From Customer-Segmentation to Self-Segmentation
By Tom Ryan
Writing in Advertising
Age, Michael Fassnacht claims traditional customer segmentation techniques
are either antiquated or have never been efficient enough in the first
place. Rather, the chief customer intelligence officer at DraftFCB urges
marketers to make "a stronger focus on enabling the consumer to
self-segment."
For retailers, this means
mimicking Amazon in segmenting identities by linking interests in one product
to another.
"An investment in
a smart product-affinity recommendation engine could be more worthwhile
than spending huge dollars against micro-segmenting the consumer base," wrote
Mr. Fassnacht.
He listed Amazon and
Apple as two experts at finding relevance around products for consumers
rather than "micro-segmenting consumers by any kind of attributes."
He also believes that
Facebook, MySpace and Google are behaving similarly.
"They are enablers
of self-segmentation and self-identification through group and interest
identities," wrote Mr. Fassnacht. "They do not place targeted,
direct communications at the center of their marketing activities, but
rather enable consumers to self-target by their own individual choices
and network preferences."
Not surprisingly, Mr.
Fassnacht spent considerable ink bashing customer segmentation practices.
First, he said an underlying problem with any consumer segmentation approach
is that most consumers belong to several segments rather than one.
Second, the
"static definition of consumer segments" is becoming less reliable
in today’s "extremely volatile society." Plunging housing, stock
or bank account values over the last six months could have "significantly
decreased" purchasing patterns for many consumers, said Mr. Fassnacht. But
life-changing events, such as a divorce or first child, that often transform
buying patterns "are
becoming more difficult to predict because consumers live their lives on
a much less traditional path than they did 10 or 20 years ago."
Finally, Mr. Fassnacht
argued that consumers are gaining more control over marketing and directing
the information they want to receive. He writes, "In truth, it’s easier
to let them choose and decide what is relevant for them than to predict
relevance based on any expensively calculated segment identity."
Mr. Fassnacht said consumer
segmentation will continue to play a "critical role" in marketing,
especially in identifying the right segment for a new product and incorporating
segment-specific needs into the design of the product or service.
"But consumer segmentation
and self-segmentation have now entered the stage of becoming equal forces
in today’s marketing discipline," writes Mr. Fassnacht. "And
this new reality will force marketers to better balance marketing investments
between targeting vs. enabling self-segmenting capabilities."
Discussion Question:
What have you thought about the limitations of consumer segmentation
techniques? On the other hand, what do you make of the merits of ‘self-segmenting
strategies’ as represented by Amazon and Apple in the article?
Mr. Fassnacht is absolutely right on his assessment but we tend to view “self-segmentation” as behavioral segmentation. If our intent is to drive, manage reward, recognize, and be relevant through segmentation, then behavior data is really not only the most useful, but it’s often the only set needed.
Traditional segmentation based on traditional marketing research is largely about attitudinal and demographic data. These data are used to put people in buckets with cute names.
This approach to segmentation might be great for developing copy but it’s not nearly as useful for generating store visits and sales as behavioral- or “self-” segmentation.
What’s most important for marketers to understand about consumers is the emotional currents that run underneath their shopping and buying behaviors. Matching messaging to the emotional pulse points matters. So does getting those messages in the right mediums for consumers/shoppers to react with them. Behavioral segmentation that helps enable this will still be useful in the future. But the key is to work with retailers to use the “double filter” to get to segmentation that is based on current consumption behaviors.
Both techniques have their applications–and their limitations.
Many segmentation schemes do in fact force consumers into boxes they don’t belong in. This has several ill effects ranging from lost sales and declining service profiles on the customer side to misalignment of whole business segments and uncompetitive strategy on the enterprise side.
But self-segmentation has limits too. For one thing, self-segmentation tends to be aspirational–focusing on who the consumer imagines himself or herself to be. Next, it may target the wrong indicators. It is probably more important to understand why a person thinks or aspires to be in a certain group than it is to monitor what group they claim to belong to.
The bottom line–there’s no short-cut (or substitute) for real consumer understanding.
I agree with Phil. “Self-segmentation” is often studied within the domain of behavioral marketing research. However, in my thirty years in the CPG industry I have always believed that self-segmentation is much more the reality than the more traditional yet more vulnerable, flawed, and erroneous approaches to customer segmentation. I think companies like Amazon.com have illuminated what has been truly the case for a much longer time.
