Do direct marketers know what they are talking about? It might seem like an odd question to ask. After all, the DM industry has had to invent swathes of new words to describe the processes it carries out. And it has benefited from one of the most consistent education programmes of any marketing sector – the Institute of Direct Marketing has put 3,000 people through its diploma course, with a further 25,000 attending training sessions.
Yet there is much evidence to suggest that many direct marketers still have not grasped some of the fundamental issues in their business. The difference between segmentation and targeting is one. How cluster analysis can be affected by missing or inconsistent variables is another. And actually applying what the data tells you to the way you market is the single biggest and most frequently missed opportunity.
When Professor Stan Openshaw gave the annual IDM lecture this year, he implied that many in the audience were not up to the mark. Try this opening remark for size: “I am going to try and persuade you all that you are being dumb in what you do.” He had a specific reason for using the word dumb – that the way targeting is being practised at the moment has no feedback loop.
He also had a specific complaint – that the power of geodemographic targeting was being undermined by trying to carry it out at too low a level of geography. “Smaller building blocks tend to destroy the neighbourhood effects that make geodemographics work,” he said. But he was also critical of existing systems for being dumb and not allowing dynamic remodelling. The ideal would be to recluster the data for every single campaign and use whatever number of clusters gave the best discrimination.
“The Prof makes a good point,” says Andy Brown, business development director at Crazy Horse 1842. “He does illustrate a basic truth about DM. The urge to over-analyse through targeting and to over simplify – the belief that when you have noted a reaction once, it is likely to continue to be the reaction in the future – have their roots in the same soil.”
While acknowledging that Openshaw’s idea is over-reliant on computer processing power and does not address the issue of cost, Brown says: “The best results for everyone in marketing will be when DM practitioners realise that sometimes the macro factors have as much influence as the micro.”
This may sound like an arcane dispute over statistics, and in one sense it is. But to put it in simpler terms – Openshaw represents the resistance, fighting against a superior force which has seized power. The geodemographic camp, for a long time the establishment, has been swept aside by the lifestyle data owners, who will eventually know everything about everybody. In their fervour to support the new rulers, many direct marketers have failed to question just what underlies their grip on power.
What Openshaw says is that, “data errors and uncertainty levels are greater in lifestyle databases than in out-of-date Census data”. If you are trying to build a segmentation system that allows you to identify your real customers within the population, such errors risk distorting the picture.
Ironically, he is supported in this view by Mark Patron, managing director of Claritas which, along with ICD, represents the new establishment of lifestyle data owners. “If you know something about somebody, that’s it. Once you start to aggregate data, you get into the fuzzy logic world,” he says. Segmentation is about aggregating data so you can spot those clusters. It is used by marketers to make absolute decisions – do we want that type of person as a customer – so anything which skews the clustering can have a big impact.
But this is also why direct marketers have become critical of the classification systems on offer. While appearing to offer sharp differentiators between groups of consumers, they are actually covering up the reality. “Clusters are not clear cut, because they can’t be if you are using cluster analysis techniques. It is not about a person being XYZ. You get characteristics pulling the data into a centre point, but the next cluster may be only slightly different,” says Andrew Greenyer, director of analytical services at The Database Group.
It is a view supported by Adrian Jarvis, senior data planner at Insight@TMW. “The benefit of building your own classification is that you don’t need to package each cluster neatly with a name. Rather than trying to force customers into clusters, you use the natural breaks,” he says.
This issue of forcing data clusters is one of the main criticisms of commercial geodemographics, which package the population into conveniently named categories. Jarvis points out that underneath, “you may have hidden clusters which, when you use a system, don’t stand out. When you let the data lie naturally, they can appear.”
Another argument against using geodemographics is that they can often cause marketers to forget about other basic macro factors which might have an impact. These would be responsiveness, broad geography (rural dwellers are more likely to respond than urban ones), and wealth.
