The Donor Tipping A/B Test

Growth and experimentation inside payments · Neon One, ~2021

Problem

The donation flow monetized through a "cover the fees" checkbox. The open question: would an optional donor "tip" model out-earn it? Both sides had plausible intuitions. Intuition wasn't the right tool.

My role

Owned the experiment design, the read, and the rollout decision.

Outcome

Tipping out-earned "cover the fees." Rolled out platform-wide as a durable revenue driver for every customer; the losing donation-form version was sunset.

Situation

"Cover the fees" was the incumbent, familiar to donors across the nonprofit space. Tipping was gaining ground elsewhere but felt riskier — would donors tip a platform on top of a charitable gift? This was a monetization change touching every donation form on a platform processing $200M a year. Guessing wrong at that scale is expensive in both directions: shipping a worse model, or leaving a better one unshipped.

The decision

Run the experiment properly. We A/B tested the two models across donor cohorts and measured the true effect on transaction volume and revenue rather than debating hypotheticals. Just as important as running the test was committing in advance to act on the result — including sunsetting the losing form version rather than keeping both alive as a configuration option. Optionality is a tax; the point of an experiment is to let you stop paying it.

Outcome

The tipping model won on revenue. We rolled it out platform-wide and retired the losing variant. Every customer on the platform got the benefit of the experiment, and the business got a monetization model chosen on evidence.

What it shows

Growth mechanics decided by experimentation rather than taste — and the discipline to ship the winner everywhere and delete the loser, which is the half of A/B testing most teams skip.