A conservation research nonprofit that sent extra volunteer recruitment emails saw more people take part in a citizen science event and more donate in a later campaign, according to a randomized field experiment involving 149,480 people.
In the group offered additional invitations, 0.41% donated during the subsequent membership appeal, compared with 0.31% in the comparison group. The difference was about one-tenth of a percentage point, or roughly 32% relative to the lower rate.
The finding concerns one organization’s campaign and should not be treated as a fundraising forecast for every nonprofit. It also does not mean that 32% of recipients donated.
How the experiment worked
Researchers Sachin Gupta, Sungjin Kim and Anirban Mukherjee describe the 2021 experiment in Marketing Letters. The paper was published in May 2026.
Participants had recently connected with the organization but had no prior record of volunteering or donating to it. Everyone received 21 baseline emails promoting a multi-day citizen science event, in which volunteers submitted observations through an app or website.
Random assignment placed about half the participants in a group that also received three extra recruitment emails. Those messages included incentives such as free online courses. The recruitment emails did not ask for money.
Seven weeks after the event, the organization invited both groups to become paying members. Researchers compared participation in the event and the later fundraising campaign.
More people gave, but gifts were not clearly larger
Volunteer participation rose from 5.72% in the comparison group to 6.29% in the group receiving the additional emails. That is an increase of 0.57 percentage points, or about 10% relative to the comparison rate.
The later donation rate was 0.31% versus 0.41%. Among people who donated, the average contribution was $45.70 in the comparison group and $44.60 in the extra-email group. The researchers did not find a statistically significant difference in average gift size.
Random assignment strengthens the evidence that the additional recruitment campaign changed outcomes for these groups. The authors also use a statistical method to investigate whether volunteering itself affected later giving. That estimate depends on assumptions, including that the extra emails did not affect donations through a separate route.
Why the distinction matters
People who volunteer may already be more interested in a cause than people who do not. Simply comparing those groups could therefore exaggerate the effect of volunteering itself.
The randomized invitation helps address that problem by comparing groups created before the event. It measures the effect of an encouragement campaign, including its messages and incentives, rather than assigning each person to volunteer.
The authors propose several possible explanations for why participation and later giving might be connected, but did not establish the psychological mechanism. Nor did the study test whether sending more email is always helpful. Both groups already received a substantial number of invitations.
What nonprofit teams can test
A fundraising team could examine whether a useful, voluntary activity introduces prospective supporters to its work. A fair test would compare similar audiences, track both participation and later donations, and account for the cost of delivering the activity and contacting supporters.
Organizations should also consider whether participants want follow-up messages and whether different forms of involvement suit their time and circumstances. A community-science task for an environmental organization may not translate to a local food bank or an arts charity.
For another perspective on measuring outcomes, see MBN’s report on a field experiment that matched small businesses with marketing mentors. Both studies show the value of comparing outcomes with a suitable control group.