Fundraising 2013

From Meta, a Wikimedia project coordination wiki
Fundraising 2013 Noticeboard
Shortcut:
FR2013

This noticeboard is for announcements, updates and testing news related to fundraising from staff at the Wikimedia Foundation.

Get involved:


2013 Campaign Wrap-up Update[edit]

Happy New Year! We successfully wrapped up the fundraiser in December, big thanks to everyone whose hard work made this possible.

We're working on a full report to share, but in the mean-time, here are a few quick facts to hold you over:

  • In December, the fundraising team ran the year-end online campaign in the US, UK, Canada, Australia, and New Zealand.
  • The campaign ran full blast for two weeks.
  • Roughly $18.7 million USD was raised from more than one million donors in December (preliminary numbers as donations are still settling).
  • Fiscal year-to-date (i.e. since 1 July 2013) approximately $32 million has been raised from online fundraising banners.
  • The team created and tested approximately 250 different banners.
  • We sent 2.6 million emails to donors from previous years asking them to give again.
  • We responded to 15,000 reader emails during the month of December.

A full report will be posted publicly in 2014. We're currently preparing to run campaigns in all other countries and languages. The multilingual campaigns will be spread out throughout 2014. Heads up: We will need help translating messages and localizing the fundraiser. If you're interested in helping out, please let us know

Thanks for all the support and keep an eye out for more updates.

Meganhernandez (talk) 13:06, 10 January 2014 (UTC)[reply]

Fundraiser Update December 16, 2013[edit]

On December 16, we started limiting the number of banners we are displaying to readers. If you continue to see banners more than just one more time (on the same device), please let us know on the meta talk page

Donations are still coming in and settling in our accounts, so donation totals are preliminary and will be adjusting in the next few weeks. So far this fiscal year, we've raised (very roughly) $27 million from nearly 2 million small dollar donors. We will raise the remaining $13 million now through the end of the fiscal year in June 2014. The team is planning to put English banners up again at a higher level around December 31 for a final year-end push. All of the multilingual campaigns will run in 2014.

We will post a full report of the fundraiser along with test reports from hundreds of A/B banner tests in 2014 once we get through reconciliation and analysis.

Thank you very much to everyone involved in the fundraiser this year. It's still not over quite yet, but we've had a very successful and exciting couple of weeks.

Meganhernandez (talk) 00:40, 18 December 2013 (UTC)[reply]

Campaign Launch Update December 2, 2013[edit]

The fundraising team is kicking off our year-end campaign in the US, UK, Canada, Australia, and New Zealand at midnight UTC on December 3, 2013.

Quick reminder: We have been running banners at a low-level since the start of the fiscal year in July. In that time, we have continuously tested new banners to find improvements that will allow us to reduce the number of banners people see each year. This is our first year experimenting with year-round fundraising. We've already raised over $13 million USD -- which is roughly equal to about how much we raised after our first full week of the 2012 fundraising campaign. During this year-end campaign, banners will be displayed to 100% of readers.

A few key facts:

  • Amount raised in testing prior to the year-end campaign: Roughly $13.5 million USD
  • Campaign goal: $20 million USD
  • Banners will run to 100% of English readers in the US, UK, Canada, Australia, and New Zealand
  • Previous donors will receive a reminder to donate email during the month of December
  • We will run campaigns in other countries and languages in 2014

If you spot any errors, please let us know on the meta talk page. Feel free to ask questions there as well.

Thanks for your help! Meganhernandez (talk) 22:17, 2 December 2013 (UTC)[reply]

October Update[edit]

Quick highlights from October

  • Approximately $2.7 million raised from roughly 200,000 donors (preliminary numbers as donations are still settling)
  • Banners ran worldwide to 13% of readers for one impression per reader
  • Roughly 13,000 fundraising emails were sent to past donors
  • Ran short tests with mobile fundraising banners

Test highlight: New Grey Banner

On the fundraising team, we're always working to minimize the annoyance of our banners while maximizing their effectiveness in bringing in donations. Even if a new banner doesn't "win" in terms of bringing in more donations, it's still a big win for our team when we find a banner that performs just as well as our best banner while limiting distraction from the Wikipedia reading experience. We know our readers are here to get the information they're looking for -- we do not want to take away from that experience in any way.

We've tested new variations of a light grey background banner. Our gold banner still came out a bit ahead, but it wasn't a huge loss for the grey banner. We're working on some new variations of lighter background banners for upcoming tests.

