Machine Learning for Nonprofits: The Essential Guide

Dec, 2020

By DonorSearch

Machine learning is a term we’ve all seen splashed across headlines and built throughout the scripts of our favorite TV shows and movies. Across sectors, machine learning is expected to grow at an annual rate of 44% through 2022, so it’s a buzzword that won’t be disappearing anytime soon.

But what is machine learning, and why is it important?

You may know what it is in theory, but are not exactly sure what it can be used for. Or it may even be one of those concepts that you pretend you understand but still feel in the dark about. Even if you do consider yourself a machine learning expert, you likely know more about the for-profit applications of it than the impact it can make on the nonprofit sector.

However, nonprofit organizations can also leverage machine learning technology to make better decisions, streamline daily operations, and focus their fundraising efforts.

At DonorSearch, we know that a solid foundation of data is critical to fundraising success. Each day, thousands of nonprofit organizations rely on our informative database to make key decisions. Recently, we’ve seen more and more of those nonprofits turn to machine learning technology as a way to elevate their use of data and make an even bigger impact on their mission.

To help you unlock some of these benefits, we’ve created this guide that covers the ins and outs of machine learning for nonprofits. We’ll cover the following topics:

Machine Learning: Overview and FAQ
Benefits of Machine Learning for Nonprofits
How to Fundraise With Machine Learning: 3 Use Cases

Machine learning is a powerful emerging trend for the nonprofit sector, but it’s still young. When surveyed, 89% of nonprofit professionals agreed that artificial intelligence and machine learning would improve their fundraising efforts, yet only 15% are using it. With this guide, you’ll be better prepared to implement a key tool at the forefront of fundraising technology and make a bigger impact on your mission. Let’s jump in.

Machine learning for nonprofits isn't as complicated to understand as you may think.

Machine Learning: Overview and FAQ

When you hear the term “machine learning,” you may be picturing science fiction movies or a robot army set to rise up against mankind. Despite the images the term may conjure, machine learning is far from fiction. In fact, you likely rely on it far more often than you realize!

Machine learning refers to dynamic computer algorithms that are able to process and analyze large amounts of data to improve performance, accuracy, and insights over time. Machine learning algorithms work by discovering and making sense of patterns. The more data that is fed into a machine learning algorithm, the smarter it will become—hence why it’s called machine learning.

Every time you ask Siri a question or click play on a movie recommendation from Netflix, you’re reaping the rewards of machine learning. However, as we’ll explore in the next section, machine learning has applications beyond the needs of the big tech world.

Frequently Asked Questions About Machine Learning for Nonprofits

What’s the difference between artificial intelligence and machine learning?

Often, artificial intelligence and machine learning are used synonymously, but they have a few key differences. Artificial intelligence (AI) refers to the broad ability of a computer or robot to perform tasks that would normally be done by humans. Traditionally, computers have been able to form data-driven tasks like complex calculations, but AI enables computers to accomplish tasks like visual recognition, speech recognition, decision making, and more.

Machine learning is a subset of artificial intelligence. With machine learning, the computer is able to learn and improve on its own without explicit instructions.

Is machine learning expensive?

While machine learning has a reputation for being prohibitively expensive, it’s not as pricey as it used to be. Generally, it’s expensive to train a machine learning model, not to use one. For example, it’s estimated that a single inference by a massive machine learning system called GPT-3 would cost a mere fraction of a single penny. Now that machine learning insights are becoming more accessible, they can often be built into other programs you’re already using, such as your prospect research database.

Who can use machine learning?

Machine learning has an extensive variety of applications. In fact, you likely rely on it every day without realizing it! Whenever Facebook suggests a friend to tag in the latest photo you posted or your inbox filters out suspected spam emails, machine learning algorithms are in play. It’s used in the for-profit sector to provide customer support, make product recommendations, and detect fraud. Nonprofits can use machine learning to optimize marketing campaigns, plan annual giving days, identify major donors, and execute other operational and fundraising tasks.

What do you need to incorporate machine learning into your fundraising strategy?

