Overview
For many years, marketing professionals have relied on a set of analysis techniques designed to help them understand the demographic and psychographic profiles of their customers and prospects. These traditional segmentations are usually derived from complex clustering techniques that map rich primary research data (usually survey based) into common groups or profiles. These groups are then given highly descriptive business names and rich descriptions and provide a framework for a wide range of marketing activities. Though such segmentations can be (and are) applied to online customers, companies that have tried to map these segmentations down to the individual level (for targeting or reporting) in the online world have mostly been disappointed. In Part I of this series, I described the biggest pitfall in extending these segmentations – the near impossibility of mapping demographic and psychographic profiles to visitors about whom we typically know nothing except their online behavior. In Part II, I discussed the advantages and disadvantages of building behavioral segmentations. In Part III, I covered different strategies for joining survey data to behavioral segmentations, when each is appropriate, and why the join is necessary at all. In Part IV, I covered basic data transformations for segmentation – focusing on describing visitor-level topic interest. In Part V, I described a Functional approach to building session profiles – and how these session-styles lend a whole new dimension to behavioral segmentation. Part VI covered the role of time-based attributes and the different ways that visitor level trends in usage can be incorporated into a segmentation. In this post, I’m going to cover one of the most challenging aspects of a behavioral segmentation – presenting and describing the segments.
A great deal of the art involved in any segmentation – traditional or behavioral – is in the description of the segments. A segment, after all, is really nothing more than a statistical agglomeration. It is not a real-world concept. Members of a segment are not a team, they do not typically share any relationship at all – they are a segment only in virtue of sharing a fairly similar set of attributes when described by the variables used in the analysis. So making these groups “come alive” for marketing professionals is an essential part of making a segmentation useful. But making groups “come alive” is not simply an exercise in marketing. The process of naming and describing groups necessarily involves a simplification and distillation of key elements of the analysis. If the analyst does a poor job of this, the segments can be highly prone to misinterpretation and resulting misuse.
In traditional segmentation, it is both a profound strength and a great danger that the variables employed are so often ones we have a deep familiarity with. Describing a segment as being made-up primarily of 12-14 year old girls, living in or around major urban centers, ethnically white, and coming from families with a median income in excess of 150K can provide a vivid picture for the marketer of the target segment. But therein lies a danger as well. For the marketers picture may be far too vivid – formed mostly of stereotypes drawn from movies, TV and even personal experience. The known facts and stereotypes around demographic variables both intensify and pollute the reaction to the demographic make-up of segments.
Behavioral segmentation's tend to suffer the opposite problem. Unlike demographic characteristics, we don’t have a welter of experience with our web site visitors. Describing a segment as being made-up of visitors whose share of mind is split between educational resources and consulting services, who visit the site 3-5 times a month, who read one or more blogs regularly and who are heavily weighted toward informational pages on the web site doesn’t provide the same kind of heavily-freighted picture.
That’s good in a way, but it can make behavioral segments seem bloodless. The marketer needs to have an intuitive grasp of the segments to incorporate them into their thinking. If the segments don’t transcend their actual behaviors, they won’t achieve that.
One of the best techniques for making behavioral segmentation carry more intuitive weight is to use VOC data as an overlay. Naturally, VOC data can include almost any of the traditional demographic and psychographic variables. And while these may, in fact, be interesting, the best type of VOC overlay for a behavioral segmentation tends to be around behavioral intent.
No aspect of VOC data integrates more naturally with web analytics data than questions around intent – what the visitor is trying to do with the web site. Intent questions segment more interestingly and provide more pointed data than almost anything else in an online survey. Those new to surveys tend to over-focus on the promoter/satisfaction aspects of the survey – and believe that being able to see the paths/pages of dissatisfied visitors compared to satisfied visitors will yield dramatic insights. This isn’t always or even often the case.
On the other hand, understanding what visitor’s wanted to do and pairing this information with what they actually did is nearly always revealing.
From a behavioral segmentation perspective, the distribution of what segment members want to accomplish is one of the most powerful, meaningful ways a group can be described to a marketer.
One problem you’ll almost certainly face marrying “intent to accomplish” with behavioral data (particularly given the session-styles approach I’ve talked about) is that intent is generally captured only for a single session while behavioral analysis is nearly always cross-segment. There is no magic bullet to solving this problem. However, if you understand that it’s likely to occur, it can influence your thinking about your survey strategy (repeat solicitations and length of survey) and the way you shape “intent” questions.
Another challenging aspect of behavioral segmentation is the number of segments that are typically derived. Most traditional segmentation schemes limit the number of groups they identify to a small handful. This makes it easier to manage and understand each individual segment.
Since the techniques most often used to derive segments allow the analyst to control for the number of groups identified, the same limits could be applied to behavioral segmentation. However, it has been our experience that this does not produce crisp results.
Behavioral segmentation seems to work better when the analyst is willing to let the tools generate more segments. These micro-segments are often truly fascinating behavioral niches with great potential for targeted marketing and messaging. But, and here’s the rub, there are often too many of them to be held easily in the head. A behavioral segmentation may work best when it spins off 16 or more segments. And that’s simply too many to easily manage.
As with VOC integration of “intent,” there is no perfect solution to this problem. But we have found a couple of techniques that do help. First, individual segments can often be grouped together into uber-clusters. Some segmentation techniques (like Self-Organizing-Maps) make this very natural. The idea is to find key behavioral descriptions that may encompass multiple segments (like “High-Value Visitors") and then provide a hierarchical view of the cluster system.
We’ve also provided a listing of top segments by key business dimensions (e.g. “Most Family-Oriented Travelers, “Most-Focused Travelers,” “Most Price-Sensitive Travelers”). Segments can be ranked against each of these business dimensions so that if a marketer has a targeting or messaging problem and wants to find the most “family-oriented” travel segments, it’s as simple as looking up that list. We call this a “Zagat” view of the segmentation based on the categorization of restaurant types in the back of the popular restaurant guides. Zagat’s categorization makes it easy for users to find the most likely restaurants for a romantic dinner or a family night out. The segment rankings by key business dimension accomplish a similar function.
One of the biggest advantages to a “Zagat” like view of segmentation is that these listings stress action-taking. Behavioral segmentation is much more likely to drive targeted messaging programs than traditional segmentation and the more this can be built into the very fabric of the segment descriptions, the more value you’ll see from your effort.
No market research company building a traditional segmentation would ever overlook the vital role that description plays in the utility and success of a segmentation project. But either because it is simply harder to do or as a result of the types of organizations performing behavioral segmentation, this critical step is a common failure point for online behavioral segmentation. But careful use of VOC integration, segment-groupings, and business groupings can help a behavioral segmentation achieve the intuitive richness of a traditional segmentation while sacrificing none of the clarity and actionability that make behavioral segmentation so desirable.
Comments