Developing an effective segmentation may require a blend of behavioural and attitudinal data, argues Pete Anderson, planning director, Underwired
The nature of mass consumerism has changed dramatically in recent years and, in sync, the landscape of mass marketing and broadcast media has changed too – the identity and the ‘voice’ of the individual have become ever more prevalent, with the feeling that individuals can make big differences. Consumers have the power either to make or break a brand, through advocacy or criticism.
The problem is that, although the world has moved on, generalisations are still perpetuated, even in our industry, where recognition of the individual and the importance of ‘one-to-one’ dialogue are considered almost sacred.
Digital space provides expanded flexibility in achieving targeted and ‘one-to-one’ marketing, with email and SMS providing cost-effective communications. Consequently, some companies are making progress in terms of segmentation and ‘relevance’. However, a startling number of advertisers and agencies alike are still not practising segmentation or targeted communications.
So why, in an enlightened, insight driven industry, are we ignoring the mountains of research, consumer feedback, trends, market forces? Is it the ‘elephant in the room’ syndrome – it’s staring us in the face but we choose to ignore it? Or are we just guilty of defaulting to the path of least resistance, or not really understanding the subject, so that the ‘hype’ is just a sales pitch with no substance?
If it ain’t broke...
A common justification for not doing something is the old adage, ‘if it ain’t broke, don’t fix it’, driven either by a real (although sometimes flawed) belief that this is the best way and it can’t be improved, or by sheer laziness or plain reluctance to deal with something that feels ‘too complicated’ and outside the comfort zone.
I once read an article by a respected practitioner in direct marketing who stated that keeping things simple was what it was all about. This is fine to a degree; over-complication is not a good thing, but they claimed that we seemed to be moving into an era of ever more complex ‘scientific marketing’ – analytics, modelling and segmentation – which is ‘over-complicating’ things. That’s like burying your head in the sand: ‘I don’t want to deal with it because it’s difficult.’ Well, whether we care to recognise it or not, the devil is (definitely) in the detail.
Thankfully, their views do not reflect the trend across the whole industry, but the subject of segmentation is still quite a broad church and open to a variety of interpretations and methodologies.
The challenges of segmentation
The problem is, there is very little discussion about how and what segmentation should be used, just that segmentation is important and that advertisers should segment their customers. Add to this that most brands aren’t fortunate enough to have a robust, comprehensive customer database that has a significant volume and depth of consistent data to apply a workable segmentation model without breaking into a sweat.
The challenges are many and varied, and there has to be a real belief that the work required will deliver something that is not only effective, but executable. There really is no point in demonstrating something that looks amazing but cannot actually be applied to a communications or customer relationship management (CRM) programme. Having a model that helps all concerned identify and describe the consumer is fine, but not being able to do anything meaningful with it makes the process pointless.
Direct marketers were early adopters of segmentation, simply because the whole principle of direct marketing is about targeting and relevance. Direct marketers are also in the best position to be able to practise and execute segmentation, since DM relies on data – customer-related data, that can identify anything from demographics (what customers look like) to behaviour (what they buy, how frequently, how much they spend, and so on).
Looking to the DM sector helps us to understand the value and the effectiveness of segmentation. The problem is that most companies find it a challenge to gather consumer data at the level of granularity required both to create and execute a realistic, workable segmentation.
Take the FMCGs (packaged goods) sector. Gathering customer data is a real challenge here since consumer purchasing is made through a retailer, usually a supermarket and, until recently, most retailers gathered only till information and transactional data that weren’t attached to any customer information. Retailers like Tesco created initiatives to address this challenge through loyalty cards, for example, Tesco’s Clubcard, while other retailers quickly followed suit with their own loyalty cards; but the challenge still exists for FMCGs to create their own customer segmentation models in order to influence category and brand purchasing behaviour.
Even where segmentation exists, the design and construction of these models can be questionable. Add to this a lack of understanding of what segmentation represents and it is clear the whole area is fraught with challenges.
Breaking into segments
The point of segmentation is to group consumers into identifiable and manageable groups – groups that display characteristics in common, whether that be demographics, channel preferences, behaviour, values, attitudes or motivations.
Far too often, though, client and agency marketers fail to remember that segments are just broad groups – groups about which we still make generalisations, albeit more targeted generalisations. Segments are not absolutes, they are not groups of consumers who are exactly alike – they are just more similar in certain attributes than if they weren’t grouped together. An example might be a demographic value, where a segment could be typified as families, with four kids, living in London and the south-east. These values are not going to occur consistently for every consumer in the segment: it just means that the segment has a greater likelihood of comprising households with those values. So, for instance, the segment might also contain some households with two kids instead of four. It just means that the segment will be ‘more likely’ than the average (whether the population, the customer base, and so on) to contain households with those traits and values. This depends how precisely the segment has been defined: if ‘4+ kids’, for example, is specified as a key variable, it ought to mean exactly that.
