Most of the time, when I get questions about predictive models, the question relates to the score. Does the score tell us “yes” or “no?” Predictive models are built on historical behaviors to try to predict a specific future outcome. The answer to the question is generally “yes,” the model should predict if a behavior will occur.
However, what really makes predictive modeling powerful is the ability to use the resulting score to help define the degree of “yes” that’s predicted. By analyzing the data, you can often determine the expected yield of an outcome and use this information to drive appropriate, efficient workflow. For example, you have a list of prospective donors that are the “yes” list. A predictive model goes one step further to rank order these “yes” donors in an intelligent way to tell staff exactly who to contact personally and who to include in other outreach tactics such as direct mail.
Also, models can predict the extent or degree of a prospective donor’s “yes.” It will tell you whether they are willing to donate $50 or $500. This empowers your staff with additional knowledge on how best to engage a prospective donor using different tactics such as phone, email or direct mail.
An essential part of creating a donor segmentation strategy is the use of an Odds Table. An Odds Table is used to determine the expected yield of an outcome to drive the right workflow. Odds Tables are built from breaking donor score results into bands and comparing those scores. This analysis is a key component of creating a segmentation strategy as it allows you to invest the appropriate level of resources based on the expected outcome. Additionally, Odds Tables can identify trends and increase speed to donation based upon historical analysis.
An example of how the information from an odds table can be used
In the chart below, the odds of a donation in Segment 9 is 71% with an average donation of $1,515. Segment 9 is the highest segment with a likelihood to donate and the highest likely donation amount. In comparing Segment 9 to Segment 7 (where the odds of a donation are 59% with an average donation of $175), it’s more valuable from a time and resources standpoint to have a different engagement strategy for Segment 7 than you would for Segment 9. You would want to use a more automated approach for Segment 7 and a more hands-on approach for Segment 9. Using the $415 national average donation for healthcare, Segment 9 prospects represent a potential donation more than 3 times the national average.
Having this knowledge allows you to apply different engagement strategies to different segments (and maybe on even approaching some prospects). This allows for a more effective donation outcome with effectively managing cost to cultivate donors.
We’ve tested many strategies with our model. Below are some suggestions on developing a strong segmentation strategy:
- Ask your predictive model vendor to help you devise an Odds Table based on their scores. They should be able to provide you with a table that shows odds of donation and expected yield in each segment.
- If you’re considering a large outreach for annual giving, the Odds Table will help you define your target audiences and enable you to create variable messages based on the expected outcome. This includes how much time and resources you invest in each segment. Different engagement methods and channels will work for different segments.
- Remember that your odds table is usually based on cumulative experience over multiple campaigns. Campaigns for Segments 8 and 9 might not be effective for Segment 7 since that segment donates in lower amounts.
- Using the segments can help you put those prospects with the highest likelihood to respond at the front of the line. Your major gift officers will likely see more interest in their outreach efforts which is a great positive step.
Using predictive models to segment prospective donors can not only create efficiencies within your organization but also increase speed to donation and ultimately return on investment with your marketing efforts. Competition for donor dollars is high with our current economic climate and most non-profits have a steep hill to climb. Make prospect segmentation a part of your strategy to drive all potential donors to act whether it be for $50, $500 or $100,000.