Non-profits and Donor Churn
Individual and corporate contributors are the lifeblood of non-profits, tied to the fabric by supporting the organization’s viability, sustainability, and advancement. Yet, non-profit groups know very little about their donor segments. It’s easy to create a profile for a reliable, faithful supporter who writes a check every year. But, why do short-term donors leave without a trace and where can organizations find future contributors who are ready, willing and able to donate? Retailers and digital marketers have perfected the art of targeted segmentation, but non-profits have yet to fully utilize technology to create profiles and intimately understand the interests, preferences and underlying behavior of their donors. Now, nonprofits can now capture critical donor base information and learn more about those who leave (churn), reduce the risk of future churn and attract new donors in real time.
Non-profits are Losing $4.2 Billion, Yearly
The magnitude of donor churn is pervasive across the industry. The “Fundraising Effectiveness Project (FEP)” by the Association of Fundraising Professionals reports that nonprofits are losing approximately $4.2 billion, yearly due to donation attrition in various forms including donation abandonment. And that is only the nonprofits studied. The actual number is much higher.
The findings of the study say it best:
“It usually costs less to retain and motivate an existing donor than to attract a new one… taking positive steps to reduce gift and donor losses is the least expensive strategy for increasing net fundraising gains.” — Fundraising Effectiveness Project
Under-utilized data and analytics can account for as much as 30-40% of losses in donations and smaller nonprofits are well advised to prioritize their donor retention strategy before (or at least concurrently with) their donor activation strategies.
The Missing Link is Data
To segment and improve your prospective donor base, you must begin by gathering critical information. The following potential data can be readily collected from a donation page without compromising any of the donor’s personal information.
- Individual’s identity
- Net worth
- Household Income
- Home value
- Luxury purchase history (luxury purchases are highly correlated with non-profit donations)
- Age, gender, zip code, latitude/longitude and a host of other demographics available free from the US Census bureau.
- Personal interests and affinities
- All the topics and phrases of interest in a visitor’s history of reading your web page
- The names of your donation page visitor’s employer is available to you (for the vast majority of visitors)
If you work in a non-profit, there are also some questions to consider in determining if a prospective donor will donate, churn or abandon their donation:
- Do I get notifications when visitors to our donation page abandon the page without donating?
- Do I know the names of the companies and the names of the individuals who most frequently visit our non-profit website?
- What is the zip code of our “almost donated” crowd who quit the donation process or tried to donate and failed?
Three Years of Donor Abandonment Analysis
In a 3-year project, we were able to identify and map 96 potential donors with a minimum one-million dollar (and greater) ability to donate but who had abandoned various donation pages, online. This following pilot study resulted in a 30% recovery rate from all abandoners: https://www.linkedin.com/pulse/donation-retention-universities-israel-kloss
|Month||# of Donation Page Abandoners||$1M and above donation capacity|
Diagram 1: Donation page abandoners by month with donation capacity.
The total capacities of the 96 abandoners (who had greater than $1 Million giving capacities) in our study was $96,000,000
By recovering just 10%, there was a potential for $9,600,000 in recovery. By pursuing the 96 abandoners alone with the 30% recovery rate proved that the above pilot study (and a similar pilot study by Dickenson University) demonstrated was possible, Conservatively, $28.8 Million could have been recovered for this non-profit.
What is Donor Churn Risk?
Donor churn risk is simply a calculation of the probability of a donor to stop donating (whether online once or recurring). This is also called donor attrition risk or donor turnover probability.
Historical Donor Data
Historical behavioral data about your donors is critical for generating an actionable donor churn risk score. If you don’t already have someone responsible for collecting (and protecting) historical data associated with the donors, make it a priority.
By exporting and crunching some of the key Google Analytics data through a program like R Studio, you can get more donor behavioral insights than most people imagine. In fact, you might be surprised about how much historical data you’re already recording. Got Google Analytics? Got log files? Almost everybody with a web site has log files. It may only be Apache log files, but that’s still data. You are measuring, whether you know it or not. You’ve got data.
While Google Analytics data is a fine place to start you’ll likely quickly need to go beyond just Google Analytics data to effectively address donor churn. There are more powerful analytics products available that will let you do things like segment traffic-to-donations ratios and do campaign personalization based on personal data about your logged-in (and other) visitors.
Why Behavioral Analytics Matter
Tracking donor-level behaviors allows non-profits many advantages:
1) Preemptive Intervention: By intercepting potential problems before your donors quit the donation process, you
can help stop loyalty problems from cropping up for your non-profit.
2) Donor Acquisition and Retention: Behavior-triggered donor churn alerts can allow your non-profit lead time on new retention strategies because, of course, donor retention is the job of everyone in any non-profit.
Let’s Dive In!
So here are some steps to help you use your internal non-profit data to slow down your donor churn rates.
Step 1. Collect and Analyze Donor Concerns
If you haven’t already, do the following:
- Make a list of observations that the frontline donor care team believes has directly lead to churn for each donor (institution or individual) for whom they are familiar.
- Make a list of observations from the frontline donor care team has heard from each donor (institution or individual) regardless of whether they believe it has directly led to donor turnover.
