Chances are good if you are in the fundraising field that you have heard the term “fundraising analytics.” You’ve probably also heard the terms “data mining,” “donor modeling,” “reporting” and “prospect identification,” too. Do these terms mean the same thing? What are the differences among them?

I asked Marianne Pelletier, who leads the HBG Analytics team, to help me put together a series of short articles designed to make sense of these terms. In each, she will describe the method and give examples of how they can be used. To continue our series, we describe the questions that Donor Modeling can answer.

Let’s begin with a case study:

A museum is in the planning stage to launch a major fundraising campaign. Their last campaign was over 5 years ago, and while they had a number of significant gifts, the coming campaign will require many more major gifts in order to be successful. After developing a table of gifts for the campaign, it quickly becomes apparent that there are huge gaps that need to be filled with prospects at every level. Significant prospect identification needs to happen.

To score the museum’s database and identify the top prospects, the museum decides to use a technique called predictive modeling, also referred to commonly as donor modeling.

What is donor modeling?

Donor modeling uses statistics tools to score a group of records using a variety of methods, including regression analysis, clustering, decision trees, neural networks and support vector machines (SVMs) amongst lots of others. Let’s take a look at just one of them, regression analysis.

Regression analysis uses calculus to find the slope of a line, which helps us visualize trends in the data. For example, we could see (based on a number of factors) which groups of people in the museum’s database have the most capacity to give as well as affinity, or connection to, the museum.

Here’s a standard matrix that is often built for major gifts programs. After downloading records and using regression analysis to score the group studied, prospects would be shown along the slope of the red line based on their relative affinity with the museum and their capacity to make a major gift.

Affinity, or “how much they love the museum” might be measured by the number of times someone attended events, or donated in consecutive years, or bought tickets to special exhibits, amongst other things. Capacity, or “how much they can give” might be found through primary or secondary research, such as a visit, prospect research or an electronic screening.

A graphic describing the relative level of a group of prospects’ affinity using a number of hearts (ranked on a scale of 1 to 3) and the relative level of their  gift capacity (ranked 1 to 3) by dollar signs might look something like this:

In this example, the top-right box represents those with greatest capacity and affinity for the organization, and the bottom-left box shows those with the least.

If you were the chief development officer at the museum, whom would you want to approach first? Your answer is likely to be those in the top right group. Unfortunately most of the time that group is also the smallest population among the scored groups, and are usually the donors you know fairly well.

Whom to select next, then? Often, two of the largest groups, represented by the larger boxes, are the $$$ ♥♥ and the $$♥♥♥ groups.  And of those, it might be hard to decide which to choose.

So, your next donor modeling study might be to look at the museum’s past track record with each of these two groups. What is your level of success in cultivating each group? What motivates them to become major gift donors?

Donor modeling helps answer those questions. The characteristics of top level donors are compared to various segments of the pool, and their scores help bubble up the best future prospects.

What else can you use donor modeling for?

Although it’s most often used to identify major gifts prospects, donor modeling can also rank groups like these:

• Annual giving prospects who are most likely to renew
• People who are likely to be good board/volunteer candidates
• Planned giving prospects
• People who would be great prospects for a specific project or campaign (like a library fund, or for endowment)
• People who would be most likely to accept a request for a visit
• Top level annual giving prospects
• Prospects best suited for a particular gift officer or volunteer

Donor modeling can even help determine the best ways to acquire new members for a member recruitment campaign. It’s a powerful tool to help your organization identify new donors, whether you’re in a campaign, thinking about a campaign, or just looking for new donor prospects.

What do you need to know?

Our series on the ABCs of fundraising analytics continues next Thursday, September 19 with a look at data visualization.

Do you have questions about donor modeling or would you like to see it at work at your organization? Contact us at info [at] helenbrowngroup [dot] com for more information.