The banners:

Gold banner
Grey banner

Results:

Results

Mobile test highlight

This is the first year the fundraising team is experimenting with accepting mobile donations. As more people read Wikipedia on mobile devices, our ability to accept mobile donations becomes more important. We're still in the beginning stages of mobile testing (with plenty of room for improvement), but this is a big step for the fundraising program.

Even with our desktop banners, we always try to maximize the use of the screen space to squeeze as much information into the banners as possible. This is an even greater challenge on tiny mobile screens. In a recent experiment, we tested banner text to see if it was better to include a salutation to readers or examples of our costs.

The banners:

Salutation mobile banner
Costs mobile banner

Results:

Mobile test results

Significantly more readers donate when the banner starts with "Dear Readers." That's a valuable piece of information to learn. Mobile fundraising opens up all sorts of new testing possibilities. We have plenty left to improve in terms of mobile messaging, design, payment methods, and user experience. Lots of lessons to learn here, keep an eye out for future updates.

Upcoming fundraising plans

The fundraising team will continue to run tests throughout November while preparing for the December launch of the English fundraising campaign. We are planning to put banners up the first week of December at 100% to anonymous readers in the US, Canada, UK, Australia and New Zealand. We will run campaigns in other countries and languages in 2014.

To briefly recap this year so far, banners have been up at a low level worldwide since the beginning of the current fiscal year on July 1, 2013. This does not mean that readers have been seeing banners all the time since July. We have our banners set to show to each reader just one time. This testing has been valuable to improve our banners while also to reach more readers. In 2012, we were able to shorten the fundraiser down to nine full days, the shortest fundraiser we've had. That's great. But we know that there are plenty of people who use Wikipedia and would be happy to donate who didn't happen to visit Wikipedia in those nine days that the banners were running. We started running banners in July to reach more people outside a of campaign that lasts just a few days. This year, we have the goal of raising the budget while showing readers fewer banners than previous years. We think this new schedule of running banners throughout the year will help us reach that goal.

Meganhernandez (talk) 16:26, 7 November 2013 (UTC)[reply]

September Update[edit]

Summer's over and we picked up the fundraising pace in September. Banner testing continued throughout September and we also ran our first email tests to past donors for this fiscal year. Emails were only sent to a relatively small batch of donors (roughly 13,000 people). In past years, we've waited until the end of the year to send all our emails. This year we're switching it up and starting earlier to improve our deliverability rates and to run more content tests throughout the year. We've started with subject line, email text, and ask amount testing in our first month of sending emails.

A few quick facts from September:

  • Approximately USD $2 million raised from roughly 150,000 donors in the month of September (preliminary numbers as donations are still settling)
  • Banners ran worldwide to 5% of readers for one impression per reader
  • Roughly 13,000 fundraising emails were sent to past donors
  • Year to date, we've raised approximately USD $7.5 million. In the same time period in 2012, we had raised roughly USD $1 million. (These totals are not including large gifts $5,000+ and we were not experimenting with year-round fundraising in 2012.)

Latest test highlight: 2012 banner vs 2013 banner!

Over the past three months, we've tested a few big changes (added the form directly in the banner) and a ton of small changes (text tweaks, highlight variations, spacing adjustments, etc). Our current best performing banner has evolved quite a bit. So to get an idea of the combined effect of all these changes, we pinned our current best banner from September against our best banner from the December 2012 campaign.

The banners

December 2012 banner
September 2013 banner

Results:

+---------------------------+-----------+--------+-------+-------+-------+---------+---------+----------+----------+--------+-----------+----------+-----------+
|           banner          | donations | clicks | ccpct | cccnv | ppcnv |  amount | amount5 | amount20 | amount50 |  max   |    avg    |  avg20   |   stdev   |
+---------------------------+-----------+--------+-------+-------+-------+---------+---------+----------+----------+--------+-----------+----------+-----------+
|   B13_0915_2012Dec_enYY   |    244    |  356   |  34%  |  56%  |  80%  | 3112.94 | 1088.73 | 2304.07  | 2874.95  | 158.66 | 12.757951 | 9.442910 | 16.834088 |
| B13_0915_2013Sept_dr_enYY |    374    |  579   |  27%  |  49%  |  80%  | 5373.01 | 1658.26 | 3535.81  | 4726.44  | 300.00 | 14.366337 | 9.454037 | 22.822161 |
+---------------------------+-----------+--------+-------+-------+-------+---------+---------+----------+----------+--------+-----------+----------+-----------+

That's a very encouraging ~ 50% increase in number of donations! For October, we're planning to repeat some past tests to get more comparisons to our performance in 2012. In the next month, we will also be ramping up the percentage of readers who see banners and sending more emails to past donors. And, of course, there are plenty of new banner ideas to test as well.