- Request that a team with the closest knowledge of donors ranks concerns and objections they’ve heard over their careers from donors in this Donor Concern Matrix Model, It can be downloaded here: http://archetypeconsulting.com/donor-matrix-model/ . The Donor Concern Matrix Model helps our non-profit customers to diagram donor concerns with 3 institutional priorities:
- Ease of addressing the donor concern
- Likelihood of addressing the donor concern
- Impact of address the donor’s concern
Diagram 2: Ease, Likelihood & Impact Matrix Diagram
- Record these concerns in the permanent record of each donor (institution or individual). If you don’t have a database, that can be solved but this guide assumes that you have one or will acquire one.
How to Use the Donor Concern Matrix
Let’s say that your donor-facing team members have heard from multiple donors that they are concerned about a lack of alignment between their philanthropic interests and your organizational priorities.
These concerns will show up in a statistical analysis of your institution’s donor churn rate (step 2).
Thanks to University of Illinois Office of Business and Financial Systems for the risk map template.
Step 2: Generate Your Donor’s Churn Risk Scores
If you’ve never done any work with a statistical package, you might want to start with a basic data mining with decision trees (see below). Statistical packages available for this step include R Studio, SPSS, SAS or even Excel’s solver features. In fact, there multiple Excel plugins available for basic statistical modeling.
Let’s start building an actionable donor churn risk score. Make sure your donor data set includes the donor concerns expressed for each donor no matter whether they are current or past donors. You must include their last donation, the amount and the full history of their donations). With all this data you can calculate a single score representing the risk that they will stop donating (the donor churn risk score).
R Studio (free) will help you generate churn risk scores from your data. Below is an example of how just a few lines of code can provide a statistically valid churn risk score for every donor in your dataset.
Diagram 4: R code demonstrating churn risk scores calculated from multiple types of models
Try it Yourself!
Check out this short tutorial to start your journey with R: https://www.youtube.com/watch?v=Gzfo4piKwdw. To be statistically valid, your dataset needs to meet statistically valid minimum for the size of your data set. You can find the minimum requirements for the levels of confidence that you seek at this site: https://www.surveysystem.com/sscalc.htm (and many others).
There are multiple approaches to churn risk scores. One is called RFM. Try out some RFM sample code (in R) here: https://github.com/apurvadeshmukh/Churn/blob/master/churnmodel.R. You can also do a simple search in GitHub for other RFM code and learn many different approaches to calculating donor churn risk.
It’s ok! Not everyone has time or the interest to learn to code in R. In this case, why not try basic data mining? Decision trees are is often considered a more accessible (and event fun) starting point for learning more about donor data. Decision trees are part of a group of data modeling methods called “supervised learning”.
Step 3: Map the Risk
After a churn risk score has been calculated for every donor, it’s time to take a closer look. You can export the data from R as a CSV file and import it into Excel, or you can get fancy and visualize the data in Tableau. Below is a very basic view of how churn risk scores look after an export from R into Excel.
This is great, but sometimes you just need a map to make it more real. I’ve included some donor latitude/longitude columns and donor zip codes for the next step, donor churn-risk mapping.
Diagram 5: Donor churn risk scores calculated in R and exported to Excel
Predictive analytics is not always as understandable in the non-profit world as a good old-fashioned map. You can bundle up the highest-risk visitors by town (over population) using Tableau (a visualization tool), so the cities with the highest donation churn risk are visible by population center.
Diagram 6: Export of donor churn risk scores mapped by city across the highest-risk population centers.
Step 4: Map Donation Abandonment
Using Tableau, you may want to overlay the donors with the non-donors (donation page abandoners). We found patterns in the results that lead to significant increases in total donations and per-donation increases for one customer by mapping donor abandonment. You can read more about our findings in these articles:
- Mapping Alumni Data: https://www.linkedin.com/pulse/mapping-alumni-data-israel-kloss
- Donor Retention for Universities: https://www.linkedin.com/pulse/donation-retention-universities-israel-kloss
There is so much under-used data in the non-profits world. The insights available for better operations are nearly inexhaustible when the data is accurate and the analysis methods are statistically sound and valid. Non-profit leaders have so much to gain from better data collection, data wrangling, data analysis and donor predictive analytics. $4.2 Billion is no chump change. There are so many great opportunities to recoup that loss. Data is the new oil. Now you have the know how to go mine it!
It would be our pleasure to assist you in your analytics projects, data warehouse initiatives, or data quality efforts. We are an analytics company with a passion for using data and technology to help our clients work smarter and make room for bigger and better ideas. To learn more, please contact the members of the Archetype team, or visit us online at: www.archetypeconsulting.com.
Israel Kloss is a Senior Consultant at Archetype Consulting in Boston, MA. He provides a broad range of data mining and modeling expertise to various clients throughout all lifecycles of the business optimization and data simulation projects. He has over 17 years of experience in web design and development, marketing automation, predictive modeling, data mining, opportunity mining, and competitive intelligence. You may see his articles on Donor Abandonment for universities and Account-based Marketing. He is also an occasional analytics conference speaker (emetrics.com, Case.org, Tableau User Groups, NEDRA, ODSC and Drive Conference)
- Top Jobs That Pave the Way for Becoming a Data Scientist in 2020 95 views | by Daniel Gutierrez, ODSC | under Career Insights, Featured Post
- 10 Best Data Science Platforms 46 views | by Elizabeth Wallace, ODSC | under Featured Post, Platforms
- Creating if/elseif/else Variables in Python/Pandas 41 views | by Steve Miller | under Python, Tools & Languages