To use machine learning, your organization needs to have a solvable problem and a robust dataset. Without the right problem or question in mind, you’ll waste valuable time and resources trying to gather aimless information. Without a solid dataset, your machine learning model will be unable to learn anything useful. But with both components, informative answers can be discovered. For example, leveraging the information in a prospect research database to identify qualified major gift prospects is an appropriate use of machine learning.

Will machine learning replace fundraising professionals?

In short, no! Don’t worry about a robot army (or just a computer program) coming to usurp your role as a fundraising professional. Machine learning can elevate, enhance, and simplify your work for more effective outcomes. With less time spent on repetitive tasks and more valuable insights from your data, you’ll have more time to focus on the areas that still need a human touch, like cultivating relationships with donors. Additionally, machine learning can help you more accurately prioritize this outreach for more profitable fundraising outcomes.

Benefits of Machine Learning for Nonprofits

While machine learning can personalize digital experiences and increase business revenue, it can also help nonprofits power social good.

Consider the following ways machine learning can benefit nonprofits and help them function more effectively:

Automate administrative tasks

Daily to-dos like sorting donor data or conducting screenings can require painstaking effort and hours of labor when done manually. With machine learning, these processes can be executed with the press of a button, saving valuable staff time.

Handle business needs like HR and finance

This is especially useful for smaller nonprofit teams that may not have a designated staff member for each business department. For example, machine learning can help you manage and streamline the nonprofit hiring process and track employee performance over time. It can also analyze financial records to minimize misconduct and fraud.

Prioritize resource allocation

Machine learning can help your organization determine which of your constituents have the most urgent or critical needs. Then, you can organize and prioritize the allocation of resources and time more effectively. For example, Crisis Text Line used machine learning to identify terms that more accurately predict the need for emergency aid, allowing the highest-risk cases to get help as quickly as possible.

Evaluate potential gift capacity

Your fundraising team is already used to using data to identify potential major donors. However, machine learning algorithms can analyze countless variables (like wealth and philanthropic indicators) with lightning speed, making the process much more efficient. Plus, over time, the algorithm will become more and more accurate when predicting which donors have the highest capacity and affinity to give.

Inform marketing and communications

Machine learning can help your organization optimize communications by analyzing engagement metrics. The algorithms look at data like open rate and send time to make informed recommendations that will improve the performance of your next email or social media post.

Improve prospect pool segmentation

Machine learning adds efficiency and effectiveness to the process of grouping donors. This helps you avoid impersonal “mass” appeals and streamlines the ask process. This way, your team can focus more on messaging that will compel donors to give rather than worrying about actually sending the message.

Predict Likelihood to Give

Machine learning can also help you predict a prospect’s overall likelihood to give, rather than just their capacity to give. Predictive models can process a wide range of available data in order to calculate a comprehensive likelihood score. This score will help your fundraising team prioritize outreach and secure donations.

More broadly, artificial intelligence can also help nonprofits sift and sort data much more quickly than staff members. With so many records on file (like personal information, individual donation amounts and patterns, event history, and more), it’s helpful to have a system that can process and make sense of this data quickly.

There are several possible use cases of machine learning for nonprofits

3 Use Cases for Machine Learning Fundraising

Intelligence University could use machine learning to plan for their annual Founder’s Day campaign.

A university alumni giving day is a great example of the power of machine learning for nonprofits.

Each year, Intelligence University plans an alumni giving day on the anniversary of the institution’s founding. Like similar campaigns for other higher education institutions, this giving day will help fund the general operating budget, campus renovations, unique extracurricular and educational programs, and scholarships to students in need.

As the Intelligence University development team starts preparing for this year’s event, they need to set a reasonable (but ambitious!) fundraising goal, start identifying candidates for major gifts, and draft campaign messaging.

While alumni are a valuable source of funding, it can be challenging to identify which former students have the highest propensity to make a significant contribution. In addition to major gifts, Intelligence University also wants to work towards a high participation rate in the campaign to get a more comprehensive view of engagement efforts.

With machine learning, Intelligence University can quickly process a huge volume of past student and donor data to determine which students are the most valuable prospects. When compared to typical major donor prospecting, machine learning can identify 4-5x as many qualified prospects in a much faster time frame.

