Smart upgrade amounts in donor emails

A newer AI model increased recurring donation upgrades by 8.6%.

What we tested

Recurring donors are one of the most valuable sources of sustainable fundraising revenue. That's why we're continuously looking for ways to help existing supporters increase their ongoing contributions over time.

One opportunity comes through recurring donation emails, where supporters can increase their monthly gift using suggested upgrade amounts. We wanted to understand whether a newer AI model could generate even more effective suggestions than the one already in use.

To test this, we compared our existing AI recommendation model with a newer version designed to generate more personalized upgrade amounts for each donor. The only change was how the suggested amounts were calculated; the email experience and donation flow remained the same.

The goal was simple: determine whether a newer AI model could help more recurring donors increase their support.

Hypothesis

If we replace our existing AI recommendation model with a newer AI model, conversion to recurring gift upgrades will increase by 12%, because more relevant upgrade suggestions make it easier for supporters to increase their monthly contribution.

Test setup

Duration	64 days
Timing	May–July 2024
Audience	Recurring donors with a USD subscription who received an installment email — ~366,000 donors across the test
Group split	• 50% control (suggested upgrade amounts from the existing AI model) • 50% test (suggested upgrade amounts chosen by a newer AI model)
Key metrics	• Conversion from email to recurring upgrade • Conversion from upgrade link to recurring gift upgrade • Average revenue per user (ARPU) from upgrade links

Key results

The newer AI model improved recurring donor upgrades without negatively affecting donor retention.

+8.57% increase in email-to-upgrade conversion
+17.69% increase in upgrade-link-to-upgrade conversion
+26.77% increase in ARPU from upgrade links
No statistically significant change in recurring plan downgrades
No statistically significant change in donor retention

Segment insights

The newer AI model performed well across devices.

+11.76% increase in email-to-upgrade conversion on desktop
+16.03% increase in upgrade-link conversion on desktop
+18.47% increase in upgrade-link conversion on mobile

Conclusion

The hypothesis was partially confirmed.

While the newer AI model did not reach the targeted 12% increase in email-to-upgrade conversion, it consistently outperformed the existing AI model across the most important upgrade metrics.

More supporters upgraded their recurring gifts, and those who upgraded tended to increase their contributions by more, all without increasing downgrades or affecting donor retention.

Based on these results, we're rolling out the newer AI model for recurring gift upgrade recommendations.

What this means for you

Growing recurring revenue isn't just about acquiring new monthly donors, it's also about helping your existing supporters deepen their commitment over time.

This test shows how we're continuously improving every part of the recurring giving experience, including the AI models that power donor recommendations. We regularly test newer approaches against the models already in production, adopting only those that deliver measurable fundraising improvements.

The result is a recurring giving program that's continuously optimized to help more supporters increase their impact, and help your organization build more sustainable revenue.

→ Explore how recurring giving optimizations can grow your sustained support.