Did you know you can increase your donation revenue by up to 15% with a single click? Optimizing your donation form for recurring monthly donations can help create a sustainable revenue stream, as well as more predictable monthly income.
Recurring donors are the backbone of sustainable revenue for nonprofits. To grow this essential base, you not only need to have the recurring option ready on your donation form, but also need to ask the right audience and suggest the right amount.
That’s where AI can make all the difference.
At Fundraise Up, we use AI to analyze donor behavior in real time and tailor the checkout experience to each supporter.
The challenge: how to show the right frequency for better results?
We’re always looking for ways to improve the donation flow and grow revenue for nonprofits. As part of our strategy, we optimize for both conversion and average revenue per donor, striking the right balance to maximize impact.
This time, we focused on a key question: which gift frequency should be shown first by default?
When a donor opens Checkout, they’ll typically see either a one-time or recurring gift option pre-selected, based on how the nonprofit configures it.
If your goal is to grow long-term revenue, it might seem obvious to lead with the monthly option. But what happens if the donor isn’t ready to commit?
Some donors do welcome the suggestion to give monthly. Others feel pressured and end up abandoning the one-time donation they initially planned to make.
We needed a smarter solution – one that shows the monthly option first only to donors likely to say yes or for larger gifts where one-time brings in more revenue, while avoiding it for those at risk of dropping off.
Identifying donor segments
Instead of a one-size-fits-all approach, we needed to understand different donor behavior types. As a result, we identified four categories:
| Donor behavior type | Behavior |
|---|---|
| Upgraders | Plans to give once, but will switch to monthly if prompted. ✅ Ideal audience for monthly ask |
| Engaged | Will give monthly regardless. Prompt makes no difference. |
| One-time donors | Will give once if not distracted, but may leave if pushed to subscribe. ❌ Avoid showing recurring option |
| Unengaged | Unlikely to give at all during this session. |
Our goal was to identify the Upgraders in real time and prompt them with a monthly ask, without disrupting the One-Timers. Easy in theory. Tough in practice.
So, we built an AI model to do just that.
The solution: AI personalization
We trained an AI model to predict, in real time, which donor type a visitor is likely to be, and personalize the donation experience accordingly.
To validate the impact, we ran a 26-day A/B test with 600,000 visitors:
- Control group: Saw the standard frequency prompt.
- Test group: Saw the AI-optimized default based on behavior prediction.
This large-scale experiment with a robust sample size gave us statistically significant insights across a diverse audience – something smaller-scale tests with just tens of thousands of participants can’t reliably deliver.
The results
- +26.76% increase in recurring donation conversion
- +4.12% lift in average revenue per user (ARPU)
- No drop in one-time donations
In short: more recurring revenue, without sacrificing short-term gifts.
A look under the hood: the technology we used
Rather than relying on standard machine learning approaches like classification or regression, we adopted Uplift Modeling: a method from the field of Causal Inference focused on estimating treatment effects.
While standard models predict whether a user will convert, uplift models estimate the incremental impact of an intervention – like showing the monthly screen – on the user’s behavior.
This approach helped us identify who is likely to convert because of the frequency default, not just who is likely to convert in general.
Our stack
Most models predict the probability of a donation itself, not “who” is likely to donate.
Uplift modeling asks a different, far more useful question that focuses on causality: “Will showing the monthly-first screen cause this person to upgrade, and by how much?”
The R-learner learns baseline behavior and treatment effect during training, but at inference directly predicts the causal uplift for each visitor.
The payoff: monthly prompts reach only the persuadable donors, one-time givers feel zero pressure, and nonprofits capture new recurring revenue without sacrificing short-term gifts.
One click to smarter giving
This experiment shows what’s possible when you stop treating all donors the same.
With AI-driven personalization, nonprofits can move beyond static donation forms and create experiences that meet supporters where they are, leading to more sustainable funding for your mission.
Ready to see what AI-optimized fundraising looks like? Request a demo or learn how AI powers better giving experiences.