At DonorVoice we do a lot of segmentation work.  In our previous lives we did even more.  We love segmenting, slicing and dicing…And yet, we have seen far too many poorly conceived segmentation schemes and an equal number of well conceived ones that never get implemented.  Heck, we’ve even participated in both scenarios and hopefully learned a little something each time.

We thought it worthwhile to recap our view on the role of segmentation, the various types and offer up a fairly new and very interesting (downright useful even) segmentation we’ve been doing based on package preference (more later).

To state the obvious, segmentation exists because firms believe they can be more financially successful by treating customers/donors differently.  If you don’t believe this or you don’t treat groups/segments differently then don’t waste your time and money.

A few important (and perhaps also obvious), semi-technical criteria for a useful and successful segmentation come to mind,

• People in the segment are very similar to one another.
• The various segments are distinct from one another based on different preferences/needs/behaviors.
• The segments are relatively stable over time.
• The segments are identifiable and reachable in the marketplace
• The number of segments is organizationally manageable.

The key with segmentation is choosing the variables to be used to create the segments.  They can, broadly speaking, be grouped into behavioral, demographic and attitudinal.  Non-profits do a lot of behavior and demographic segmentation, very little attitudinal.  They miss a big opportunity with this omission.

There is room for debate (and therefore, plenty of it) on the technical aspects of a segmentation, particularly one of a statistical variety.  There are a myriad of statistical methods and approaches and there are certainly bad ones produced all the time.

That said, there is no “right” and “wrong” segmentation schemes per say (which is different than saying there are not bad segmentations).   In fact, it is the case that every customer and every donor belongs to lots of different segments, plenty that aren’t identified or focused on by the organization and plenty that are, with the choice of which segmentation scheme to use based on who, within the organization, is doing the slicing and for what purpose.  For example, geography obviously matters for event based activities even if those in a similar geography are very different in terms of financial behavior on the donor file and thus in very different segments when doing direct mail plans.

Having a multifaceted segmentation strategy, one that recognizes the value of multiple “looks” at the exact same donor, can only help the understanding and marketing effectiveness.

One recent segmentation scheme we’ve been employing – that only adds to the multifaceted understanding of donors – is based on direct mail package preference.  Specifically, we use a set of scores assigned to a myriad of direct mail component “parts” (e.g. different outer envelope formats and copy, different letters, reply forms and premium inserts) as determined from a tool and survey methodology employed in the commercial sector to assist in making product development choices when the options for size, shape, color, brand and price (for a tube of toothpaste for example) can be almost infinite…just like developing a direct mail package.

(You can find a high level video overview of the tool here if inclined)

Using these scores to identify segments means we find groups who exist because they have different, dramatically so as it turns out, direct mail package preferences.  These groups exist because of package differences – it is what defines them.  The big question is can these segments be described and profiled in a way that they become identifiable and reachable in the market?  In other words, can we act on the segmentation, which is based on complex attitudinal data measuring preferences?

So far, yes is the answer.  We’ve found older, wealthier segments (these are descriptive characteristics, not defining) who prefer a expensive to produce note card package.  However, for renewal with high average gifts, the net income economics are very favorable.  This same segment has strong, negative feelings about packages with bright, colorful 4 color images on, for example, the outer envelope.  On the same donor file however, there is a segment who prefers this type of package over all other types.  This segment is younger and less affluent.  There is also readership and viewership data to further describe and target these demographics.

Identifying these market segments on your file, or in your acquisition universe (also doable) is only achievable using attitudinal data to derive the groups and demos and behaviors (e.g. readership) to define them – just the type of 360 look that most sophisticated, innovative organizations are looking for these days to better understand and market to their file with all the financial upside it can deliver.