3 Types of Predictive Models to Improve Your Fundraising

The use of artificial intelligence (AI) in the nonprofit sector has rapidly expanded—just over a year after these tools became widely available, over half of nonprofits use AI solutions. 

Nonprofits are turning to these solutions because of their tremendous potential to bolster their fundraising campaigns and other efforts. Predictive analytics in particular empower nonprofits to make strategic, data-driven decisions that set their fundraising programs up for long-term success. 

In this guide, we’ve outlined three types of predictive models that can support your nonprofit’s fundraising efforts, pulled from BWF’s fundraising predictive analytics guide. First, we’ll explain how predictive modeling works. Then, as you browse this list, identify the models that apply to your nonprofit’s giving program that will work best for your fundraising strategy. 

How does predictive modeling work?

Predictive AI utilizes data, statistical algorithms, and machine learning to forecast future outcomes based on historical data patterns. Doing so enables informed predictions of events or behaviors and empowers nonprofits to make data-driven decisions, strengthen donor relationships, and enhance fundraising effectiveness, maximizing impact and advancing their mission.

To develop accurate predictive models, you’ll need to use two types of data: 

Training data is added to your models to help inform them, provide necessary context, and build them to your specific needs. For example, if you were creating a model to help predict which mid-level donors have the potential to become major donors, your training data could be information about past mid-level donors who upgraded their giving. 

• Testing data is added to the models to determine the accuracy of their predictions when given new information. Building on the example, you could add data about current mid-level donors to your model to help test it and see whether it generates accurate predictions about who is most likely to become a major donor. 

It’s helpful to clean your nonprofit’s database before starting predictive modeling. This ensures the information you feed into your models is updated, accurate, and robust. We also recommend working with an AI fundraising consultant who can help you leverage best practices when getting started with predictive modeling. 

Let’s look at some models you can create using your nonprofit’s data to help guide your fundraising efforts. 

Giving Behavior Models

Modeling donors’ giving behaviors is an effective way to predict how your audience will engage with your organization in the future based on their past interactions. Here are a few different types of giving behavior models your nonprofit can use: 

• New donor acquisition: Measures the rate at which your nonprofit brings new donors on board.

• Recurring giving: Assesses how many donors become ongoing supporters rather than one-time givers.

• Giving channel preference: Pinpoints donors’ preferred giving methods, whether online, in-person, or via direct mail.

• Likelihood of renewal: Determines how likely donors are to continue giving to your organization year after year.

• Next gift amount: Evaluates how much donors are likely to give the next time they donate.

• Campaign priorities: Defines which specific fundraising campaigns donors are likely to prioritize.

You can personalize your donor communications based on the insights these models provide. For example, you can determine the right channels to engage donors, what gift amount to ask for next, and which projects and programs to emphasize in your fundraising messages. Donors will be more likely to respond positively to your outreach when it speaks to their interests. 

Scoring Models

Scoring helps your organization determine which segments of its overall donor population best suit specific fundraising campaigns or initiatives. The following scoring models can help support your donor prioritization efforts: 

• Engagement scoring: Measures how engaged donors are with your organization through various metrics, including email open rates, event attendance, and social media engagement. 

• RFM segmentation: Helps group donors based on their gifts' recency, frequency, and monetary amount. 

• Estimated giving capacity: Assesses how much donors are capable of giving based on wealth metrics (including stock and real estate ownership, past donation amounts, etc.) and warmth metrics (a personal connection to your cause, a history of involvement as a donor, board member, or volunteer, etc.) 

• Grateful patient scoring: A model used in healthcare fundraising to help determine which former patients are most likely to engage with certain fundraising appeals or campaigns. 

These models help pinpoint the specific donors who will drive different campaigns to success, whether it’s your annual giving campaign or a major capital campaign. As a result, you can focus more of your time and resources on soliciting those highly valuable and engaged supporters, helping to drive a higher return on investment (ROI) for each campaign. 

Giving Program Success Models

Lastly, you can use predictive modeling to forecast outcomes for specific giving programs. These program models include: 

• Principal and major giving: Determine how many principal and major donors your nonprofit will be able to recruit and how much you expect each donor to give. 

• Planned giving: Project the success of your planned giving program based on the number of donors involved and their projected value.

• Annual giving: Assess your annual giving program from multiple angles, such as discovering which donors are most likely to renew or increase their annual giving. 

• Mid-level giving: Evaluate the success of your mid-level donor outreach efforts and your ability to upgrade mid-level donors. 

• Grant giving: Measure your likelihood of winning certain grants and your anticipated share of overall funding from grants. 

Your giving programs drive your organization’s fundraising efforts, and using predictive modeling can help you make the right data-driven decisions that keep costs low while increasing giving. 

How to choose the right predictive models for your needs

With all these options, how can you determine which models will work best for your fundraising efforts? Follow these steps to choose the right predictive tools for your nonprofit’s needs: 

1. Assess your fundraising goals. Start by outlining what you hope to achieve from the predictive modeling process. Be as specific as possible when setting these goals, which could include increasing annual giving, preparing for a capital campaign, expanding your legacy giving program, and other aims. 

2. Determine what kinds of data you have available. What supporter information do you already have at your disposal? You’ll need access to comprehensive, clean data to train and test your models. Start by auditing your donor database and identifying potential gaps to fill with audience research or data appends. 

3. Reach out to a fundraising consultant. Developing accurate predictive models can be challenging, so we recommend working with an experienced fundraising consultant to clean your data and construct your models. These professionals can help determine which models are most useful for your fundraising needs and ensure you have the right strategies in place to make the most of the insights you gather. 

4. Ensure your models are scalable and sustainable. Work with your consultant to ensure your models can grow with your organization. Ensure your modeling processes integrate seamlessly with your other data-gathering and processing tools, such as your CRM or marketing platform. This helps increase your models’ long-term quality.

These steps will help your organization move forward with the solid foundation it needs to make predictive modeling a core component of your fundraising strategy.