Wikipedia Fundraising Design Needs[edit]

The Wikimedia Foundation is the non-profit organization that hosts Wikipedia and other websites that are created and maintained by hundreds of thousands of volunteer editors. We provide technical, legal, administrative and other support as a foundation on which our volunteers editors work.

We raise almost our entire budget in an annual fundraiser that you’ve probably noticed. We display eye catching visual messages in banners (like the gold banner above) on the top of Wikipedia articles asking readers to donate.

The primary goal of the fundraiser, however, is not to raise money, but rather to educate Wikipedia readers about how who creates Wikipedia and how it operates. We think thousands of brilliant people collaborating to “make the sum of all human knowledge available to every human being” is a pretty beautiful thing, and we want the world to know about it.

Each year that we’ve run the fundraiser, we’ve gotten much better at telling our story while also becoming more more efficient at raising money. But we want to do even better. We want to catch the attention of more readers while making the donation process as simple and easy as possible.

We need a special kind of designer to help us achieve that goal. Our banners are not flashy or beautiful -- and we've learned from testing that they shouldn't be (though we are always happy to test old assumptions with the right designs). But they need to look nice and professionally designed -- in other words, they should not look like 1998. In our experience, most talented designers tend to get very bored making small variations to our plain and simple banners. Amateur designers (e.g. someone who knows how to use photoshop but doesn't have a serious background in art and design) sometimes don't have the problem of getting bored, but they have the problem of not being able to make new designs look good.

We are looking to connect with designers who can make beautiful designs, even when they're extremely simple designs -- designers who will enjoy learning everything they can about our past tests, what's worked and what hasn't, and use their own creativity and curiosity to come up with new designs to test. At the bottom of this page you'll see a banner we recently tested that made a simple change: move the form from the expanded banner right into the banner itself. That's a relatively small design change, but in one test it improved our donation efficiency by about 10%. What are additional changes to that design we should make? How can use space more efficiently? How can we use color or effects to make it look nicer? Those are the kinds of design changes we're looking for. We're also excited to test radical departures -- but we have limited time with which to test, so we need the radical departures to come after you've gained experience with our problem space.

Take a look below at a few examples of the evolution of our banners, some ideas we'd like to test, and send us a message if this looks like a project you'd be interested in working on: megan@wikimedia.org

Banner Testing Background[edit]

Over the past few years, we've dramatically improved the efficiency of our fundraising banners. We run A/B tests of banner messages and designs to find the best-performing banners. In 2009, we started with text-only banners, but we quickly learned that our "personal" appeal" banners with a photo of our founder performed about 50% better.

Banners from 2010

Lessons learned:

The "Personal appeal" banner beat out the other one-liner slogans. We've tests hundreds of slogans created by ourselves, community members, marketing and advertising experts. It turns out that our readers don't respond to slogans -- they love and are curious about Wikipedia and want to learn about it. The "personal appeal" promises a chance to learn more about Wikipedia.
We got a huge increase in clicks and donations by adding a photo of our founder.
The fancy graphic backgrounds on the Jimmy banner did not do any better than the plain background Jimmy banner.
Banners do not need to be heavily stylized to be effective. In fact, we think our readers tend to ignore our banners if they look like typical online ads.
Include a call to action ("Read now"/"Please help" button).
Adding "Donate" button lowers click rate and lowers total money raised. (But of course increases the donations/clicks rate.)
We think the plain HTML "Please help" button performed better than the more designed buttons, but we haven't tested very many variations yet.

Banners from 2011-2012

The fundraiser had a very different look in 2012 compared to previous years. Prior to 2012, our highest earning (and most widely recognized) banner included a photo of Wikipedia founder Jimmy Wales asking people to click to read his personal appeal. In 2012, we experimented with a new approach by featuring key Wikipedia facts directly in the banner. This new "facts banner" performed much better than any banner we had run before – in some tests, bringing in about three times as many donations as the "personal appeal" banner.

