January 10, 2019 Kirstie Kiernan and Gurjeet Singh
Rather than relying on evaluating the effectiveness of past fundraising efforts and basing decisions on opinions and experience, predictive analytics provides additional guidance on what will likely be the most effective campaigns, whom to target and how to allocate resources to maximize fundraising results.
Predictive analytics is a set of techniques and technologies that extract information from data to identify patterns and predict future outcomes. Based on a variety of statistical techniques and software technology, predictive analytics helps to understand the relationships between data points and identify patterns within the data, as well as which factors contribute to the prediction. This analysis can be configured to show prediction based on various factors and can be refined further over time as more information is included in the analysis.
Most nonprofits house the type of data that can fuel detailed analysis, which results in actionable insights. They have donor information that often includes a wide array of demographic information, historical behavior information and information about how donors responded to past fundraising campaigns. This type and breadth of information can quickly be converted into predictions and more effective fundraising campaigns. Even if the nonprofit only has hundreds or thousands of donor records — as opposed to hundreds of thousands or more — that is sufficient for creating effective predictive analyses.
What can be predicted
Predictive analytics can help identify the people who are most likely to donate and those unlikely to give. To identify potential donors, managers can examine past donor information to identify the characteristics that most accurately determine whether someone donates. Unlike traditional analysis methods that only examine past donation information, predictive analytics leverages information such as age, income, lifestyle, past donation information and associations to nonprofits with similar missions to pinpoint donors. With this information, fundraisers can more precisely target a pool of potential donors to maximize fundraising results.
For example, a nonprofit with a list of 3,000 past donors and 2,500 potential donors might only be able to directly contact 2,000 donors through in-person meetings, phone calls and/or direct/digital mailing due to budget constraints. Top determine which of the 5,500 potential donors to contact, fundraisers can assign a donation probability to each potential donor based on historical donation information and each potential donor’s characteristics to then target only the donors with a high probability. This helps reduce the overhead of devoting resources to individuals or groups who are unlikely to donate and maximizes the donor conversion rate.
Going one step beyond identifying donors, predictive analytics can be used to predict donation amounts. Fundraisers can assign both a donation probability and an expected donation amount if they donate. This is an expected value for donors, and this information can be calculated to optimize the fundraising campaign. If a fundraiser identifies five high-value donors who only have a 40 percent donation probability, targeting those might still be more valuable than pursuing five low-value donors who have a greater than 90 percent probability of donating.
Predictive analytics can be applied to almost any area of nonprofit operations. While improving fundraising is often the first goal, predictive analytics can be used to improve other areas of the organization. Several examples of these are:
• Mission-specific goals;
• Operational performance;
• Cost forecasting; and,
• Community and government outreach.
The Bigger Picture
Predictive analytics is an important method for improving your fundraising process. Just as major retailers, financial institutions and healthcare companies are utilizing predictive analytics to maximize revenue and reduce costs, nonprofits have an opportunity to make use of this technology within their own organizations.
Regardless of the volume of fundraising you are doing or the makeup of your donors, you can benefit from applying predictive analytics to your fundraising campaigns this year.
In the traditional fundraising process, several steps are typically employed across different layers of a nonprofit’s data. First, key donors are identified and targeted. This may be done based on selecting key individual donors, by a prior donation level threshold, and/or demographic information. Next, past campaigns are assessed, and new campaigns may be discussed and evaluated. Finally, a plan is developed and executed to drive fundraising. For this entire process, the rigor of data analysis and the evaluation of past campaign effectiveness may vary by organization but, at a high level, the processes are similar: organizations make use of data and personal judgment to drive future fundraising efforts.
The predictive analytics process runs alongside this methodology to augment it, which acts as an advisor to existing activities and decision-making processes. Predictive analytics offers a way to look at the information in a new way by incorporating your existing methods and institutional knowledge. Predictive analytics can be run parallel to your process to offer new ideas, prove or disprove existing ideas and approaches, and provide a way to gauge how effective new approaches to fundraising will be.
Core steps are outlined in the accompanying graph. They include:
✲Identify the organization’s goals and survey historical data to map the goals to key data points. In this step, the organization determines which questions it wants answered and whether the data it needs is available;
✲ Develop a predictive model and analyze and visualize the data to derive insights. This step allows nonprofits to derive forward-looking analyses and findings from historical data. Specialized software and/or custom-developed logic in a programming language, such as Python or R, is used for the analysis;
✲ Evaluate the results to determine whether the analysis was effective and, if so, how to apply the findings for meaningful actions; and,
✲ Finally, an iterative process of refining and re-running the analysis is performed based on the findings and changes.
A major misconception about predictive analytics is that it can replace a fundraising team or will serve as a stand-alone fundraising strategy function. A predictive analytic model is only as effective as the information and guidance that is provided to it. Performing predictive analytics effectively requires institutional knowledge and refinement. Predictive analytics is a statistical and technological way to utilize data based on institutional knowledge, so it is useful only if it is designed, implemented and evaluated by data and industry experts.
Pilot predictive analytics
Many fundraisers choose to implement predictive analytics when the board or executive leadership team recognizes that its fundraising efforts could be improved. At that stage, an organization might have sufficient data to understand what worked well in the past. But, it would likely rely on comparisonsbetween past approaches and new approaches, market research and small test campaigns to evaluate new ways to raise funds. Nonprofit leaders might choose to engage a data scientist team or purchase a predictive analytics package if current fundraising techniques and test ideas — which require an investment of time and resources — aren’t delivering the desired results.
Predictive analytics is an important method for improving your fundraising process. Just as major retailers, financial institutions and healthcare companies are utilizing predictive analytics to maximize revenue and reduce costs, nonprofits have an opportunity to make use of this technology within their own organizations. Regardless of the volume of fundraising you are doing or the makeup of your donors, you can benefit from applying predictive analytics.
Kirstie Tiernan is managing director, Technology & Business Transformation, at BDO. Her email is email@example.com. Gurjeet Singh is senior manager, Data Analytics & Automation, at BDO. His email is firstname.lastname@example.org.