To a degree, Fassnacht is correct and static definitions of consumer groups have not really worked for some time. A good example is looking at the “Hispanic” market as one monolithic group when we know how segmented it really is. I think the same thing is happening with “Asian” consumers and the sub-segment “Indian” consumers. I believe the latter is currently underserved in the supermarket industry at large–and don’t tell me it’s because they shop only at their own groceries!
Basically, this whole conversation gets back to one simple tenet–listen to what your consumers want! There is a place for complex algorithms and they make great copy, but nothing beats talking to people and watching how they shop–whether it’s in supermarkets, department stores, or discount stores.
Customer segmentations in general are often static and inflexible, especially during changing environmental factors. As Fassnacht argued, customers tend to fall into multiple categories and sometimes those categories are oppositional. Many companies prefer to rely on their own segmentations simply based on the fact that each company prefers to identify their customers in unique ways, not to mention the ability to control the segment identifications.
The problem with allowing customers to self segment is two-fold; the first of which is the reliance of customer owned segmenting. Once a customer is able to self-segment, the issue becomes the discrepancy of which segment a customer “feels” they fit into, rather than what segment actually applies to them.
The second issue with self-segmentation is the sheer amount of segmentations that can be created based on self-identification. Segmentations inherently exist so marketers can create broad-stroke advertising and promotions based off of a semi-collective group of individuals. If companies allow customers to self-segment similar to Amazon and Apple, the complexity of developing unique and customized campaigns becomes enormous. Some companies are able to capitalize on self-segmenting simply because they are based online, which allows for larger amounts of flexibility and individual customizations of advertising. Applying this technique to traditional media outlets such as television, radio and print becomes nearly impossible, based on the inflexibility of those media outlets. However, like many companies, there is still the need to mix the traditional with the innovative since it’s not necessarily the segmentations that are antiquated, but rather the media platforms for which these segmentations exist.
If we were discussing this in the 90s, I might agree, but we are in 2009. It has not been age, income, and sex for years. Even the addition of education and married status would not be acceptable today. Successful retailers understand market segmentation. They also understand that too narrow segmentation will not result in a profitable store. Extreme segmentation may work on the Internet, but not with brick and mortar.
For a store to be profitable, there need to be sales which come from a customer base. Many retailers have two or more target markets that can co-exist.
Self-segmentation sounds good, but so does market research. It is more likely to identify a fad than a trend. Because there is no financial commitment, self segmentation is more along the lines of hopes and dreams than reality. There is nothing wrong with hopes and dreams, but few can be in the business successfully.
Traditional customer segmentation can be over-simplified and static, while self-segmentation can be overly-detailed and complex. What’s a marketer to do?
The key question to ask here is “why?” Why do we want to sub-divide our shoppers into groups? Why do we want to have micro-tailored messages? The answer: relevance. Not only relevance to the shopper, but relevance to the marketer. If I’m selling toilet paper, I don’t care whether the shopper only buys low-sodium foods; but if I’m selling low-salt chips, then I sure as heck do.
So, what marketers really need are purpose-built segments. You are introducing a new line of products in a different department? You need a segment of customers who have bought your brand before and who shop that department. Your stores have increased the selection of hot prepared foods? You need a segment of people who buy frozen entrees and other convenience meals.
For most retailers, this type of on-the-fly behavioral segmentation is challenging at best, and requires too much work to make sense for all but the largest (and therefore, least-tailored) behavioral segments. We believe that with the right tools, marketers from both retailers and manufacturers can finally take advantage of chains’ vast loyalty data resources to build a segment of the right shoppers for each and every need.
To make analysis, planning, and strategizing easier, a variety of characteristics are used to segment consumers (either by outsiders or themselves) into a box. Once the box is identified, that group of characteristics is attributed to the group in general over time.
One major problem, as identified in the article, is that consumers belong to lots of boxes, change from one box to another, and don’t all fit into that box equally well. This is true whether the box is imposed by researchers or by self-selection.
Unless data is constantly collected and analyzed in several different combinations, segmentation is problematic. Those pesky consumers are a moving target.
Fassnacht has an excellent point. The other half is being able to rapidly iterate and measure marketing techniques. A static segmentation too often leads consumer-oriented companies to develop programs based on the “theory” of what will appeal to those segments. That’s true even as they move to micro-segments.
The reality is that consumers will tell you what they like through their actions. The measure of a program is the consumer response. The most successful companies will use a great engine to suggest what consumer might like (a la Amazon), but then will test repeatedly, roll out what works, and then iterate as the environment changes.
To some extent, grocers have been doing the Amazon thing for decades.
How different, really, is “Customers who bought NO BS Sales Excellence also bought NO BS Business Excellence” from having a display of Old Bay and other seafood seasonings right under the fish case?