With the algorithm handling this herculean task, Intelligence University’s development and alumni relations professionals will be able to focus their attention on making personal connections with individuals or creating emotionally compelling appeals.

The Intelligence County Animal Shelter could use machine learning to inform their #GivingTuesday marketing.

An animal shelter's communications plan is a great example of the benefits of machine learning for nonprofits.

The Intelligence County Animal Shelter is hard at work planning their fundraising campaign for #GivingTuesday. This won’t be their first year participating in this worldwide day of giving, but they want to step it up a notch and achieve a higher fundraising goal than last year. While it’s possible for a team member to process and analyze historical campaign data on their own, the processing and reporting required would be challenging and time-consuming.

Machine learning can help the animal shelter efficiently learn from their past #GivingTuesday campaigns to increase its fundraising total.

A machine learning algorithm can help the animal shelter determine:

Which send times led to the highest conversion rate
Which platforms drove the most engagement (email, social media posts, direct mail etc)
Which content elements were the most useful for encouraging clicks (images, videos, etc)

The marketing and communications team can take all of this information and use it to craft more compelling and effective appeals and posts. Additionally, machine learning can help the team create more accurate segments of donors and target each group with a personalized message.

Finally, machine learning can identify which of these segments (and which individual donors) are most likely to make a generous contribution, so the fundraising team can follow up and conduct outreach. Like in the other use cases, these predictions are based on a combination of existing donor information and prospect research data.

Intelligence Hospital could use machine learning to raise funds for a facility expansion.

A hospital's grateful patient program is an excellent example of how machine learning for nonprofits can work.

Intelligence Hospital is one of many medical institutions that rely on fundraising initiatives to offer state-of-the-art equipment, improve their facilities, pay for supplies, and other important needs.

Right now, Intelligence Hospital is planning a capital campaign in order to build a new wing of the building that will be tailored to the needs of children with cancer. While there is a wide variety of potential donors that Intelligence Hospital could solicit for this healthcare fundraising campaign (such as doctors, corporate partners, or health-related foundations), one of the most common groups to pursue is grateful patients.

A former patient is likely to someday give back to a hospital if they are especially grateful for the care and treatment they received. By making a donation, a grateful patient can express their appreciation and provide support to the individual staff members and departments that were most meaningful during their care.

However, due to the intimate and delicate circumstances of medical treatment, it can be awkward to make fundraising asks to past patients. Development staff may worry about coming across as insensitive, or worse, violating HIPAA. Plus, timing is especially important, as patients should never be approached while still at the hospital but are most likely to donate soon after their discharge. While beneficial, grateful patient programs can be challenging and time-consuming. This is where machine learning can come in.

Intelligence Hospital can use a machine learning algorithm to identify potential donors from past patient records and automatically screen new patients for giving potential on an ongoing basis. This information can help gift officers conduct outreach to the most qualified prospects and secure the funding needed for the new children’s cancer wing.

As a real-life example of this in action, Futurus Group’s powerful Gratitude to Give model takes information from DonorSearch’s prospect data set and identifies potential donors based on gratitude. And because the model will continue to evolve and learn over time, the insights discovered will only become more and more accurate.

Wrapping Up

In the next few years, machine learning is likely to become more prevalent throughout fundraising and other nonprofit operations. With such clear benefits and use applications for this technology, it’s great news for the nonprofit world that machine learning is fact rather than fiction. To start making use of this forward-thinking innovation, look for a platform like DonorSearch that has the technology built in.

For more information about machine learning and other innovative fundraising technology, explore the following additional resources:

Artificial Intelligence for Nonprofits: Fundraising’s Future. While machine learning and artificial intelligence aren’t interchangeable terms, the two technologies have similar functions and applications. Learn more about AI’s potential for nonprofits in this article.
Major Donor Fundraising: 13 Effective Strategies for 2020. Machine learning can help your nonprofit identify and target potential major donors. This guide offers other useful strategies for major donor fundraising.
Prospect Research: The Ultimate Guide. The benefits described in this article are only possible when machine learning algorithms can draw from a detailed data pool, like a prospect research database. Explore the basics of prospect research here.