Meganhernandez (talk) 16:20, 3 October 2013 (UTC)[reply]

July & August Update[edit]

We're two months into our new fiscal year and our new method of running banners at a low level (showing to 5% of readers worldwide for one impression per reader). A few key facts from July and August:

  • Approximately USD $4.7 million raised from roughly 350,000 donors
  • More than 40 A/B tests run of different fundraising messages, banner designs, translations, and payment methods

New Highlight Test

Before we placed the donation form directly in the banner, readers used to have to click on the banner to reach the form. One way we got people to click, was by making one of our most powerful lines in the message link blue: "If everyone reading this gave $3, our fundraiser would be over within an hour." Once we put the form in the banner, we realized it didn't make sense to have that sentence link blue (since readers no longer need to click on the banner). But we also knew that the "If everyone..." line was very compelling and that we wanted it to stand out to readers. In a recent test, we tried a new way to highlight the "If everyone..." line and had pretty promising results:

Highlight Test Round 1

Blue link banner
Highlighted banner
banner donations clicks ccpct cccnv ppcnv amount amount5 amount20 amount50 max avg avg20 stdev
blue link 1179 1879 26% 47% 76% $15,708.38 $5,173.00 $10,653.47 $13,910.35 $500.00 $13.32 $9.04 23.235085
cyan highlight 1425 2234 27% 48% 77% $17,075.16 $6,172.82 $12,367.02 $15,762.39 $200.00 $11.98 $8.68 16.811178

Highlight Test Round 2

We saw an encouraging increase in number of donations by highlighting the "If everyone.." sentence. But was that cyan color too bright or annoying? We'r currently testing another variation with a light yellow highlight, and it looks it's performing about the same as the cyan banner. (This test is still in progress. We'll post the final results once the test ends.)

Cyan Highlighted banner
Yellow Highlighted banner
banner donations clicks ccpct cccnv ppcnv amount amount5 amount20 amount50 max avg avg20 stdev
cyan highlight 4337 7127 27% 46% 75% $53,469.29 $18,879.22 $37,645.40 $48,305.58 $500.00 $12.33 $8.68 19.774109
yellow highlight 4418 7226 27% 46% 75% $55,361.24 $19,279.91 $38,776.10 $49,925.49 $500.00 $12.53 $8.78 20.453388

What's next

We're heading into September with plenty more test ideas. We're also increasing the percentage the banners are running (from showing to 5% of readers to showing to 10% of readers). The banners are still only set to show to each reader one time. As always, we're looking for help spotting errors. If you see more than one banner on the same device, spot an error in a translation, or come across any error at all please let us know on the meta talk page.

Mhernandez (talk) 13:48, 2 September 2013 (UTC)[reply]

First two weeks of the new fiscal year: July 15, 2013 Update[edit]

Banners have been up at a low level (showing to 5% of readers for just 1 impression) for two weeks. Where are we at? So far, we've received roughly 130,000 donations totaling about USD $1.8 million. Daily total donations are going down each day; we need more time to figure out where donations will level off. Overall, things have felt pretty smooth for the team. We’ve caught a few errors, have been making improvements, and running tests each day.

How to help: We are looking for feedback! Our banner hiding setup is still pretty new to us. The banners are set to show to each reader just one time. If you read Wikipedia anonymously and see more than 1 banner impression, please let us know on the meta talk page. Banners are also now set to NOT show up to readers who have cookies disabled. If you have cookies disabled but still see a banner, PLEASE tell us. We also have a bunch of tests up with different payment methods, currencies, and translations. If you spot an error anywhere, please tell us the error and which country you are located and we will follow up with more questions.

Tests

In these first days of banners running, we have run about 20 different tests of banner messages, design, translations, and new payment methods. Some tests are still running since we have not received enough donations to get a large enough sample for the results to be statistically significant. I’ll post the results below of just a couple of tests that we’ve completed.

Form vs no-form test

Back in May, we first ran a short test of a banner with a form and we had some pretty encouraging results. We brought the form back again for a test, and saw an increase in donations once again.

