Between Christmas and the end of January, Alex Gordon-Brown and I ran a donation match for AMF. We decided to triple (Gift aid included) and raised almost 14,000 pounds.
Our motivation besides raising money that wouldn’t have been donated otherwise was to find people who haven’t been involved in the EA community so far. We hoped we could try to draw in newcomers and get them to donate, in the hope they’d donate more in the future. For those reasons, for evaluating the impact of our match, we aren’t looking too closely at the money we have raised officially (not even taking into account the obvious concern that some of those donations would have happened anyway).
Unfortunately, most of these benefits are difficult to assess. People who donated seemed to mainly come from EA-affiliated circles with some people being “core EAs” and some people being more losely connected. We also had a few donations by people whom we suspect to be Alex’ coworkers. This flags up a neglected benefit of Earning-to-Give-careers - more access to people who are able to donate large sums too. Those donations were also the highest we got, which highlights this benefit further.
There were a few donations in the thousands, a decent chunk in the hundreds and many below a hundred pounds.
We decided the amount based on how much we thought was achievable, but still high enough so it would bring out higher donations. So far, donation matches were capped at 4-figure sums*, while the largest donation we got was already more than 5,000 dollars. We hoped to anchor people with a high target. This seems to have worked. A smaller goal probably wouldn’t have brought out as many >$1000 donations.
Ben Kuhn mentioned in his post that matching a donation more than 1:1 doesn’t seem to be worth it (this post appeared after we started the match). Since we wouldn’t have gone for a higher amount even if we had done only 1:1 matching, there wasn’t a cost to tripling. [Edit: Ben pointed out in the comments that this result was based on a direct mailing campaign, so it's unclear how well it translates to Social Media fundraising.]
On the cost side in general, we put a couple of hours in, but very few overall. The donation match was shared on Facebook and we had a few private conversations because of it, but again, very few. The money we put up for the donation match would have been donated anyway to AMF.
The other cost was the match not filling up, though the effects of that are unclear. It might demotivate people in the future if there’s a match with a similarly high goal. I think this is offset by the experimental value the match had.
Most donations happened in the first and the last week of the match, that’s also when we tried to draw attention to it the most. We could probably have done the match over a shorter timeframe, so donations would have been more regular. This might have resulted in people being more motivated to donate, compared to when it looks like it’s come to a halt.
Another reason behind the timing was the intention for the match to be Christmas related, since people are more likely to donate during that period. We were a bit late for that unfortunately, so in the future (combined with the reason above), we’d rather do a match only for December. If you have any suggestions how to improve (our) donation matches in the future, we'd be happy to hear them.
No worries. I probably should have titled the post "Did Donation Matching Work in Two Direct Mail Campaigns in Minnesota?", but then nobody would have read it.
Hmm. Maybe you're right. At any rate, you're certainly correct that it's not unusually disingenuous.
This is something I'd be interested to see research on--is the average donor aware that the money would have been donated anyway? I suspect the answer to this is "no," at least for some people, since matching campaigns frequently say something like "double your impact" which implies that the money would not be donated otherwise. I also suspect that the people who don't know this would be put off if they found out.
(It's worth noting that some matches do involve the donor influencing the matching funds; for instance, CFAR's December 2013 fundraiser explicitly stated that the last donors had promised to donate elsewhere if the match wasn't filled, and HCEA's December 2013 fundraiser allowed donors to choose between GiveWell top charities and split the matching funds the same way.)
If people knew that most matches weren't actually matching, we'd expect saying "I plan to donate £60k to AMF, please join me!" to have about the same effect as donation matching (unless for some reason people enjoy playing the game where everyone acts like the individual donors are moving double/quadruple the amount they donate). That doesn't seem likely to me, but intuitions are pretty frequently wrong about this kind of thing.
A few random observations:
(A) The final level of the match probably does increase our giving to AMF relative to other places, albeit not by the full amount (my back of the envelope is that each £1 given to the match increased our giving to AMF by £0.5), That's not dissimilar to the structure of the CFAR match.
(B) Most people, even among my heavily selected friends, don't naturally think in terms of counterfactuals where altruism is involved. Seriously. When conversations I had with people did touch on counterfactuals, the assumption was that the money woul... (read more)