Lessons learned:

The facts banner beat any other banner we'd ever run, even the Jimmy "personal appeal" banner. And it beat it by up to 300%.
The facts banner partly works by eliminating a major obstacle in our donation process. By placing the key information directly in the banner at the top of Wikipedia articles, we eliminated the need to click on the banner to read our fundraising pitch which increased donations. This seems like an obvious innovation, but we tried many combinations of designs and text that did not beat the Jimmy banner before we finally found a winning combination.
Including the form (amount options and payment methods) directly in the banner performed better than having to click on the banner to reach the form.
Adding a side tab with a donate button once you scroll down the page increased donations. (See an example with the side tab.)
Highlighting "If everyone reading this gave $3, our fundraiser would be done within an hour" increased donations.
Adding "If everyone reading this gave $3, our fundraiser would be done within an hour" to the side tab made a huge difference.
Yes, that long paragraph really does better than a short snappy slogan. Turns out people who read an online encyclopedia like lots of facts and information. We've tested hundreds of variations of the banner text, and we found that people actually do read the text and are persuaded to donate when presented with the right combination of honest facts. Wikipedia readers do not donate to marketing or gimmicky slogans.
We've tried hundreds of variations on this paragraph, constantly finding small improvements by changing a phrase or word. We think there are still big improvements to be discovered in the text, but finding them is getting difficult for us.
Bright or intense colors performed better than softer colors.
Having the text together in a single paragraph performed better than breaking the text up into multiple sections or columns.
We've run lots of tests on the ask amounts in the radio buttons. These lower ask amounts work best with the facts banner.
Adding large donation amounts (e.g. $500, $1000) doesn't seem to increase the rate of large donations. If it does, it's only by a tiny imperceptible amount.
Adding $5 or even $3 increases the number of donations by a lot, but also reduces the average donation. In the end it usually makes more money. Even when it doesn't make more money, if the difference isn't great, we prefer to get more donations.
The banner took up less screen space than our previous "personal appeal" banners and ran at the very top of the page – out of the way of the article space. We do not want fundraising banners to interfere with the Wikipedia reader experience.
The info "i" makes a significant difference.
The facts banner has the additional advantage of educating ALL Wikipedia readers (not just the select people who click on our messages) about who creates Wikipedia and how it operates.

Our Current Best Performing Banner

Next Steps[edit]

We only show readers a few banners each year, so the amount of tests we can run is limited. We have a few ideas we'd like to try out, but we're not professional designers so we are also are seeking new test ideas from designers around the world.

A few ideas we'd like to try:

Small form when you click on the side tab (rather than the entire banner re-expanding).
Having just one line of the banner ("If everyone reading this gave $3...") stick at the top of the page when a reader scrolls down the article.
New banner placement. We've tested lots of banners at the top of the page, but want to experiment a bit with new design for banners on different parts of the page.
New graphical elements to get more people to click.
All new design ideas we haven't tested or even thought of yet.

A few details about our testing process:

When testing a new idea, we always try to keep all variables constant, except for the one new test variation. For example, if we are testing a new line of text, we keep the color, size, font, graphical elements, etc. constant between the test banner and the control banner.
We could be much more aggressive with our banners, but we try to keep the annoyance factor to a minimum to not disrupt the reading experience. We don't use popups or animations.
We conserve space as much as possible. Each year, we manage to make our banners smaller, yet more effective. We want to continue to decrease the page space our banners occupy by making them as small as possible.
The facts are important! Turns out people who read an online encyclopedia like lots of facts and information. We've tested hundreds of variations of the banner text, and we found that people actually do read the text and are persuaded to donate when presented with the right combination of honest facts. Wikipedia readers do not donate to marketing or gimmicky slogans.
The text we have has been tested against hundreds of variations. If you want to change the text, your best bet is to just change it a little bit. We think there are improvements still to be found. But before you suggest a new text, please look at some of the hundreds of other texts that have lost.
We only show readers a few banners each year, so the amount of tests we can run is limited.

Additional Background[edit]

We publish updates with our test results. For more background info on our testing, check out: