The Basics of RFM Segmentation - Segments
The mail / no mail decision is made a group of names at a time. Every name in a group is mailed or not. These groups are called segments.
The most popular and easy to implement segmentation for the house file is RFM, for recency, frequency, and monetary value. RFM has the advantages of being simple and easy to understand. When properly set-up, and enhanced with product purchase information, it is more than adequate for most B2B direct marketers. Segments defined using RFM are called RFM Cells.
Pros and Cons of Segments
Using segments has several advantages. They are transparent: everyone can understand who is in a segment and why they are there. Segments are consistent and easy to control over time, once you settle on a workable scheme. This allows results to be validly compared from campaign to campaign and year to year.
The major disadvantages of segments are: defining them; and the tendency for segments to multiply. Coming up with a solid set of segments that gives you the level of control you need to plan and measure your mailings is a challenge. For example, when using RFM you have to decide on how finely to pull apart recency. Do you need weeks, months, quarters or years? Do six time buyers need to be broken out from seven time buyers? Is it worth breaking out extremely high value customers? These questions can only be answered by practical experience and looking at results. You may have to do some testing as well. The number of segments can multiply rapidly as you add various enhancements or wish to consider more ranges in one of the RFM dimensions. The best solution for this is flexible software that allows you to easily define different RFM structures for different purposes.
Attributes of Good Segments
You cannot throw random names into a segment and expect the result to be useful. Segments should have these characteristics:
Similarity (within a segment)
To justify mailing all or none of the names in a segment, we should have some reason to believe that the names are similar. RFM provides that justification for the house file.
Distinctiveness (between segments)
While the names within a segment are expected to be similar, the names in different segments are expected to differ. Otherwise why have two distinct segments?
Over time we may learn that the behavior of two groups turns out to be very close and we merge them. For example, we would probably see a big difference between 2 time and 3 time buyers and therefore expect to see a difference between 3 time and 4 time buyers. Only experience will show if there actually is a difference.
Comparability (over time)
Segments are used for both planning and results analysis over time. We want to be able to compare a segment's results from campaign to campaign and year to year. For RFM segmentation, this requires stability in the RFM structure.
In prospecting, this requires careful planning in how to order list. A long term prospecting plan (perhaps better, an approach to prospecting) is needed to ensure segments will be comparable over time. For example, our plan (or approach) would dictate that all lists be broken out and coded by year group, even is we are not interested at the moment. In the future we will almost certainly find it valuable to look back and see how the various year groups performed.
Size (large enough)
Segments should be large enough that meaningful results can be read. There is no point in having a lot of cells with 100 names in them.
A prospecting segment is a particular select from a prospecting list or database, such as "fourth quarter Hello Direct buyers" or "SIC 35xx with 1-4 employees."
Prospecting lists tend to be small, so year groups are often used as the segment.
House File Segments using RFM
RFM (for recency, frequency, monetary value) is the most popular method for creating house file segments. It is easy to implement and understand.
A carefully designed RFM scheme meets the criteria discussed above:
The next two sections of this article discuss how to create an RFM scheme.
Next section: RFM Values