“Wealth is one of the most important factors in a model, but one which geodemographics can only broadly indicate,” says Paul Robinson, director of SDM. “As geodemographics are only indicative at the aggregated level, then their power will be lost if applied to neighbours rather than neighbourhoods.”
This is precisely the point which many direct marketers currently appear to have forgotten. The arrival of large-scale, lifestyle databases, which carry individual details, has encouraged the view that companies can target a prospect by their characteristics. If your product appeals to golfers, find those people on a database with an interest in golf.
So seductive is this idea – and so in line with the way direct marketing as a whole has been selling itself – that one important fact has been forgotten. Targeting is not the same as segmentation. Targeting is glorified list rental – picking out the variables which suggest a person will be interested in your offer. It also tends to downplay the importance of those neighbourhood effects which Openshaw believes derive powerfully from geography.
Ian Liddicoat, micromarket- ing director at LVB Draft Worldwide, says that does not matter: “Postcodes don’t buy anything, individuals and households do.” He recognises that lifestyle data may be flawed, but equally, geodemographics may not show any discrimination at all between consumer groups for some products, especially packaged goods.
“You can make geodemographics work for packaged goods products, but only when you have filled in purchasing activity, because the volume of products sold varies tremendously from one house to another,” he says. Two households next to each other may be identical in every respect – age, income, lifestyle interests. But if one has children and the other does not, their consumption patterns will diverge hugely. Geodemographics alone would not pick that up.
Likewise, George Antoniou, chief executive of Alchemetrics, says: “Relevance is so importance. Do you own your own home? If you are sending out a mailing for secured loans and only five out of 20 homes in a postcode are owned, it would be crazy to mail those other 15 because it is wasteful.” He suggests that geography should be considered as just another variable alongside financial status, lifestyle and lifestage.
The DM industry’s response to the problem of discriminating at a finer level has either been to rush towards individual-level targeting, or to try and force geodemographics down to a lower level of geography below what it was intended for. Patron warns that, even with the richness of lifestyle data, there are risks that skews will emerge from refining systems too far.
“It is an intuitive thing to go to the lowest level, but the data works against that. You have the neighbourhood effect and something which is almost the opposite of a regression to the mean. What happens is that you might find two postcodes covering one block of flats. Using lifestyle data, you may find a group in one postcode which is different enough to a group in the other to convince you they are actually different,” he says. But the danger is that the surveys used to create those clusters may simply be at opposite ends of the distribution curve – the reality is that the block as a whole has more in common than in difference.
For marketers carrying out small volume activity, none of this might matter. Jarvis notes that a five per cent variance in a total audience of 200,000 may not make a lot of difference, since eight per cent of any given data set are likely to have moved or died in any case. But at a national scale, even one per cent variance could mean 200,000 people.
The apparent refinement of lifestyle data targeting combined with the relative crudeness of many marketing objectives has led to some of the sloppiest thinking in any marketing discipline, however. “In my experience, there are customers who segment and realise they have five target clusters. Then the artwork comes in, and they have produced only one pack which goes out to everybody,” says Jarvis.
This is what Lester Wunderman has described as “mini mass marketing”. Feeling confident that the segmentation process has made them intelligent about their prospects, marketers slip into the same old shiny salesman’s suit. Their patter varies not one iota between one group of prospects and another.
That is why a campaign like that to launch the Goldfish credit card stands out. Not only has the target market been segmented, the marketing has been too – a huge range of different creative messages have been developed by DP&A to talk to the different characteristics of each cluster.
It is hard to blame many direct marketers for missing the point. Even Openshaw managed to miss out the difference between segmentation and targeting, probably because he has only ever done the former, as one observer noted.
Segmentation is a strategy, targeting is an outcome. The latter flows from the former and should feed back into it. The problem is that too many direct marketers are locked into an approach they know already works, and only go after a market they know already buys from them. As Patron succinctly puts it: “If you only own a hammer, everything looks like a nail.”