). We completed testing two ways of selecting monthly donations – a radio button at the top of the form, and a button under the current credit card or Paypal payment choices. The button at the bottom was the clear winner. Presuming recurring donations work well, we plan to introduce them into our giving options permanently.

As always, share your feedback & comments - Deniz (WMF) 20:31, 11 December 2010 (UTC)[reply]

December 9: Chapter Overview[edit]

This year, 12 of our local chapters are participating in the Fundraiser and accepting donations in their own countries. We've asked some of their representatives to give us a short update about the progress of their campaigns so far.

Wikimedia Deutschland

Wikimedia DE has had an incredible fundraiser so far, they have received about 40,000 donations, 100% more than last year already. They started their campaign running last year's Jimmy banner, and recently introduced the updated control banner. The new banner is performing three times as well as the original, and their average daily income is as high as their most successful day of fundraising in 2009.

The chapter has also been asking their donors for stories, and have received comments or stories from one third of all donors. In the coming weeks they plan to release an appeal from Pavel, including a video they've produced; and an appeal from editor Church of emacs. Once they've reached their goal, Wikimedia DE will continue running a different campaign that they are in the process of designing.

Wikimedia UK

In terms of numbers, Wikimedia UK believes they have a lot of growth still to come. This year, donations have primarily come through Paypal, with only a minor percentage of checks. The chapter has been frustrated by the lack of an automated reporting system their donor database, they've had to input all their information manually. They are still working to find the best system to handle both checks and bank transfers in order to accurately record and thank donors. WMUK's volunteer team hopes to establish a credit card gateway in the near future, but it has proved to be time consuming to set up.

What have they learned so far? "Metrics are everything." Their testing has been extensive and invaluable in optimizing their processes. WMUK's findings confirm the results from our past few months of testing; the control banner has consistently outperformed other banners. They hope to have a paid staff to help them reach their fundraising potential next year.

Wikimedia Sverige

The Swedish chapter is also running the control banner. They are using Payson, a local Paypal equivalent, to collect online donations. The chapter recently added their bank account information to their landing page and found it increased donations “a little.” Unlike Payson and other payment gateways, bank transfers have no transaction fee; however the bank doesn't collect important donor information, such as email addresses, to help build a donor database.

Wikimedia Israel

The Israeli chapter has a much loftier target this year because they are hosting Wikimania Haifa this summer. They are running the green control banner, and have localized their landing pages in six languages, bringing in 1.5 times the total of last year's fundraiser already.

Wikimedia Russia

The Russian chapter sites are using the control banner, with considerable success. Their projects continue to grow rapidly, and the chapter has received seven times last year's contributions.

Wikimedia Nederland

The Dutch chapter is running the control Jimmy banner and testing changing the text of the banner. They are currently exploring possibilities for new banners.

Wikimedia Australia

This is the first year WMAU has participated in the fundraiser. They had difficulties with Paypal at the start of the fundraiser that have since been resolved. Despite the setback, their efforts have been incredibly successful.

Wikimedia Austria

This is the Austrian chapter's first time fundraising. Their donation-per-impression rate is the best among the chapters, despite having missed a few days of testing.

Wikimedia France

The French chapter has exceeded last year's donations. They ran a localized campaign at the start of the fundraiser, and have now begun running the Jimmy banners being seen on other projects. They will be introducing new banners shortly to run alongside Jimmy's appeal.

Wikimedia Magyarország

The Hungarian chapter is running the control banner and have received three times the donations in under a month of fundraising than their entire campaign last year. They are finding that the donation abandonment rate on Paypal is quite high, it looks like only 1 out of every 10 people who click 'donate' complete the transaction. They believe this is because the Paypal interface is only in English – and can't be localized.

To combat the number of abandoned transactions, they have added their bank account details under the donate buttons, which seems to have made a positive influence; they are still waiting for data from their bank. Next year WMHU is looking into hiring a developer to write an online banking system so they can bypass Paypal.

Wikimedia Italy

The Italian chapter is also running Jimmy's appeal, and have doubled the number of donations from last year. They have found that people are likely to join the Italian chapter once visiting their site. As a small chapter, this has been incredibly valuable for them. Prior to the fundraiser they had about 180 members, and currently have more than 300.

