Our new GreenBook Directory site is live!
Your #1 strategic guide to consumer insights.
Qualtrics X4 The Experience Mgmt Summit
Brand & Retailer tickets for all IIeX events now start at just $99! Get or give one today!

Four New Approaches To Consumer Segmentation In A Digital And Social Age

Traditional consumer segmentation is at the heart of marketing practice, yet it simply does not work that well because it is rarely very actionable. Here's how that can change.

Woman hand with fruit mix


By Joel Rubinson

Traditional consumer segmentation can be maddening.  It is at the heart of marketing practice to group consumers into segments based on their needs and self-stated psychographic profiles and to then attempt to target a high priority segment with new offerings and advertising. Yet, it simply does not work that well because it is rarely very actionable.

I remember once conducting a massive new product forecasting study where we were asked to put segmentation questions into the study.  There was virtually no difference in purchase intent across the segments!  Tell me again how this segmentation helped with innovation?

Remember Best Buy’s commitment to customer centricity? The segments simply did not lead to impactful redesign of their stores.

Take an attitudinal segment and try to target them on analogue TV.  You often wind up looking at demographic indices for that segment to place media which turns the rifle shot into a scattershot.  OK, you hit the mark but you hit a lot of other targets too. And, while I am airing our marketing research dirty laundry, have you ever tried to score the same people into attitudinal segments at different points in time? My experience suggests that you will probably only classify 50-60% into the same segment.

Jim Stengel, the former CMO at Procter said at a conference in October 2012, “Get close to the consumer and do something with it”.  In segmentation, marketing research never seems to get to the “…do something with it” part

…but here’s how that can change.

Segment moments. I am much more interesting to Ford or General Motors when I am looking to buy a car then right after I make the purchase, am I not?  I happen to be on a diet now which makes me much more interesting to Atkins, Dukan and Weight Watchers than I was a month ago. Moments segmentation is better for innovation and for media strategy intended to influence the path to purchase. In a digital and social age, moments become directly targetable because I, the consumer, do things differently on my self-guided tour of the internet depending on my current goals, giving out forensic signs. I seek out different content, I search for different terms, I like different things on Facebook, and different products show up on my frequent shopper data.  All highly targetable without needing to water things down with demographics. (For an example of moment segmentation on understanding smart phone use and motivations, click here.) (The supplier was InsightsNow, Inc., I was the consultant, and AOL and BBDO were the clients.)

Segment for ad targeting based on brand loyalties. Increasingly, we can merge digital and social data with frequent shopper data for ad targeting. (Facebook just cut a deal with Datalogix to do this for example.) Rentrak and TRA have each merged TV viewing with frequent shopper data. The “so what” is that a marketer can now target their advertising to “switchables”.  Who are they?  The consumers who buy your brand occasionally but not most of the time.  You will find a much higher response to advertising and promotions from switchables than from those who are completely loyal to either you or some competitive brand.

Segment people as shoppers. Do I plan purchases or decide in store? Do I like to explore to find new meal ideas?  Do I like to sample the gourmet cheeses? Do I like to sniff the fragrances of air fresheners and shampoos? Do I probably have an infant at home given diaper and formula purchases? All of these have clear action implications for category adjacencies, store layouts, and specific shopper promotions delivered in a customized way, increasingly via mobile apps. Is there a consumer attitudes segmentation that can claim the success of what DunnHumby did for Tesco in the UK? Not that I know of.

Segment people based on targetable interests and values. Rather than create a psychographic battery of questions for segmentation and HOPE that we can target segments, why not flip this around? Why not analyze the interests, cultural values, and lifestyle characteristics that are available via Facebook or Google profiles and create segments on factors that reflect those actionable characteristics? That way, you can take your segmentation and do something with it! Furthermore, every ad campaign becomes a test that you have the segment right that you are targeting because they should exhibit greater response.

Final point.  Traditionally, marketing research seeks to create a small number of segments that each represent sizable opportunities.  This is still important for motivating innovation ideas, but when it comes to ad targeting, you can have many micro-segments as advertising is micro-targeted…served up one impression at a time in search, and with real time exchanges.

Please share...

