Fundraising 2012/How Wikimedia revenue grows
This was written in May 2012 by Zack Exley, the Wikimedia Foundation's Chief Revenue Officer at the time
One of the WMF fundraising team's duties is to report to the WMF board and Wikimedia movement our estimation of short term and long term fundraising potential. The fundraiser has changed a lot over the past few years. We're raising a lot more money now. A lot of people might not understand how the new fundraiser works and where the recent big gains have been coming from -- and may understandably be developing the false impression that we can grow as fast as we want to.
Below, I want to explain the WMF fundraising team's view of the future of revenue growth. Not all of this is backed up by hard data. In some cases, it is nearly impossible to get hard data. For example, when we rely on surveys, interviews, and group discussions to understand donors, the results are tainted by Self-selection bias. Therefore, sometimes our advice is based on a combination of hard and soft data plus gut instinct shaped by thousands of interactions with donors and active testing against our whole donor base.
What determines how much money we can raise?Edit
Our revenue is primarily a function of the size of our readership, their potential willingness to donate, their ability to donate, and the power of our banners, landing pages and donation processes to inspire and facilitate their donations.
We can’t change the size of our readership or its ability to donate. We can only change our banners, appeals and donation process. We know that there are a lot of people who would donate if they would only click on the banners and hear what we say in the appeal, but who don’t click, because they are not moved by the banners. And we know there are a lot of people clicking, who don’t donate, but who would if we had an appeal that spoke to them.
Here’s an example of the first kind (this is a Wikipedia user who participated in a video user test session on Usertesting.com):
She is an example of why we’re always trying to find ways to improve the banners. But over the past two years, we have only discovered two major improvements to banners: adding an image (obvious) and adding a green, leafy background to the images (not so obvious). We’ve tested hundreds of different messages and images, with only two successes. Each of those two successes, however, is responsible for about a 30% increase in donations. Our successes improving appeals have been more frequent.
Outside of the fundraiser, we only have two hours of testing available to us per week and can only test during seven or eight months of the year. Therefore, we discover most of our improvements during the fundraiser itself.
This creates a big problem for predicting how much we can grow the fundraiser each year: We can only count on the improvements that we discovered the previous year. We can not predict what improvements we’ll discover in the upcoming fundraiser.
For the past few years, we’ve made big promises without knowing where exactly the gains would come from. We were confident that there was enough low hanging fruit left such as adding pictures to banners and simplifying forms. Now, however, there are almost no obvious gains left to make. I believe that we will be able to continue to make large increases in our fundraising every year. But we will probably have a year or two over the next decade in which we fail to find big improvements. Those will be years when our revenue is almost flat, only increasing (or decreasing) in proportion to the factors we don’t control: our readership size and its ability to donate.
What I’d like to propose is that the Wikimedia movement transition to a revenue growth model that only counts on the revenue increases that were locked in in the previous year, and then uses each fundraiser to discover the following year’s increases. (Note that below, when I talk about revenue goals, I'm referring to movement wide goals that add together revenue from WMF, and the three chapters that are payment processing: Britain, France, Germany and Switzerland.)
A total $45,000,000 (USD) Wikimedia revenue goal this year would be a step halfway toward that arrangement. We think that, with hard work, creativity and luck, the movement can probably raise that amount without harming long term revenue. As I'll explain below, however, raising that much money requires making some creative new discoveries that we can't be sure will come. If they don't come, then to get to $45 million, we'll have to extend the fundraiser significantly or run a second campaign in the spring or summer.
A $40 million goal would not require big new discoveries (just a few little ones), but we would try to achieve them anyways, to give the movement a greater range of options for 2013. A $50 million goal would require bigger discoveries than we can promise. Therefore we would have to start the fundraiser much earlier or schedule a second campaign in the spring or summer.
What are the major innovations that have increased Wikimedia revenue?Edit
Each year, a few big innovations were responsible for almost all of our growth. Here they are:
2009: A. Ran the "Personal Appeal" from Jimmy. It was the first effective, direct pitch. The phrase "Please Read: A personal appeal from..." had a click rate that was miles ahead of anything else, and also attracted donors in particular.
2010: A. Ran the personal appeal from the very beginning of the campaign, instead of "working up to it." B. Added a picture of Jimmy to the personal appeal banner, increasing click rate by about 1/3rd. C. AB tested appeals, dramatically increasing our donation rate. And Jimmy wrote a version in his own words -- Temple of the mind! -- which increased donation rate even further. D. Radically simplified the donation forms, increasing donation rate significantly.