Last week the chapter experimented with magazine advertising, picking a publication whose content is released under a CC license and which has high distribution and readership within Italy (about 250,000 copies are sold a week). So far the ad hasn't spurred a particularly large response, but they are waiting to fully evaluate the outcome.

Wikimedia CH

The Swiss chapter is the only one that doesn't have an online payment gateway. Donations to the chapter are made via payslips deposited at the post office (the standard method for contribution in Switzerland) which is working incredibly well for them. WMCH is running the control banner, and have localized landing pages in Italian, French, English and German.

Translation Update

Volunteers continue to localize messages at an incredible rate. We are still asking the community to help us get these published!

Number of languages ready:

  • Jimmy's Appeal: 80
  • Core Messages: 40 published, ~40 more partially translated
  • Benefactors Page: 64
  • Social Media Messages: 48
  • Kartika's Appeal: 14
  • Lilaroja's Appeal: 5
  • Abbass' Appeal: 2
  • Gomà's Appeal: 7

If you would like to write an update on your chapter's progress, or have more to add, please share! And as always, tell us what you think! - Deniz 01:23, 9 December 2010 (UTC)[reply]

December 7: Donor Stories & More Editor Appeals[edit]

See the stats for today's test

Throughout the Fundraiser we've been collecting donor stories through our Thank You page. We've received over 25,000 comments and stories so far. Philippe took the stories from just the last four days and created this word cloud. It's incredible to see the way our donors think and write about us – we can learn quite a bit from visualizations like this. If you have a story you'd like to share, send it to stories(_AT_)wikimedia.org.

Yesterday, we added a new editor appeal to our testing rotation for US projects. We used Victoria's appeal to test both banner text and landing pages.

Landing Page Test:
We ran three versions of Victoria's appeal: two letters and one two-step landing page with the “Imagine a world” quote on the second step, which used both versions of her appeal. We ran this same test last week with Sage's appeal, but the numbers were not significant enough to discern a definitive answer.

Banner Test:
We tested whether changing the text on the banner would influence the click-through rate. One of Victoria's banners read “Who edits Wikipedia? I do” with a “find out more” button, and the other text was our control “Please read: A personal appeal from Wikipedia author Victoria.” Our control banner did noticeably, and consistently better; which is encouraging as we've been running the 'better' banner since the launch of our campaign. We have removed the "who edits" banner and are running the “please read” version at 40% now.

We had changed Sage's banner to also say “please read,” to try and raise the click through rate, but it remained low. When a donor did reach the letter and donation form, the conversion rate was higher than other editor appeals. This is a strange case, but may be due to his banner 's design varying so drastically from the other appeals. We have taken Sage's appeal down for the time being to allow other banners to run at a higher percentage.

Current Banners
  • Victoria @ 40%
  • Lilaroja @ 40%
  • Gomà @ 5%
  • Kartika @ 5%
  • Abbas @ 5%
  • Jimmy @ 5%


Today we will be testing an appeal from Liam, and are anticipating embedding the Contribute video into the landing page to test this week.

We have translation requests for Abbas', Goma's, Kartika's and Lilaroja's appeals currently. Please help us translate these if you can. --Deniz (WMF) 18:55, 7 December 2010 (UTC)[reply]

December 4: Weekly Recap[edit]

See the stats from today's test

This week was focused on testing Editor Appeals on US projects. We ran several versions of each letter; testing the length of the appeals, the amount of the suggested donation within the appeal's text, and adding a photo of the editor on the landing page. We also looked at which messaging, themes, or individual stories resonated with donors.

We completed testing of Thursday's new appeal from editor Lilaroja, and are now running her most successful appeal, the “dollar, a love note, and a broken heart” letter. This version has a fundraising mention at the beginning of the letter, and is more reader focused than other appeals we have run. The letter also gives specific examples of many different people supporting Wikipedia in whatever way they can.

On Friday we introduced five more editor appeals: three letters from Abbas, and another two from Sage. What's unique about this batch of editor letters is the length. Abbas' letter is the longest we've run yet, and Sage's the shortest.

