Donor database or crystal ball?

publication date: May 24, 2018
 | 
author/source: Matthew Dubins

Donors are the beating heart of your organization. For that reason, wouldn’t it be great to know what they are likely to do with their donations next year?

To help you and other organizations in the social profit sector, I’m happy to share with you my process for predicting future annual donor behaviours, and test results that show how you can use these predictions to gain a much higher return on investment.

Ingredients for Predicting Donor Behaviour

To get useful predictions, the biggest ingredient here is a donation history file. You may have heard that past behaviour helps predict future behaviour, and that is especially true when predicting future annual donor behaviour. Let me show you an example donation history file that contains all columns of information necessary to make great predictions:

Predictive Modelling - The Actual Crystal Ball

Using the above columns of information, you can then feed the data into 3 predictive models, that score each donor based on their likelihood to:

  • Give at all next year
  • Upgrade their giving next year
  • Reactivate next year

The final scores for each donor will be based on considerations such as:

  • Did they give at all in recent years?
  • Did they give to a popular appeal?
  • How long has it been since their first gift?
  • Which appeal type is their preference?
  • Which fund type is their preference?

Sensible Recommendations Based on Predictive Scores

What I like to do is to take the donor scores that come out of this predictive modeling process and create plain English recommendations. I call my process ‘DonorFocus’, by the way.

  • If they gave last year, and have an upgrader score in the top 30%, your next step would be to ask them for more money next year.
  • If they gave last year, but don’t have such a high upgrader score, target them for renewal with the same ask amounts.
  • If they haven’t given in a year or so, and have a reactivator score in the top 30%, you should target them for reactivation.
  • For everybody else, please leave them alone :)

Retrospective Test Results

To test this, I took gift history files from several organizations, and submitted data up to the second last year available. Once I had recommendations for the following year, it was easy to see how valuable those recommendations would have been. The two key metrics here are donor rate, and upgrader rate. These just mean what percent of donors gave, or upgraded, in the last year on file.

Test Org. 1 - Higher Education

 

 

Look at that ‘Upgrade’ group! Using past data, this organization could have identified a group of real superstars. The donor rate was phenomenal, and the upgrader rate was over 5 times the rate of the ‘Same Ask’ folks!

Test Org. 2 - Disease Awareness & Support

Yet again, the ‘Upgrade’ group are the top performers from both a donor and upgrader rate perspective.

Want higher donations from your superstars, and less money wasted contacting the disinterested? Predictive recommendations are a game changer!

Matthew Dubins is Chief Donor Scientist at Donor Science Consulting: a truly Canadian consulting agency using predictive analytics, data visualization, dashboarding, and address correction to help you do better fundraising with the help of your data! You can reach him at matt.dubins@donorscience.ca.



Like this article?  Join our mailing list for more great information!


Copyright, Hilborn Stanois Inc., © 2011-Current. All rights reserved.

Free Fundraising Newsletter
Join Our Mailing List


 

Hilborn:ecs