Research
2026 holiday shopping trends: The impact of holiday ads, AI use case shifts, the impact of rising costs and more.
Learn more
For years ‘big data’ has been talked about and utilized in a variety of ways. The most common methods in which non-profits leverage big data is to identify prospects for new donor acquisition and to model housefile donors to determine seasonality, ask and offer. It’s important to note that all these predictions are made possible through the power of machine learning algorithms.
While we have always viewed mid-level and recurring donors as valuable, prospecting these high-value donors (whether on or offline), still poses a challenge to direct response fundraisers. With today’s machine learning capabilities, data is available—at our fingertips—to enable us to make behavioral predictions. For marketers who leverage database cooperatives, machine learning algorithms expose a vast amount of data that can now be utilized in ways that have never been possible. For example, if you’re trying to grow your sustainer or mid-level program, in a standard regression model, there may be thirty to fifty elements that weigh heavily to make the prediction. With machine learning, there are thousands. Finding the difference between one potential donor to another, across thirty variables is good, but across thousands of variables, there’s likely to be an even greater distinction introducing new levels of success. As one of our data scientists frequently says,“Where regression model is algebra, machine learning is calculus.”
While almost any model can help re-activate a previous donor or identify a new donor prospect, machine learning enhances that capability by using more data elements to make the requested prediction. Another benefit of machine learning is predictive analytics. Machine learning helps to break down the complexity of data-sets within seconds. And with the incorporation of seasonality, machine learning models help to Identify the timing and amount of a donor’s contribution based on a contributor’s transaction history leveraging seasonality, donor demographics, and so much more.
Every cause has an opportunity to grow, especially when you know more about your donors. With machine learning, marketers gain insight into thousands of donor characteristics. This unique knowledge can be used to drive donor acquisition, conversion, retention and upgrades to help grow donor engagement and value. In addition to gaining insight into these thousands of characteristics, machine learning helps non-profits to:
Identify new, unique audiences. In the non-profit community there’s a lot of concern about over-communicating to prospects and donors. For example, the age segment of 65+ receives one-hundred different non-profit packages in their mailbox each week. Identifying new unique audiences enables non-profits to find new constituencies and to ensure proper cadence of communications.
Improve performance. Over the past several years, many non-profits have witnessed increasing costs to acquire a donor and have reported a perception of a shrinking prospect universe. With machine learning models, we can identify highly responsive, unique pockets of prospect audiences. Additionally, machine learning models can be leveraged to identify unique pockets of high-value prospects resulting in the same decreased net cost per donor yielded through increased merge retention and higher revenue generation. Regarding the housefile, machine learning models can drive variance in the identification and subsequent cadence of lapsed reinstatement initiatives driving a larger number of lapsed donors to reinstate their giving, ultimately driving an increase in overall reinstatement long-term donor value.
Utilize every element of data. As marketers are integrating machine learning models into their programs, it’s reassuring to know there’s so much available data to leverage. For example, as you’re receiving repeat donations, take a look at the month or season of the donation, what spend looks like outside of the philanthropic cause that’s being supported, the types of apparel the donors purchase for attending events, if they have children, grandchildren—the list goes on. Before you know it, you’ll have ten-thousand pieces of information that you can model and identify to see if ‘Joe Smith’ should be in segment number one or ten for direct mail; and in digital activation it can more effectively refine those large audiences that are necessary to hit your reach metrics. This vast amount of data is going to help to identify unique audiences.
At Epsilon, we continue to innovate our data offerings and have incorporated machine learning techniques to take our modeling solutions to the next level. We work with our non-profit partners to help:
To learn more about how non-profits are securing the second donation, listen to our webinar with our client of the ASPCA titled Speed to second gift: the metric that will make your retention soar.