We're also conducting another test with Sage's appeal where the second page of the two-step payment form does not have a copy of Sage's letter – but a banner with the “Imagine a world” quote. We're interested in these findings as they may simplify the deployment of these pages on all projects.

Banners Currently Running
  • Jimmy: 5%
  • Kartika: 5%
  • Gomà: 5%
  • Lilaroja: 28%
  • Abbas: 28%
  • Sage: 28%

We've lowered the percent at which both Kartika and Gomà run as they've been up for a few days. These percentages will change as we introduce new editors and adjust for performance.

Findings from this week of testing
  • The original two-step payment form we've been using is the most effective, it performed better than the new one-step process.
  • Adding an editor's image to the landing page did not significantly affect donations.
  • The click-through rates on the editor banners continue to be on par with the winning Jimmy banner, but bring in fewer donations.
  • Many donors appear to relate better to letters which focus on readers showing support instead of individuals editing.
  • There has been a positive response to the new editor banners, the variety keeps our campaign interesting.

On Monday and Tuesday of next week we will introduce two more editor banners into the testing mix. We love seeing editors on banners. Want to be on one? Write an appeal!


We are still working on writing a script to automate the posting of testing numbers – we'll have them for you next week, promise :) As always, tell us what you think! - Deniz (WMF) 01:36, 5 December 2010 (UTC)[reply]

December 3: Tweaks to the Fundraising Meta Pages[edit]

I visited the Wikimedia Foundation offices today in SF, and Philippe (WMF) immediately put me to work. They are relentless here! So Deniz (WMF) and I decided to take a pass at the pages here to make them a bit easier to navigate.

We decided to focus on making it as easy as possible for people who have not been actively engaged in the fundraiser to get a quick overview, then dig down into the data. To help us figure out where some of the problem areas were, Deniz used me as a guinea pig (since I am part of this target audience), watching me attempt to find information on the site. We took the results from that, came up with a list of over 20 usability fixes, then went about fixing what we could (about 19 of them).

Some highlights:

  • We made the Fundraising Update link more prominent on the main Fundraising 2010 page. We also eliminated the "Check out the data" link at the bottom, because it was sort of a catch-all. The Update page is a better route to that information.
  • We simplified the sidebar on the subpages, moving the updates and stats sections higher, and removing outdated links.
  • We removed a few tabs (although we're still having trouble with templates; help?!), and we reordered them to make more sense.
  • We went through the rest of the content, updating outdated information, and adding explanatory text and labels where appropriate.

One of the big challenges right now is getting test data up on the wiki as quickly as possible. Jalexander and Azariv have been working hard on that, so there surely will be a solution soon.

Comments and, more importantly, further improvements welcome! --Eekim 00:57, 4 December 2010 (UTC)[reply]

December 2: More Editor Appeal Testing[edit]

See the stats from today's test

Over the past two days we have continued to roll out editor appeals for testing on English projects for users in the US. On Wednesday, we introduced two different appeals from Joan Gomà (1, 2), and ran them alongside the winning Kartika letter from Monday. We're still working on the data from Wednesday's test, but the longer Gomà letter has out-tested the other two appeals. Today, we added two appeals from User: Lilaroja.

Both the editor and Jimmy banners have a similar click-through rate, however, Jimmy's appeal receives approximately twice the number of donations.

A few takeaways from the first round of editor appeals:
  • The most successful letters have a catchy, short first line. For instance, “It wasn't wrong, but it wasn't right enough.”
  • The length of an appeal matters; longer letters have a higher donor conversion rate.
  • A higher ask in the letter ($20, $35, $50 or whatever you can to keep Wikipedia free) leads to an increased average donation.

As testing continues to determine the best version of each editor's letter, we're putting out community translation requests and building localized landing pages to move the banners out on other projects.

Have feedback, suggestions? -Deniz (WMF) 03:35, 3 December 2010 (UTC)[reply]

December 1: Translations - from Afrikaans to Yiddish[edit]

On November 15th, we launched our annual fundraiser with translations in over 40 languages, surpassing last year's total of 39 localizations. As of today, we've doubled that number. Our messages have been translated in over 80 languages, spanning six continents - making this year's campaign the most inclusive to date.