17 responses to “Four New Approaches To Consumer Segmentation In A Digital And Social Age

  1. These are hardly new and apart from the social media overlay have been in use in segmentation for as long as I can remember Anyone in retailing, fast food and on premise alcohol consumption have been segmenting on these dimensions for years. Who didn’t know that the purchase occasion, the consumption occasion, the shopping expedition and the brand loyalty segments were not useful segmentation models? Liquor marketers knew how critical the drinking occasion was as far back as the 1970’s and most research in the on-premise environment was about that. Since the 80’s every major retailer has been working with brand managers to find ways to affect loyalty shifts. Really it is almost ludicrous to suggest targeting on values is anything remotely new. This stuff was around in the 70’s. Am I missing something here or is this just another social media puff?

  2. True these approaches are not new, but it is false that everyone has heard of, understands or uses them. There are many ways to segment some of which fit specific circumstances better than others, so a reminder is not a bad thing.

    There are arguments for segmentation by “jobs-to-be-done,” outcomes sought and need-states.

    There are also arguments on the limits of segmentation [see the book “How Brands Grow” on heavy users], discussions on how method dependent segmentation can be and the possibility that we buy for reasons unknown to us/ not the reason given in the study.

    Hope that more is published on the subject since large, expensive, underused segmentation projects have occasionally been “The last project the researcher” does before exiting the firm. .

  3. A one-size-fits-all approach to segmentation (be it one generic application framework for all companies, or one custom solution intended to serve all parts of a single business) is doomed to fit no one very well. We find that the most actionable segmentations are those that are custom-crafted to meet the priority applications (e.g., strategy development, innovation, communication, targeting, etc) of a specific business in light of its unique business situation.

  4. Agree that no one size fits all. Some of the more obvious approaches to segmentation are often neglected e.g. price/ value [consider the automotive industry, P&G value vs. premium products, Intel chip series], pack size [e.g.affordable single packs for less affluent households in asia, Sam’s Club value packs, etc.], design, distribution channel, technology used, etc. It can be useful to list all the potential dimensions for segmentation before choosing one as a focus.

  5. I like Joel’s set of paradigms for segmenting the market, for one big reason – it provides some substantive qualities for research practitioners to look for as new marketing opportunities. We need more of these, and as marketers we all need to think through potential segmentation solutions from time to time, not just focus on whether we’re using the right statistical methodology. The latter is of course essential, but without first using “marketing imagination” to guide the segmentation focus, our tools are useless, as Bill Tanner suggests above.
    The heart of our practice, as marketers, is to (a) target the groups in the population who are (b) especially responsive to the substantive content that a good segmentation provides. These four offerings, can and should offer guidance as input to the statistical machinery behind the segmentation solution. In other words, we may have a great car that will take us to new places, but without an experienced driver…I’m sure everyone gets the point here.

  6. Joel, this is a very interesting article. I like some of your ideas about other analytic techniques to improve on segmentations. However, why “throw the baby out with the bath water”. Segmentations combined with psychographics and demos can be highly effective depending on what your looking to know and what you are analyzing, In the past some segments which I either created or borrowed from like “Power Players”, “Road Warriors”, “Schedulers”, and “Sun worshippers” was highly effective in understanding travel behavior for hotels and airliners. It was so effective, in fact, that the NY Times wrote a whole article about the segments. Techniques and ways of analyzing data can always be improved, but let us be careful about criticizing tried and true methods.

  7. Thanks, Joel. Agree with you concept of targeting a moment. We talk about targeting based on where the consumer is within the purchase funnel.. And have had great success by crossing geo-demographic segments with purchase funnel markers.. To hit the right person with the right message and level of investment, recency and frequency as they move through awareness, consideration, intent, purchase.

  8. Joel’s message to think about how you might want to use a segmentation as you think about what characteristics to look at is a good one. If you need a segmentation to guide shopper marketing decisions, for example, certainly talking to, learning about, and segmenting shoppers makes sense.

    That’s only part of the actionability issue though. Ideally, you want to know which group of shoppers or customers to prioritize as you think about where and how to spend your marketing budget. And the only way to do that is to figure out what characteristics are predictive of ROI to the brand.

    While perhaps helpful when it comes to developing messaging, just looking at social media data and developing segments around professed interests and attitudes doesn’t necessarily tell you anything about how much of your time and resources to put into developing message for any of those groups.