2011: A. Worked with editors, WMF staff and donors to create appeals mostly in their own words that performed much better than 2010's editor appeals. B. Began accepting donations in nearly every currency and many new payment methods. C. Added a few lines to the Jimmy appeal that made a big difference in donation rate. D. Added green leafy backgrounds to banner photos. This -- believe it or not -- increased the donation rate by about 1/3rd.
Can’t the fundraising team at WMF and chapters just keep discovering more innovations like that?Edit
Here's the hard truth about the new stage of innovation we're now in: In 2011, if we hadn't stumbled upon innovations C and D, we would have missed our goal significantly. Some of the payment processing chapters probably would have missed their goals by even more, because they relied primarily on WMF-made banners and appeals for their income, and they were already pushing close to maximum capacity by running longer campaigns and relying more heavily on Jimmy.
This is what I really want the movement to understand: Stumbling across both of those innovations was just dumb luck. OK, maybe we can call it “smart luck” actually, because intentionally created a creative and testing program designed to find those kinds of lucky innovations.
But there was no guarantee that we'd find any! In 2011, there were no obvious breakthroughs left such as adding an image to a banner or simplifying the forms left to implement.
As we got closer to the fundraiser, I felt sure that we were going to miss the goal by a few million dollars. I figured that we would have to run a second fundraiser in the spring. We had tested hundreds of new things, but none of them gave us significant gains. And then, just a few weeks before launch, we discovered that someone photographed in front of a leafy green background did 30% better than in front of the white background (which had beaten every other background we had ever tested).
So how much more can we raise in 2012?Edit
I think I’ve explained the problem we face in 2012: To significantly increase our revenue, we need more significant breakthroughs in our banners or landing pages. But we have no idea what they might be. And we only get one or two shots per week to look for them between now and mid-November. When the fundraiser begins, then we’ll test five or six appeals per day and as many banners as we can come up with. I have faith that through that process, we’ll discover some big increases. But there is no guarantee that we will. Recall that of hundreds of banners created by our team, the Wikimedia community and professional designers and ad people, none came anywhere near the sparsely designed “Please Read: A Personal Appeal...” banner.
I have faith that we will find significant improvements this year, and most of the following years into the foreseeable future. I’m just saying (pleading) that we shouldn’t count our chickens before they hatch.
Can't we just increase the length of our fundraiser to raise more money?Edit
Adding days to the fundraiser does raise more money, but the marginal increase produced by each day is very small. This would only change if we broke the fundraiser into two parts and ran two separate 2-4 week campaigns. But we don’t know how well that would work yet. Another possibility would be to run banners at a low level year round. I’m not advocating for either of those strategies, and there has been significant resistance to any deviation from the once-a-year model from many members of the community, chapter leaders and WMF staff.
Here’s why adding days to the fundraiser doesn’t work: When you add days, you’re adding the worst days. For WMF, those are $150,000 days. If we have to guarantee an additional $5 million, and couldn't count on any magic new appeal or banner being discovered, then we'd need to start the fundraiser about 33 days earlier than last year, or October 14th. And even then, we can't be sure that the $150K days would hold -- they could go even lower for all we know.
We face this limit because almost all of our potential donors visit Wikipedia over the course of a 5-6 week fundraiser. Most donors give within the first few times they see the banners. Another portion wait, intentionally or subconsciously, for the very end of the campaign. Therefore, when you extend the fundraiser, you’re not reaching significantly more donors.
Therefore, when you lengthen the fundraiser, you are only adding that small number of donors who are both (A) likely to donate and (B) visit the site very rarely. After the first month of the fundraiser, we are scraping the bottom of the barrel and relying on that small pool of donors. We generally panic for a few days then. Then, we break out the end-of-year messaging which unlocks that 11th Hour donor pool.
This is why I’m arguing that the way to increase our revenue is to create more powerful banners and appeals, not add days to the fundraiser.
The reason I say that splitting the campaign in two would create more revenue is because I think many donors would experience that as two separate campaigns, and two different occasions to give. This might work to raise more money -- we don’t know yet. But it might also erode long-term revenue potential and reader goodwill if we become known as “one of those organizations that’s always asking for money.”
One more reason to be cautious about counting chickensEdit
One more reason to refrain from counting on big gains before we learn how to make them is this: The larger our budget grows, the higher the proportion of potential donors who don’t like how big it sounds.