For 42 languages, including 5 Indian languages, this year is the first time they've had localized versions of the appeal letter on their projects.

Hundreds of community volunteers have been deeply involved in translating the fundraising messages, and we continue to publish localizations daily. Ultimately, our goal is to represent as many of Wikimedia's 270 language projects as possible, and thanks to our dedicated international editors, we are well on our way.

We still need help localizing pages. Don't see your language? Visit our translation hub and help us localize the page. If you see any incorrect or rough translations – let us know and we'll fix it!

As always, tell us what you think, and check back Friday for this week's testing re-cap. - Deniz (WMF) 01:42, 2 December 2010 (UTC)[reply]

November 29: Editor Appeals & New Landing Pages[edit]

Today we ran three versions of a personal appeal written by Wikipedia editor Kartika alongside our control banner. The winning editor appeal continued to run throughout the day to test three variations of our donation form.

Currently, we are running three landing pages - our original two-step payment process, as well as a version with the editor's image, and a one-step donation form. We are aiming to simplify and improve the usability of our donation process by continuing to test and tweak the landing pages.

Tomorrow we will continue to test editor appeals; we'll post detailed test results for both the editor appeal and the landing page tests this week. Until then, share your feedback. - Deniz (WMF) 03:39, 30 November 2010 (UTC)[reply]

November 23: Week One[edit]

The 2010 Fundraiser has had a fantastic first week – to maintain the momentum it kicked off with, we continue to test and tweak our banners and landing pages. Most regular visitors to Wikimedia sites likely noticed we tested 12 variations of the Jimmy banner throughout the week, along with 5 versions of an editor appeal. We tested new images, background colors, changing the wording slightly and adding “Read Now” buttons to the banners.

  • We found that our control banner, which the Huffington Post dubbed the “soft-toned supplicant,” the “periwinkle petitioner” and the black background banners performed best. The click-through rate increased when the “Read Now” buttons were added.
  • We slightly adjusted the ask string by adding three more options. It performed better than the original, which had 4 suggested donations ranging from $20-$250. People like options, lesson learned.
  • On November 13th we tested community-submitted messages against the Jimmy banner in 12 languages, and once again the 2.26% click-through rate of the Jimmy appeal was unmatched by any of the other messages.
  • This week we will be running a second test of the Editor Appeals. Our initial test of Kartika's appeal had a higher click-through rate than our control Jimmy banner, and was the first banner we've tested to “beat Jimmy.”

As always, tell us what you think! -Deniz (WMF) 15:20, 23 November 2010 (UTC)[reply]

November 4: The Schedule[edit]

A memo to Wikimedia community, friends, staff, and other stakeholders.

On Monday, November 15, we will launch the 2010 annual fundraising drive for the Wikimedia Foundation. As you know, our funding model relies on the support of our friends and community members. Our average donation is about $25, and we have received more than 500,000 donations in the lifetime of the foundation. This year, we have to raise $16,000,000. That’s our biggest target yet, but it’s still only a tiny fraction of what the other top-ten websites spend on their operations. It’s critical that we reach our goal to maintain the infrastructure necessary to keep Wikipedia and its sister sites running smoothly.

We are a community that does great things, and does them routinely. As we begin to bring this year's fundraiser to a close, we will launch our 10th Anniversary year! It's hard to believe, isn't it? What would the world be like, if the wiki hadn't launched? If we hadn't jumped in to grow it? If we hadn't financially supported it? The world would be a far different -- and far more sad -- place, I think. This 10th anniversary year provides an opportunity for reflection and introspection, but it also provides a chance to refocus: to plan, to build, to grow. We've just completed the strategic planning initiative, and emerged with a cohesive, defined plan for the future growth and development of the Foundation, the projects, and the movement. Now is the time.

So let's get going.

Since August, a team of dedicated staff members and volunteers has worked to develop the fundraiser for this year. We committed early to radical and full disclosure of all the data we had, in keeping with the spirit of the transparent nature of the Wikimedia movement. We quickly identified three major points in the donation process that were "levers" we could pull to optimize the process: banner messaging, banner design, and landing/donation pages.