  9. Here’s my contrarian POV- To me, the “new” segmentation is based on new sources of information. Location information is allowing point in time identification of the micro-segment that is in your store. Web browsing behavior on their tablet segments them on a purchase funnel level and product level. Third party data sources tells you their demographic, attitudinal, and behavioral profile. Graph analytics segments according to their social network and how influential they are. Given our massive compute power, and the move to ‘in-memory’ systems, we’re able to marry our analytic environment to our transaction environments, combine all of those variables in our model, score the in-store prospect in real time, serve up the next best action, and make targeted offers on the spot based on the their “segment of one”. Now we’re talking new age segmentation!

  10. And with that, @Tony just encapsulated the very essence of why “big data” analytics will lead marketing decision making and MR will be relegated to one spoke on the wheel supporting that process. It’s not an unimportant spoke though; the “why” will still be vitally important, although the “who, what, when, where, and how” that MR used to be one of the main sources of will be driven by the model Tony outlines.

  11. what a great conversation thread on my blog! thanks all. Bill Tanner’s link is very important, especially the part about only 50% of respondents being classified into the same segment if they retake the test. I saw that sobering finding repeatedly on the attitudinal segmentations I did before I got religion. Lenny and Tony totally get what I was trying to communicate; these are NOT the same data researchers have been working with.

  12. It’s true, too many consumer segmentations lack consistent use within businesses. I think this is because many segmentations are a bit faceless – often tricky to recall details & differences and therefore difficult to socialise within organisations. Okay, some people write eloquent pen portraits and even illustrate typologies – but these ‘definitions’ are often rooted in the category and tell you little about who these people really are. And unless they are behaviourally based, they are often disappointing at separating groups on a brand share basis too. Since the troubles with segmentation are clear, there are a number of firms doing radically different things to get to better results. At BrainJuicer we’ve been helping to bring segmentations to life by understanding consumer targets beyond the category and looking at how these key segments live their life as a whole, as defined by social media. As well as doing the business in terms of insights and idea generation, DigiViduals® help get teams behind key consumer targets and give segments a life way beyond the research de-brief. But there are surely others.

    Thanks Joel throwing down the first gauntlet! Any new ways of lifting the often sterile outputs of consumer segmentation has got to be a good thing – let’s hear about a few more!

  13. Re Sharon’s request – anyone doing anything with visual segmentation.

    At a consumer segmentation level check out who segment individuals according to their responses to quizzes where the answers are images rather than text and the orientation is to psychological profiling. They have an Emotive Targeting segmentation in an Analytics tool that you can tag on your website

    I guess a more tactical approach is dynamic creative optimisation which effectively segmetns audiences based on their response to the visuals they are viewing in ads. Example specifically for DCO

  14. If I want to do customer segmentation for telco, in the same way as you have referred based on people interest or characteristics, what would be the most common aspects I should be looking at? I am situated in Myanmar.

  15. I just love being six years late to a conversation, but I am doing some background work to create segmentation templates for a small MR tech company, and I ran across this. General Mills, a food CPG, has done lots of segmentation studies over the years, and at some point there was challenge from business unit presidents about the value. They are/were expensive. The average shelf life was 18 months, and then they were relegated to a dusty shelf because no one could figure out how to take action on them. I was the head of analytics there (I retired a half dozen years ago), and at one point I tried to ban them. Instead we did a review to better understand which segmentation studies were successful/had impact, and why. The factors that differentiated between actionable and non-actionable projects had little to do with either analytics or data sources. The factor that was most significant was whether there was a formal organizational mechanism for communication and implementation. Think workshops in which teams are persuaded to own the results, and in which teams work through the mechanics of implementation (now, how do I build the digital media plan). The next most significant factor was upfront homework; often plain old fashioned focus groups that did a good job of delineating customer behaviors and motivations, and language. Surprisingly, whether there were upfront conversations with key stake holders was not a significant factor. The most “techy” factor had to do with the experience level of the analyst. Experienced analysts had a reasonable understanding of the business context and model. Importantly, they were very creative in their approach to analysis; they iterated through multiple statistical analyses for variable screening and segmentation, they sometimes created new variables.

Join the conversation