Form banner
No-form banner
banner donations clicks ccpct cccnv ppcnv amount amount5 amount20 amount50 max avg avg20 stdev
Form 30768 45970 32% 51% 78% 409546.50 137624.66 291007.49 374763.04 759.15 13.310794 9.458122 19.008618
Noform 28145 39919 38% 57% 83% 383271.73 127404.56 273390.08 349347.47 1000.00 13.617756 9.713629 19.807021

HOWEVER, for that form test we actually changed our control banner in a couple of ways to hold variables constant between the form and no-form banners. The new form banner didn’t have the Amazon or monthly donation options we usually offer to donors. We removed these options from our control banner so we could isolate the form vs no-form variable. Next, we tested the form banner against our old control banner that included monthly recurring and Amazon options and our control banner came out ahead.

Next step here is to test a form banner that also includes the Amazon and monthly donation options. We have a design and are working on getting this setup right now. High hopes for this new banner!

New English text test

We’ve switched up our banner text a little and have seen a boost in donations from the change. The text tweaks are a fusion of ideas that came out of conversations we've had with donors. The new text has an extra "ad-free" mention, says we "survive on donations" rather than "run on donations," and highlights that "now is the time we ask." We’ll keep using this new “fusion” text going forward.

banner donations clicks ccpct cccnv ppcnv amount amount5 amount20 amount50 max avg avg20 stdev
Cntrl 28554 41786 34% 53% 80% 384555.71 128619.74 274357.62 351478.11 1000.00 13.467665 9.608378 19.807512
Fsion 30359 44103 35% 55% 80% 408262.52 136409.48 290039.95 372632.40 759.15 13.447825 9.553673 18.998346
Control Text
New Text

We'll keep testing and monitoring daily totals. It feels pretty good to start of the new fiscal year. As usual, our readers have been sending in messages since we've started running banners again. Here's a quote for you from a reader: "If knowledge is power, then Wikipedia is a nuclear reactor." More updates to come! Meganhernandez (talk) 16:49, 16 July 2013 (UTC)[reply]

July 1, 2013[edit]

Happy New Year! Today is the first day of the 2013-2014 fiscal year. We have some changes in the timeline for running banners and plenty of new ideas to test.

A quick recap from last year: The 2012-2013 fiscal year was the first year we broke up the fundraising campaign. In past years, we had always run banners all over the world in different languages during a year-end campaign in November and December. Last year, we broke that up and ran banners in English speaking countries at the end of the year, and ran banners in multiple languages in the spring. Last year was also the first year we started experimenting with limiting the number of banner impressions readers see. For a recap with plenty of details from last year's fundraiser, please see the 2012 Fundraiser Report.

This year, we want to continue to test more in these new areas of (1) spreading the campaign out from just a couple weeks at the end of the year and (2) with limiting the number of banners readers see. We're starting today by putting banners up at a low percentage (showing to about 5% of anonymous readers for only 1 impression per reader). Whenever we have banners running, we always run a test to find more effective banners, messages and donation flows. During this first test, we'll be running an English message test, banner design test, and several different tests of message translations in multiple languages. We'll post another update once we have some results to share.

Why are we starting in July? With the launch of the new fiscal year, we are starting up our experiments with spreading donations out more throughout the year. We are starting at a low-level (~5% of readers will see just 1 banner impression). As we get results and figure out daily averages, we can adjust the traffic. But it's a good place to start for now.

A few reasons why we're excited about this:

  • We're aiming to reach more readers throughout the year (not just people who visit in the few weeks that we run banners at the end of the year).
  • We can get into the fun part of the fundraiser (A/B testing, finding issues, testing tech improvements live, learning about our readers, etc).
  • Instead of being a mad scramble at the end of the year, we get to ramp up consistently and make improvements year-round.
  • The fundraising team can work & improve with more consistent roles throughout the year.

How you can help: We are looking for feedback! Our banner hiding setup is still pretty new to us. The banners are set to show to each reader just one time. If you read Wikipedia anonymously and see more than 1 banner impression, please let us know on the meta talk page. We also have a bunch of tests up with different payment methods, currencies, and translations. If you spot an error anywhere, please tell us the error and which country you are located and we will follow up with more questions.

More info to come. Here's to the 2013-2014 fiscal year! Meganhernandez (talk) 16:11, 1 July 2013 (UTC)[reply]

Form Banner Test: May 19, 2013[edit]

We tested a new banner that has the donation form directly in the banner. The idea here was to eliminate a click required to complete the donation process. With our current control banner, readers must click on the banner before selecting an amount and payment method. With the new banner, we're hoping to increase donations by shortening the donation process.