Banner messaging: Wikimedia fundraising has always been driven by site notices -- banners -- that run at the top of project websites. We’ve known for years that different banner messages drive different numbers of people to click through and donate. Therefore, this year we began the fundraiser by inviting community members to propose new banner messages for us to test.

Almost 900 people were involved in the creation and discussion of potential banner messages. We tested dozens of iterations of banner designs, including both graphical and text, and we will continue to do so.

Many of the new banners did well. Unfortunately, none of them came anywhere near the 3% clickthrough rate of the winning banner from years past: “Please read: a personal appeal from Wikipedia founder Jimmy Wales.”

But we’re going to keep trying. Our research indicates that banner wins because it is simple and direct with no attempt at marketing or manipulation. So we’re going to test, “A personal appeal from Wikimedia editor _____” and later in this memo, I’m going to invite you to be that editor and write an appeal for us to use in the fundraiser.

Banner design: In our testing this year, we also quickly learned that graphical banners perform almost 100% better than text banners with the same message. Because of this, we will obviously be using more graphic heavy banners than we have in past campaigns.

Landing/donation pages: Once a user clicks a banner, they land on a page that asks for a donation and provides payment options. We have spent a lot of time and energy optimizing those landing pages. Optimization of donation forms is an art and a science that involves messaging, graphic design, and usability research.

We will have iterated through roughly 40 different designs before landing on the ones that we'll launch with. We are committed to encouraging people to beat us at our own game: we invite chapters and affiliated groups, organizations, and Wikimedians to create their own landing pages that they believe will work better than the ones we're running. If we see some that are exciting, we'll test them, and run the ones that perform best!

In countries where there are Wikimedia chapters, the chapter has the option to create their own landing page to test along side the default. We hope that chapters will beat the default everywhere there is an attempt. In countries where there are no chapters, we’d like active Wikimedians to contact us about doing the same thing.

As we proceed through the campaign, we'll be constantly testing. We'll test messages, banners, and landing pages. We'll also test timing, and font size, and hundreds of other small variations. But we're doing it all with an eye to integrity in data analysis, and an understanding of not only what the data tells us, but what it doesn't tell us. Our decisions are grounded in fact and well reasoned theories: not hunches or educated guesses.

One thing is very different this year, though. Once we hit our goal - and we will hit our goal - rather than immediately removing all banners, we're going to use some of the banner space (with a reduced banner size, frequency, and using targeted appeals) to ask people to contribute - not financially, but with their knowledge. We will target readers, and encourage them to become editors. It seems logical to us that this reader conversion effort should flow naturally from our fundraising campaign: both are forms of contribution. We also believe that it will yield financial payoff in years to come by embedding new people deeply into our community and instilling them with our key values and an understanding of the greater mission.

This is an aggressive campaign. It's an entirely achievable goal, however. The only way to have it work, though, is to have full buy-in from the community. Will you reach out to the people near you (either physically or virtually) and ask them to get involved? Tweet that you donated. Write a blog post about it. Deliver four donations from friends with your own. Help new users who make their first edit as part of the contribution campaign.

Here are some key things to know:

  1. On November 15, we will launch the fundraiser.
  2. You will begin to see banners consistently on the sites beginning on Friday, November 12 as we do full scale functional testing.
  3. This is a "contribution" campaign, celebrating all kinds of contribution.
  4. Our numbers are reasonable and attainable, but still a stretch.
  5. There will not be success without the full and active engagement of the community.

We've billed this as "the fundraiser you can edit", and it's true. Community volunteers have been deeply embedded in our planning, including in all of our testing. Community suggested messages were requested and tested. We truly think of this as a fundraiser that is co-created by various parts of the community.

There are still ways that you can participate directly, right now. We’re going to test appeal letters from Wikimedia editors. If you think you can write a letter that will beat Jimmy’s, please go to the meta page and sign up so we know to expect your letter. You can also just send one to me by email: donate(_AT_)wikimedia.org.

I'm honored to be leading the effort this year, and ask you to join with me in making a contribution on the first day of the fundraiser.