Originally, segmentation was designed around demographics, because what consumers looked like was the main criterion for grouping them together. It was also a fair bet that if you belonged to a group of consumers that were, for example, a middle-income household with two children, both adults earning, living in the South East, you’d behave much the same as the next household with the same profile. The use of ACORN classifications (still in use, but now largely superseded by MOSAIC, both being geo-demographic classifications of residential neighbourhoods, placing groups of households together based on their similarities at postcode level) to broadly describe an audience based on their similar life-stage profiles was, and still is, widely used for broadcast and ‘above the line’ media planning, where awareness activity has traditionally relied on mass-marketing techniques to reach broad groups of consumers.
But broad segmentation descriptions of an audience, like ‘ABC1’ consumers, while appropriate in certain circumstances for media planning and the like, tend to be overused and are increasingly inappropriate these days for describing groups of consumers who need to be targeted using integrated activity. After all, ‘ABC1’ broadly describes the majority of UK households earning a decent income. Not very targeted or specific, yet it’s an audience description that still appears, albeit less frequently, in client briefs to agencies where integrated or relationship marketing techniques are required.
The challenge for the agency planner is then to drill down, to gain greater understanding of the audience groups in order to provide those ‘real’ motivational insights that inspire great ideas and creative. At Underwired our challenge is to find a way for our clients, strategically and creatively in a digital space, to break down barriers and find a way into consumers’ hearts and minds. How to achieve this is the most powerful insight that we can find, the insight that helps us understand what makes a consumer tick. The ‘thing’ that makes a person engage with a brand, consume a brand message, and then take action; it enables us to connect the strategy to the creative, helps us find that big idea that can break down the barriers to brand engagement, and gets consumers to open their wallets.
The output, the manifestation of the idea, is the single-minded proposition. Focusing on one single-minded proposition for a broad audience is all very well for a long-term brand or awareness campaign played in an offline, above-the-line space, but consumers are more savvy and selective than ever – and so single-minded propositions to an audience of one large group can frequently alienate more people than they engage.
The concept of single-minded propositions still works. But now, thanks to segmentation and ever-more-targeted communication channels, different ‘single-minded propositions’ can be targeted to different audiences. So now the single-minded proposition becomes relevant to consumers in each segment, which means that more people are engaged, yielding better response, less wastage, and greater cost effectiveness. The result? Improved ROI. Do this on a regular basis and it becomes a ‘relevant’ dialogue with the consumer – a relationship (so now we’re talking eCRM). The result? Improved CLTV (Customer Lifetime Value).
In an ideal world...
So what is the ideal? What is the most effective way to segment consumers into groups so that communications can appear as relevant and specific to each individual as possible?
The ideal will differ, depending on circumstances – the business and marketing objectives, the marketplace, the brand, the customers, communications channels, available data and research, available resource and budget. But, in my experience, segmentation works best if the main discriminator is the most significant factor that has an impact on response and purchase – and whether this factor can be used as a means to define and group customers into clusters of similar traits.
Rather than demographics, it’s preferable to group consumers into attitudinal and behavioural groups. These are more predictive factors, dynamic components that have a more direct correlation to a response or an action. For example, if a barrier to a product is a perceptual one, then the most significant way to segment the audience could be by attitudinal type – in other words, ‘I only buy products I trust...’
In addition to segmentation, brands should still strive to reach consumers with some real data-driven, one-to-one communication in order to optimise the consumer’s relationship with the brand: but segmentation does provide a cost-effective means to target a group of customers with a message that is common to them, and therefore relevant. Used as a filter, segmentation can be even more effective as a means to isolate a group of customers – in order to then target further within the group, pointing even more relevant messages to sub-sets of customers within the segment. By using attitudinal and behavioural data as the main criteria for identifying segments, the most dynamic components are being used, ensuring that the targeted message takes into account all customers in a group that have common motivations and behaviour.
Hearts and minds informing communications
By including attitudinal data in the mix, a strategy for communication and investment can be made, potentially a brand retention route for the high-spending segment, and possibly a couple of routes to test for the other. One may be to test communications with testimonials from similarly profiled consumers, or to further identify specific ‘barriers’ through a customer survey for this segment, and communicate appropriate messages accordingly.
Using traditional profiling and demographic segmentation techniques, the second segment could have appeared potentially as valuable as the first. The problem is that if the attitudinal barrier is an absolute factor in determining behaviour that can’t easily be broken down or changed over time, then the level of investment in this segment is not going to be very robust.
Finally, it is worth remembering that, just because customers fall into segments, it does not necessarily mean that they will always be in that segment – behaviour can change and alter over time, so segmentation should be dynamic; and just because a segment exists, it does not mean it should. Just like profiling, segmentation can sometimes identify customers that should not be there, that actually cost money to keep and may never move up through the value chain.
One thing’s for sure, good segmentation is always going to perform better than no segmentation at all – because ‘relevance’ is everything!