Discerning habits

If we condense the time that modern man has walked the earth into a single day, then supermarkets have been around for less than a minute. Yet in that same amount of time, marketers have radically transformed the way customer relationship marketing is conducted.

Before the mighty food and household emporia dotted the landscape, we had local shops, which knew their customers, talked to them regularly, understood their tastes and determined whether they had relatives coming for Christmas and would be buying a bit extra. Likewise, the man from the Pru, doing his weekly rounds, could spot that Fred Smith at number four had a bit of extra cash, because he had bought a new car and seemed to be doing a lot of decorating lately.

But in the minute just gone by, the personal element of customer service has disappeared: shops, financial institutions and other businesses have lost touch with their customers.

However, in the past few seconds, help has been at hand: computers and EPOS systems have made it possible to gather large amounts of data about the transactions being made. Once again, institutions can know their customers’ food preferences, and when they last bought a car.

Market research and other data collection methods are also accumulating mountains of information. The task for the next minute is to translate that data back into useful knowledge about individuals.

While there are plenty of data analysis tools around, research from AnswerSets, a developer of customer segmentation tools, found that half of the 1,000 surveyed sales and marketing staff thought they got the answers to marketing questions only “occasionally” right. Fewer thought they got answers right “most of the time”. Incompetent technology or poor quality data were blamed.

Dun & Bradstreet European development manager Richard Lloyd says it is not surprising that data is often of poor quality, considering how some of it is gathered. For instance, if you collect a prospect database from a campaign where consumers fill in a coupon and return it, most coupons are too small to write clearly and at length. To add to the problem, the information at the receiving end may be keyed in under pressure, and quite possibly by people outside the UK who may not understand the country’s addressing system. The potential for errors, therefore, is great. Equally, it is not unknown for sales people to make up some of the data.

On the other hand, data can come from too many sources, so that the information cannot be integrated. Everyone within a company who has some contact with customers should have at their fingertips all the information the company has about its customers. If that information is to hand, dealing with customers and answering their queries is much more effective.

Alchemetrics chief executive George Antoniou says: “It’s pointless having data dotted around in different company departments. You need a whole view of what your customers are like, not least because you don’t want to mail them for a product they already have.”

Even when data is reasonably reliable, for example when it comes directly from transaction records, it can soon become unwieldy. Todd Merry, a partner at relationship management agency Miller Bainbridge, says: “Too much data is being collected. Transactional data at customer account level is too dense to analyse with today’s tools on a regular basis. One-off efforts can yield results but are outdated as soon as they are finished.”

Merry says American Express, for example, makes good use of data, but it has far more of it than it is able to use: “It probably uses only about five per cent of its data,” he says.

There is simply too much data to move around, he adds. If all the transactional data was sent to the headquarters in Phoenix, Arizona, to be sorted, it would take six months for the round trip. “It has created a monster that’s not accessible,” he concludes.

As well as being inaccessible, a massive database is also costly. A cost of &£1 per entry to clean up a database is manageable when it contains only 50,000 entries. But when you have millions of members, it turns into a hefty bill.

However, Merry says: “It is better to have the data than not. Preserving past data is better than tossing it out – if nothing else, it may show trends or seasonal patterns which are useful.”

Jan Bourke, head of data planning at marketing consultancy Perspectives, disagrees. “You must be able to turn the raw data into something of value. The key is to be ruthless about the reasons for collecting it.”

When Boots launched its Advantage Card loyalty scheme in September 1997, it followed two years of research. Boots customer insight strategy manager Helen James says: “The information is changing the way we operate. We can now analyse the behaviour of groups of customers, such as the effect of marketing activity and the impact of promotional offers on behaviour over time, and make decisions about store layout, ranging and promotions based on this input.”

One example of a surprise link that has emerged from the Boots data is the number of people buying films and photo frames with new baby products. “Like many large retailers, we are organised along product category lines, so it never occurred to us to create a special offer linked to picture frames for the baby products buyer, yet these are the very things new parents are likely to want,” says James.

But what of the future? Will the data being collected now still be of use in ten years’ time?

Kevin Codron, marketing manager at IBM Global Business Intelligence, which supplies Boots with its systems, says: “Of course, some data will be obsolete but much of it will be valid. As we understand more about the lifetime value of our customers, we realise that data covering the past six months of buying history is of limited use.

“But a broad history of data held in a consistent format and analysed in the right way will provide a much deeper insight as customers move through different stages of life. The buying behaviour of a student today will be very different from the same person in 15 years, when they perhaps have a career, live in a different area and have a young family. We need to understand their needs and preferences as they move through these stages, recognising the consistencies and the new characteristics, and tailoring our business to these triggers.”

Boots now aims to identify which customers it should value and retain, and which would be more valuable if it focused on them more. It may well be more profitable to encourage existing customers to buy deeper into the range than attract new ones, encouraging them to, say, buy some dental floss with a toothbrush.

AnswerSets chief executive Nick Kellet agrees that existing customers are the key to future profitability: “We should be making it hard for them to leave. That means understanding and learning so much about them that it is very inconvenient for them to go anywhere else. We should make staying so easy that the action of leaving is like shooting yourself in the foot.”

Attitudinal information willbe vital too, although collecting qualitative data can be expensive. Market research gives Boots, for example, an understanding of what is driving customers’ behaviour that appears on the database. It is pointless directing a major marketing effort at people whose attitudes mean they are unlikely to become more valuable.

Mark Smith, marketing director of database consultancy Quadstone, believes one-to-one marketing is only a step away. “Most retailers which do direct marketing look at broad groups,” he says. “But they should look within those groups at the actual people who buy. Contact with individual customers is better than static demographic information.”

Smith believes the next big thing is e-commerce. “It enables you to change offers by the hour and watch whether or not people buy,” he says.

E-mail also offers an incredibly cheap way of getting in touch with people, compared with direct mail, suggests Shaun Doyle, chairman of campaign management specialist Intrinsic, which handles data for Sainsbury’s, among others.

Looking to the future, Doyle suggests that pattern search techniques will improve radically as the tools to analyse behaviour become more sophisticated and reliable. He, too, thinks that attitudinal data will become more important to marketers, with loyalty card application forms asking more psychographic questions to create better prediction models.

However, Doyle warns that data protection legislation may become more stringent, making it increasingly difficult for marketers to collect and keep customer information. It is already happening in Germany, for example, where the government has introduced limits on how long data can be retained.

“People you don’t know hold in-depth information about you,” he says. “Consumer pressure may force the Government to clamp down on organisations which abuse the information they hold.”

Doyle also suggests there is a copyright issue here too. There are currently unanswered questions about who actually owns the information on a person’s marital status and date of birth. He envisages that consumers may recognise the value of the information about them and form lobby groups to pool data and charge for access to it.

All the more reason, then, for organisations to discover which information is the most valuable, and learn to recognise patterns within it that will help them to sell in the future.