If you have any questions or comments, I'd love to hear them. Please tell me what you think by writing to donate(_AT_)wikimedia.org. Philippe (WMF) 00:35, 5 November 2010 (UTC)[reply]

October 28: Payment Methods[edit]

On Thursday we tested PayPal v. Credit Card donations, running one Jimmy banner for two hours with 5 landing pages at 10% each for logged out users on English Wikipedia.

  • Appeal 22: A simplified version of our traditional landing page with most side links and the comment box removed, the lowered ask string, and buttons for both credit card and PayPal payments.
  • Appeal 23: Radio links for payment methods that included major credit cards and PayPal, with one “continue to donate” button.
  • Appeal 24: Had an ask string and separate buttons for both credit card and PayPal, with a text box for email accompanied by an opt-out checkbox.
  • Appeal 25: Had an ask string and gave donors the option to contribute by credit card only.
  • Appeal 26: Identical to Appeal 25, except it only gave donors the option to give through PayPal.


The success of Appeal 22 showed that removing text boxes and any unnecessary text from the form makes a big, positive difference. Predictably, adding an email section to the form reduced donations. Appeals 25/26 were run to test selection bias between the two payment methods, these had a significantly higher click rate than Appeal 22, but when not given payment options, donation abandonments were twice as high as the rate on the other forms. Having both PayPal and credit card payment options is clearly important. Appeal 23's radio payment buttons also showed that too many options reduces donations.


Check out the detailed stats or the summary, & tell us what you think!Deniz (WMF) 20:11, 3 November 2010 (UTC)[reply]

October 26: More Community Banners[edit]

Posted by: Deniz Gültekin
On Tuesday we tested the largest number of banners yet: a total of 53 community proposed and Jimmy banners on English, German, Spanish, French, Italian, Dutch, Swedish and Portuguese Wikipedias. We also ran banners on Italian Wikiquote, and geo-targeted the Portuguese banners for Portugal, Brazil, and users who had their language set to Portuguese.

See all banners | View summary stats | View all stats

We increased the size of the text banners to match the Jimmy Appeal banner, and made them visible to anonymous, non-logged in users only. We ran the banners at 50% each total, except for the German test, which began at 10% and increased to 40% as to test their server strength. We used a significantly cleaned up version of the landing page for the test, removing the public comment box, and a lot of the text around the donation options.


Stay tuned for more updates, and of course, give us your feedback. --Deniz (WMF) 19:58, 28 October 2010 (UTC)[reply]

October 25: More Testing[edit]

Posted by: Philippe Beaudette
As you know, we've done a tremendous amount of testing and optimizing of the fundraising pages during the past few weeks. Over the weekend, we did a lot of analysis of the numbers and what we've learned so far. The biggest finding was that the amount of knowledge we gain from each test is huge, and we really feel like one more round of testing would make a sizable difference in the overall performance of the fundraiser.

With that in mind, we're going to push the launch date back to either Friday the 12th or Monday the 15th.

We believe that we are well positioned for the most successful fundraiser in history. Our hourly totals (during an off-peak hour) during the testing period are higher than the hourly totals of the highest hour of the biggest day of last year's campaign. We honestly believe that pushing the launch back will allow us to get a few more percentage points of performance out of the landing pages and that, when applied to the fundraiser as a whole, could potentially raise the overall totals substantially.

We're aware, of course, that this fundraiser is critical for funding chapter programs over the next year. We want it to be maximally productive, and we think you'll be very pleased with the results this year.

If you have any questions, as always, please feel free to email me (philippe(_AT_)wikimedia.org).

And as always, give us your feedback! Philippe (WMF) 03:59, 26 October 2010 (UTC)[reply]

October 19: Suggested Donations[edit]

Posted by: Deniz Gültekin

Earlier this week we ran one banner at 60% for logged out users, and six landing pages for two hours. The landing pages tested ascending and descending suggested donations or "ask strings," and three variations of the Jimmy appeal letter. For this test we also removed the public comment option and the photo header from the landing pages.

We learned from the October 14th test that the suggested donation of $20, $35, $50, and $100 brings in more donations than the original $35-$250 suggestion; so this week we experimented with a descending donation suggestion starting at $100 versus ascending from $20.

