Randomized Testing for Online Fundraising Appeals

Michael Giberson

Following up yesterday’s note on randomized testing in free legal aid, here is another kind of applied experimental work: The recently ended Wikipedia fundraising campaign made extensive use of randomized testing to explore just which appeals generated the most revenue. “If everyone reading this donated $5” vs. “If everyone reading this donated $10” (about the same). “Only 4 days left” vs. “4 days left” (“Only” seemed to increase donations slightly).  Many variations tried.  Links and some data here.

HT to Al Roth at Market Design.