How to use analytics to ensure you target the right donors with your marketing efforts
Segmentation is often misunderstood.
It’s one of the most powerful tools we can use to target existing donors. Segmentation also establishes the framework and vocabulary that allow us to use valuable insight to inform our contact strategy for donor audiences.
Unfortunately, segmentation is often misunderstood in practice. We tend to fail to balance tactical and strategic priorities with reliable mathematic assumptions and predictive analytics. Basic business rules play tug-of-war with analytical best practices, often resulting in a segmentation that provides so much detail that it can’t be implemented effectively.
But, there’s hope. Segmentation can help your organization when it’s arranged properly. Here’s how.
2 approaches to segmentation
There are many types of segmentation an organization can leverage to optimize marketing efforts. Here are 2 that we use often:
- Tactical Segmentations tend to focus on optimizing one or more specific outcomes like value or life-stage.
- Market Segmentations utilize more descriptive qualities to inform strategy and tend to be defined by characteristics that shift less over time – demographics, attitudes, needs, etc.
 |
| There are many categories of constituent dimensions. Tactical Segmentation tends to focus on behavior and Market Segmentation tends to focus on identity. |
Even though market segmentations can help, tactical segmentations are essential to effective targeting strategies.
Segmentation by Recency, Frequency and Monetary value (RFM)
One of the more common forms of tactical segmentation is a value-based approach that leverages historical behavior data to subdivide donors into meaningful groups. This approach is foundational to our RFM segmentation strategy.
Some of the most powerful and accessible attributes that predict future behavior are past behaviors.
- Recency of behavior
- Frequency of behavior
- Monetary value of behavior
These 3 components can be represented through a variety of data transformations. The process of classifying RFM requires a marriage of science and art to arrive at a framework that creates reasonable differentiation across the file while accurately predicting future donation behavior.
We start with the Recency component of RFM. Recency shows us that donors are most likely to give in the time immediately after their gift. The likelihood of future response declines rapidly during the subsequent months, then more moderately over time.
 |
| Recency of last response is best at differentiating donors in the first few years following a response. |
In determining how to distribute groups within a given RFM component, it is important that we use not only the distribution of population, but the strongest differences and similarities of performance to inform the various break-points, creating fewer, larger segments of very similar donors.
Frequency components can help us improve on the common single vs. multi segments with deeper classifications of multi-gift donors, 1x Annual (1 gift 0-12 months from last gift) vs. 2+ Annual (2+ gifts 0-12 months from last gift). Monetary segments are influenced by the most common donation sizes on the file as well as the change in predicted gift size.
 |
| With the ability to differentiate “Multis” into “1x Annual” and “2+ Annual,” we can leverage stronger predictive power in our segmentation. |
 |
| Monetary groupings among low dollar ranges make the most sense when informed by how common those gift sizes are within our marketing program. |
The results of RFM segmentation in action
We conducted a validation exercise on a large ministry fundraising program, testing re-selection of a historical campaign using this new RFM segmentation vs. their legacy segmentation criteria, and found significant improvements.
The revised value-based segmentation was better at reducing net loss (23% lower) and realizing net gain (11% more within the campaign) than previous RFM segmentation criteria. This simpler segmentation also contained far fewer total cells to evaluate (147 vs. 988), creating a simple and efficient measurement process to select and deselect cells from any given marketing impact.
Do you have any questions about RFM segmentation? Please send me an email at scaldwell@masterworks.agency.