Peer-to-peer experiment

Initial research

As described in more detail in our shallow review of peer-to-peer fundraising, external research on this method is weak. Peer-to-peer (P2P) is a relatively new technique and, although a few charities have mastered it (Charity Water), there is little data analysis even with the low standards of general fundraising research. Our initial foray into this field came in the form of the Charity Science walk, and was based primarily on recommendations that charities run events like this, rather than evidence favoring peer-to-peer fundraising in general. We were inspired to explore it further when birthday fundraisers run by our co-founders, Joey and Katherine, raised significant amounts of money and when others started to get similar results.

Methodology

As noted above, the Charity Science walk was as much an experiment in running an event like this as it was an exploration of peer-to-peer fundraising, and its distinct methodology is covered elsewhere. Our typical peer-to-peer fundraisers have been run online. We carefully investigated the best software to use and decided that we needed to prioritise attractiveness and ease of use, deciding to use CauseVox (though Classy also had many strengths, particularly in the US). We paid for a low-fee account with CauseVox and directed most people towards it (with deworming being the default cause, and other GiveWell recommendations selectable), but also offered AMF's DIY fundraising facility as an alternative option that was more appropriate for some people. We individually contacted large numbers of people in the runup to their birthdays and Christmas, and spread the idea of peer-to-peer fundraising through the effective altruist community. The Life You Can Save later contacted us saying that they were keen to help with this, though other organizations were not; most promotion in 2014 was done by the core Charity Science team.

Our individual experiments in peer-to-peer fundraising steadily got better returns, and as the field looked more and more promising, we decided that investing further resources in it was justified. We followed up our Christmas fundraisers with another sponsored fundraiser, which, unlike the Charity Science walk, was international and allowed lone individuals to participate. We called it Experience Poverty, though this name prompted significant pushback and we later started referring to the fundraiser as Living On Less. The event involved participants getting sponsored to spend only a certain amount of money on food for a certain number of days; in the US our default suggestion was to eat on less than $2.50 a day for three days. We used this opportunity to experiment with several different things including: 

• Allowing people to seek sponsorship for a demanding sacrifice
• Signing people up via local effective altruist groups, employing the network of these that one of our staff members had built up, in addition to our personal networks as employed in previous fundraisers
  
• Making initial attempts to recruits fundraiser outside the effective altruist community (although we ended up with little time to spare on this 
• Investing more staff time in a particular peer-to-peer fundraiser

Hypothesis

In experimenting with peer-to-peer we wanted to test a number of hypotheses (not all of which we expected to be true). As described above different experiments tested different approaches, but between them they gave us evidence on the following questions:

• How much money could we raise via peer-to-peer fundraising? Money moved remains our primary metric.
• Could we spread this beyond the effective altruist community, with which we have particularly close connections?
• Could we get local effective altruist groups to raise money?
• Do matching funds increase money raised?
• Do seeding funds increase money raised?
• Which sorts of peer-to-peer events work best?

Results

The cost all all P2P events including staff pay (but not counterfactual time) was about $6000 about $125,000 was raised not including any matching, seeding and excluding some but not all counterfactual donations (this is about $350 an hour). Our top fundraisers made a huge difference in how successful the campaign was. The average amount raised for each event ranged from $160 to $240, with the average of all fundraisers being $200. Out of 300 participants in total, about 40% of all money raised came from the top 3 fundraisers. Our average fundraiser raised 60% less if they have done another P2P campaign in the past. 

Conclusions

This was our most successful fundraiser to date and is extremely promising to scale with higher net worth individuals, people outside the EA community, and with other activities that don't involve directly donating immediately. Our most successful fundraisers were often well off, older than average individuals, and raising from their work peers and friends. This makes us lean towards a workplace P2P event and workplace giving in general. We have contacted and talked to these bright spots a little bit and plan to talk to them more so that we can determine how to reach out to similar individuals.

We are also find that we are less confident about donation matching than we'd previously written when we cited matching as a major factor in our peer-to-peer fundraiser success. After running a few more campaigns and comparing the matching and non-matching campaigns, we have found no difference between the two. The spike we got near the end of the campaign which we attributed to matching beforehand also occurred during our campaigns where we did not match anything. Additionally, some external research also suggested matching was not as strong as conventional funding wisdom would suggest.

You can learn more about lessons we learned from running peer-to-peer fundraisers here.