The Banners:

Potential Pros:

  • Eliminating one click to donate & shorting the donation flow (we know that extra clicks and steps usually hurt conversion)
  • Quicker/easier to start the donation process
  • Possibly more noticeable banner design with the amounts in a box

Potential Cons:

  • Too much of an ask right up front (maybe donors need the extra text on the dropdown page to motivate them to give)
  • The form could distract readers away from reading the banner text (which we know does the work of convincing readers to give)
  • Too busy of a design layout

The Results:

It looks like the new form banner has about an 8% increase in donations, great news! This is our first test of the form banner this year, there are plenty of other variables to test here to optimize this concept (different design, displaying fewer amounts in the banner, adding the recurring option to the banner, etc.) Encouraging results.

banner donations clicks ccpct cccnv ppcnv amount amount5 amount20 amount50 max avg avg20 stdev
B13_form_form_dr_en 1365 2157 32% 47% 76% 17179.51 6056.18 12310.90 15632.81 300.00 12.585722 9.018974 18.251252
B13_form_noform_dr_en 1255 1760 38% 57% 84% 16704.96 5690.39 11916.70 14976.91 250.00 13.310725 9.495378 19.372926

d/bi confidence test, clicks/bi confidence test, Wilcox test of means: p= 0.004175

Mhernandez (talk) 11:00, 20 May 2013 (UTC)[reply]

April Update[edit]

Thank you for all of the helpful comments that were sent in while banners were running! The fundraising team ramped up testing in March in various languages and countries that were not included in the year-end 2012 campaign. Approximatley $5 million USD was raised from this campaign in the month of March. Banners were taken down on March 31, 2013. We are currently analyzing the data and will publish a report with the donation info from money raised in the English year-end campaign in 2012 as well as from 2013. Meganhernandez (talk) 14:12, 3 April 2013 (UTC)[reply]

Tech Update: March 29, 2013[edit]

The fundraising technology team is excited to announce that we're moving to Limn for public data presentation. We aren't quite there yet but we have done the first step -- we've commissioned a server that will serve sanitized data to the public. Its available at samarium.wikimedia.org. We will be opening up more data as time allows too.

Mwalker (WMF) (talk) 23:14, 29 March 2013 (UTC)[reply]

February 27, 2013[edit]

We have another update on fundraising testing. We have been testing at low levels for almost three weeks. We are preparing to increase the percentage for a 24-hour test so all anonymous readers in the countries and languages where we are running campaigns will see a banner (just 1 banner impression). If a reader has already seen one banner over the past three weeks of testing, they should not see another banner during the 24-hour test. The test is set to start at 18:00 UTC on Wednesday, February 27 and will run worldwide, excluding the countries where banners already ran in December 2012 (US, UK, Canada, Australia, New Zealand, Germany, France, Switzerland).

Again, hiding banners from readers who have already seen a banner is new for us and our goal is to minimize the number of banners our users see. If you read Wikipedia anonymously and see more than one banner, please let us know on the fundraising meta talk page: http://meta.wikimedia.org/wiki/Talk:Fundraising_2013

We are also testing out new translations and donation pages in many many countries, so if you see any errors or have any feedback, it'd really help us out if you'd send it to us: http://meta.wikimedia.org/wiki/Talk:Fundraising_2013

Thanks everyone,

Meganhernandez (talk)

February 6, 2013[edit]

Hi all,

We wanted to let you know the fundraising team is starting up testing in February. We're starting at low levels, so most people will not even notice banners to start with. 5% of anonymous users will see a banner just one time. We are not showing any banners to logged in users.

There was an announcement in November about us splitting up the fundraiser this year. Just a very quick recap: We ran the end-of-year campaign in November and December in the top 5 English-speaking countries (US, UK, Canada, Australia, and New Zealand). The campaign was successful and we were able to take the banners down a few weeks ahead of schedule: http://wikimediafoundation.org/wiki/Press_releases/Wikimedia_Foundation_raises_25_million_in_2012_fundraiser

We are now working on starting up testing in countries that were not included in the end-of-year campaign. We'll be working on translations and optimizing our donation pages in many countries over the coming months.

We will post a report of the year-end campaign with much more detailed information and will send a note to this mailing list when it's available.

We always need help making improvements in different countries and languages. If you have any suggestions, please do leave us a note on the fundraising meta discussion page: http://meta.wikimedia.org/wiki/Talk:Fundraising_2013

Thanks!

Meganhernandez (talk)