Sometimes high-tech merely automates the obvious.
Fassnacht’s first argument against segmentation is that the “static definition of consumer segments is becoming less reliable in our extremely volatile society, especially in today’s economic climate.” True. But it’s also the best reason for companies to segment customers, and then devise different strategies for each. As he points out, there are more customer segments now than there were a year ago because economic conditions have created more of them and ever-changing cultural conditions create them.
He neglects to point out, however, that segments are opportunities. The more the better. Customer segments by their nature are subsets of the larger customer base, defined by demographics and behavior. They are never static. They never were. If for example, a consumer electronics company sees its customer base falling in high-end purchase activity, and gaining in low-priced products. That is not an argument against segmentation. It’s a signal that operations and marketing need to define that segment and treat differently than it did a year ago.
I agree that traditional methods of segmenting customers are outdated. The practice began as a way to describe consumers. The fallacy of this practice is that while consumers may live in the same neighborhoods, have similar incomes and be representative of the ethnic backgrounds of the segment, they do not purchase goods and services the same way as others with similarly described consumers. To correctly segment customers, it is important to measure buying behaviors and preferences. How will consumers respond to promotional items, how do they respond to price increases or decreases, how will assortment and space adjustments impact buying behaviors? These and other factors are the real measure of a consumer segment.
Just because a person is a college student does not mean they spend money the same way other college students do. To deliver the 4 P’s correctly, retailers should examine buying behaviors and segment their customers on those behaviors. As decisions are made in the merchandising process, the buying preferences can correctly influence/drive the available choices in a given store for a given set of consumers. We can return to the intimacy of the single store owner/operator, but it will be realized only when we understand the specific and unique needs of each group of customers who frequent a given store.
I liked Ryan’s comments on this, but I tend to favor self-segmentation based on personal behavior, rather than personal opinions or attitudes. This is illustrated here (See the fourth view of the quick trip).
I intend to write more on this later, under the heading Science, Segmentation and Creativity, but here I will simply summarize a few points. As a scientist, I like to approach things from scratch, with a clean slate. And this means having some understanding of how we “know” the world and anything in it.
The foundation to knowing, is dividing things, or making distinctions. The simplest is dividing this from that. As someone once said, the world is divided into two kinds of people: Those who divide the world into two kinds of people; and those who don’t. 🙂 I suppose this is a trivial way to divide the world. But it IS one way of segmenting, or dividing. So segmentation lies at the foundation of science.
Faced with a large number of seemingly disparate items, people, or what-have-you, the first thing a scientist must do is attempt to sort them into groups. That’s all on earth that segmentation is.
But the second element of segmentation is naming the groups. It is what I call “The Adam Problem!” What to call all these animals that we see.
So there are two essential elements in segmentation, the first is sorting, and the second is naming. The fact that the sorting can be assisted by sophisticated mathematics, can easily give false confidence in a segmentation scheme. The math is powerful and helpful, but even if the data is dead-on accurate, if the variables fed in are not the “correct ones,” it will be a case of GIGO – garbage-in; garbage-out. So which ones are the “correct ones?” God only knows, and he’s not telling.
The reality is that the researcher selects variables from data at hand, or that can be readily obtained, with no particular guarantee that it is relevant to whatever purpose is intended for the segmentation.
The selection of the variables involves a massive amount of by-guess-and-by-golly. (And judgment, too.) The mathematical processing can easily give an unjustified scientific patina to the output – but at least your sorting is now done.
The next hazardous piece is naming the segments. There is nothing in the mathematical process that spits out names for all these animals (segments.) Often, it is not entirely clear what to name the beasts, so the researcher will use their own judgment in naming them. Judgment in, and judgment out, and a lot of judgment controlling the math, too: how many segments, how distinct must they be to be considered a segment, etc.
Then the real erosion begins, when the assigned names take on lives of their own, as if the verbal, mental constructs associated with the words, were held in common by the creator of this scheme, and all the users. Suddenly the named segments produce offspring that may have only a weak and tenuous relation to data that is, conceptually, very far away. But to an entire industry, possibly, these words-names may be accepted as received truth.
On a lighter note–or perhaps not :-)–if we are to believe the philosophy of the Vedas, the Universe has a head start on “self-segmentation” and “customization of consumer experience” technology. According to it, the world and our experience of it is “Maya,” an illusion product of our mind, and we are free to create and mold it, and experience it as long as we hold the illusion.
If that’s the case, our techies and marketers have a long time to go before they climb that technology curve.