It’s very difficult to figure out what non-donors think. It’s even hard to figure out what donors think. We can survey people on our sites, but the sample is radically skewed toward people who feel positively toward and are interested in us -- because that’s who clicks on banners, donates and agree to take surveys to help us. (We will eventually do some kind of general-population telephone surveying, but that will be very expensive.)
Nevertheless, among the people who donate and answer surveys, there are some troubling data: When asked “If you had to take a wild guess, how big do you think Wikipedia's annual budget is? (In dollars.)” only 6.1% answered “Tens of millions”. And we know from many of our group discussions and user tests that most of those people are probably thinking 10 or at most 20 million. (This survey was given on May 16, 2012, only to U.S. donors. If anyone is skeptical of this number, we can repeat the survey and repeat it in different countries.)
When we tell them that our budget for this year will be around $45 million, many accept that and trust that we need the money and will spend it wisely.
But many others are shocked in a very negative way. This probably explains why in 2010 a progress meter running up to our $16 million goal increased donations, but that a meter running up to $25 million did not seem to help in 2011. I have to assume that many of the donors who made the comments listed below would not have donated if they knew our budget, without any further explanation.
I believe that it’s possible to convince these donors that our budget is justified, but most donors are not going to give us the chance to do that. We only get a few seconds of their attention and if we’re going to raise our budget, we need to use those seconds to inspire a donation, not explain budget details. (Nevertheless, we are going to put a lot of effort this year into finding better ways of explaining Wikimedia budgets and programs.)
Here are some of the comments in response to “Wikipedia's budget is $45 million per year. Please tell us what you think of that.”
"much more than I thought"
"I'm surprised it's that high! I'm surprised you're able to receive enough donations. I am also surprised that the budget is that high. I'd be curious as to where that figure comes from."
"seems like alot to me"
"sounds like a lot, my compnay budget is half of that for 400 plus emplyees"
"It can only go up and that worries me."
"I am surprised, i thought Wikipedia managed to operate off of A LOT less, but I still realize the majority of Internet based companies operate off of budgets that would make $45mil seem like pennies."
"It makes me wonder what the employees make."
"It's higher than I would have guessed with just 95 employees. Either the employees are making a lot more than I guessed (I guessed $80K average) or the servers cost a lot more than I guessed"
"Spend it wisely"
"Damn gimme some!!! I thought only 5 million"
"Holy crap !!! That's 10 times more money than I guessed ! How do you manage to raise that every year, especially at start-up? How is the money spent (salaries, technology, marketing, finance, real estate, legal, etc.)? Amazing that wikipedia is still a free service started by a simple vision !!!"
On the other hand, many were positive:
"Good for you!"
"It sounds very high, however I have nothing to compare it to as I do not know everything that goes into running something so large."
"I'm inpressed; that's not a lot considering how significant a website you have and how popular it is among webusers."
"Fine. I assumed less money and more nerd info fan boy types rather than professionals."
"When I think about it, your international support (and gratitude) that figure shouldn't surprise me."
Would knowing our budget is $45 million make people less willing to donate? We asked and the answers were mixed about evenly or neutral. (I asked for plain text answers, so it’s not easy to count. I wanted to get a sense of how people would answer and will ask multiple choice in the next survey.) But I don’t know if self reporting is that accurate in this situation where donations are often made on a split-second, emotional basis.
This response may be more representative than the data suggest, as it may reflect the thoughts many have in that split-second decision:
“Less likely. 45 million is a lot compared to a household budget and it makes my individual donation seem less useful knowing how small the effect will be.”
I’m not trying to make a strong argument about how our donors view our budget. I’m just arguing for caution and for the principle of waiting until we’ve shown we can raise more money before counting on too much more money.
Wikimedia projects are funded by an incredibly unusual revenue model. We are showing that a top website, a universal educational resource, a central pillar of the world's information infrastructure, can thrive without selling its soul and independence to private investors or advertisers. This revenue model is one of the things that ensures our ability to serve our sites for free with all content under a free license.
The fundraiser is also turning into something that educates Wikimedia users about how our projects and movement work. Tens of millions of users click on banners and read the appeals every year. The primary mission of the WMF fundraising team is now to improve that educational function. We believe that we can discover efficiency gains each year that allow us not only to keep increasing the movement's revenue as needed, but to also focus the banners and appeals more on teaching users about our movement. As seen in the video example I gave at the top of this page, many users fall in love when they finally learn what we are. Imagine future campaigns that focus on recruiting new editors more than new donors. That's what we're trying to get to.