The landing pages had three different versions of the Jimmy letter:

  1. Last year's letter
  2. Alternate text version 1
  3. Alternate text version 2

What did we learn?

  • Appeal 9 was most successful during this test, with a .89% donation rate (number of donations:number of banner clicks) and 251 donations.
  • Landing pages with descending suggested donation amounts had a higher average donation
  • Alternate text version 1 appealed to donors most

We're currently running another test, I'll update you on our findings tomorrow! And as always, give us your feedback!-Deniz (WMF) 00:56, 22 October 2010 (UTC)[reply]

October 14: Landing Pages[edit]

Posted by: Deniz Gültekin

Yesterday we completed our 11th test, running one Jimmy banner at 60% to anonymous users only, with four variations of the donation landing page. This test was to assess what variables on the landing page lead higher donation completion, what caused donors to abandon the transactions, and if lowering the suggested donation threshold affected the average donation. During our two-hour test the banners raised a total of $41,966.

This week's variables included our typical suggested donations of $35, $75, $100 and $250 and a lowered set of donation suggestions of $20, $35, $50 and $100. We experimented with removing the public comment option to see if it had any influence on whether people “abandoned” their donation before completing it, and ran landing pages with and without Jimmy-image mastheads.

  • Landing Page 1: Acted as the control in the test, it featured the Jimmy header, public comment box, and suggested donations that began at $35 and went up to $250.
  • Landing Page 2: Had the Jimmy header, public comment box, and suggested donation ranging from $20-$100
  • Landing Page 3: No header, public comment box and suggested donations of $20-$100
  • Landing Page 4: Had no public comment box, the Jimmy-picture masthead and suggested donations starting at $20 as well.

Landing Page 4 was the most successful landing page, with .98% of individuals who clicked the banner making a donation – bringing in $11,822 in two hours. It was the only page that did not have the “leave a public comment” option. Interestingly, Landing Page 1, the control page used during last year's fundraiser, raised the least, with only .65% of people who landing on the page making a donation - earning $9,262. However, Landing Page 1, which had a suggested donation amount starting at $35, had the highest average donation at $33, on par with last year's average of $34. The lowered suggested donation did bring down the average donation amount by approximately $6, but increased the total number of donations received by more than 100 on each page.

On Tuesday, October 19th we'll be running our next test, running three versions of the Jimmy appeal with new suggested donation amounts and comment options.

Tell us what you think on the Talk Page! --Deniz (WMF) 22:08, 15 October 2010 (UTC)[reply]

October 12: A slew of community banners[edit]

Posted by: Philippe Beaudette

Today we significantly stepped up our testing of community suggested banners - testing almost forty different banners that were community suggested against the control, the "Jimmy" appeal. We ran a mix of community proposed banners and consultant banners, which all had significant community support. We're still putting together the results (you can see them on the spreadsheet) but it's fairly clear that Jimmy's appeal still emerged the favorite.

In addition to the 40 english banners, we also tested German and French banners. The same held true for those projects, we haven't found the magic combination that "beats Jimmy" yet, but we're narrowing in on it!

The message that came in second to Jimmy was the donor quote banner. It had a .25% click-through rate, compared to Jimmy's 3.68%. We would like to experiment further with donor quotes and test more.

We want to test adjusting the layout and design of banners as well; altering size and color, and introducing images and animation into text-only banners.

Check out the full stats, and leave us your thoughts on the talk page! Philippe (WMF) 00:38, 13 October 2010 (UTC)[reply]

October 11: Small changes bring big results![edit]

Posted by: Philippe Beaudette
Today's testing was a little out of the norm: we ran at an abnormal time, and for two hours. We tested a total of 12 banners, all variants on the "Jimmy" theme. Several elements were varied: some were text only, some included a "Donate now" button, some included a full image banner, and some included a photo of Jimmy's face on the left. The full results from the text are on the spreadsheet, but here are some really fascinating things we learned. In some cases, they're logical conclusions, but it's great to have data to support them.

  • Having a "donate now" button increases our conversion rate but decreases our click-through rate.
  • People want us to be polite. Saying "please read" gets us more clicks and more donations.
  • Images are critical. Banners with images do better than text-only banners. We didn't use images last year, so this was a huge learning for us.
  • Having a "donate now" button and not saying "please read" is a total killer. People don't donate.
  • Saying "please read" without a "Donate now" button gets us the most clickthroughs.
  • On the donation page, having a "big blue button" affected performance. The small grey buttons worked better.


Check out the full stats, and leave us your thoughts on the talk page! Philippe (WMF) 04:03, 12 October 2010 (UTC)[reply]

October 8: Jimmy brings down our pages![edit]

Posted by: Deniz
Today we ran six variations of the Jimmy landing page and donation form in order to evaluate which layouts and appeal letters perform best, as well as to test our server strength.

The different versions of the Jimmy landing page were up for 2 hours to test certain elements of each one. We wanted the test to determine which of the following are most successful:

We received much higher traffic than we had anticipated or scaled for – it slowed page loads to a crawl, and brought down our donation pages at one point. Some individuals who clicked on a banner experienced load times as slow as a minute, for many others the page timed out. For 10 minutes, our servers gave up all together and users received a database error page.

What we learned: While the fundraising team didn't get reliable data about landing pages from this test, we did establish that we need to expand our server capacity in order to handle the load from banners as successful as the Jimmy appeal.

Our next test will take place on Monday, October 11th at 20:00 UTC/12:00pm PST. We will be testing different variations of Jimmy banners then to get an idea of the best performing layout and design of banners. Join us on Webex or IRC.

Discuss this entry on the talk page.

September 23: Banner & Appeal Tests[edit]

Posted by: Deniz
This week we ran four separate banners, and tested seven variations of landing pages. We used the Jimmy Appeal banner as a control to test the rest of the messages against, as we know it is our most successful banner. The Jimmy banner was up for a total of 30 minutes, as the text was not clickable in some versions of Firefox and had to be taken down mid-test. However, while only being up for half the test, it had a 5x higher click through rate over it's nearest competitor (Be Curious), and brought in $1,686.56.

  1. Shorter copy on landing pages leads to more donations.
  2. Banners need an ask, without an ask there are no donations.
  3. The Jimmy appeal was the most successful banner by far. It was up for 25-40% of the test, and still received 5x more clicks than its nearest competitor, Be Curious.
This is not an ad. Donate today to make sure you never see one on Wikipedia. landing pages:
  • CC form CC form #6 was the most successful, raising 3X the donations of the CC form #5 & CC form #7.
  • CC form #7, which had the longest copy, had the largest number of total clicks, with only 2 donations and the least amount raised.
Stay curious. Donate today to keep Wikipedia free and strong. landing pages:
Creative

We have to get some new banner templates and landing page templates. Time for community involvement there!

September 16: What We've Learned[edit]

Posted by: Deniz

  1. The "Donate Now" button does not improve click rate, but increases conversion rate.
  2. Banners with an ask lead to the best conversion rate. (ex. banners which explicitly say "donate now" are clicked with the intention of donating; vague banner statements, such as "stay curious" or "for today, and tomorrow" are clicked out of curiosity, and may not convert.)
  3. A soft ask is more effective than a direct ask (supported by you v. donate now)

The banners that did the best fell under the following categories:

Protect
Ideological
Reader Supported

The banners that failed fell under the following categories:

Emotional
Ambiguous/curiosity

Banners to Keep

  • Fifth largest - without donate button
  • This is not an ad. Make sure you never see one on Wikipedia. - try on single click system
  • The sum of human knowldege, for a small sum - try again with corrected quote
  • Sum of knowledge - with donate button
  • Free knowledge for all: make it happen - for Facebook
  • Wikipedia:share - with copy on landing page
  • For today, and tomorrow. Please donate - For logged in users with better landing page copy

Banners to Eliminate

  • Free knowledge for all
  • Wikipedia is there you everday. Be there for Wikipedia today. Donate.
  • Be curious.
  • For today, and tomorrow. Please donate. (eliminate for anonyous)
  • One day people will look back and wonder what it was like not to know

Landing Pages

  • Keeping 7 & 8
  • Got rid of 9
  • 1 could be turned into a letter
  • Try 8 for logged in users
  • Mixed Reviews for 1 & 6