In reception at the west London offices of Dunnhumby there are four clocks hanging on the wall, showing the time in London, Cincinnati, Seoul and Melbourne. They are there, says Martin Hayward, to impress upon visitors and staff how much the company, which specialises in consumer data, has grown over the past decade. “We could justify having Toronto, Paris, Cape Town and Shanghai up there too,” he adds.
Drive to the head office of the supermarket chain Tesco in Cheshunt, a drab commuter town north of London, and you will see that the retailer, which owns most of Dunnhumby, also has a clock wall. This has 12 faces, telling the time from Seoul to Warsaw. Soon Los Angeles will be added; Delhi might be next.
In the 12 years since Tesco hired Dunnhumby to help with its customer loyalty card, Clubcard, Britain’s biggest retailer has set up shop around the world, while also becoming a giant on its home turf. Every four weeks, two-thirds of British households shop at Tesco. It takes nearly ₤1 in every ₤7 spent in the nation’s shops - no one else comes close. And part of that success is down to Clubcard, which has tracked the shopping habits of up to 13 million British families for more than a decade.
“Contrary to popular belief, Tesco’s most significant competitive advantage in the UK is not its scale,” Mike Tattersall, the retail analyst at Cazenove, wrote last year. “We believe that Clubcard, which conveys an array of material benefits across virtually every discipline of its business, is Tesco’s most potent weapon in the ongoing battle for market share.”
Martin Hayward, who has been director of consumer strategy and futures at Dunn-humby for the past two years, settles down in his chair in his small office around the corner from the lifts. “We use your purchasing behaviour to create a picture of the kind of person you are,” he explains, looking - in a pink shirt with the top buttons undone - far more like a media executive than a stats-man. “We often liken what Dunnhumby does to standing in the queue at the checkout. You are looking at someone’s shopping, you know, and you’re thinking, ‘Good God, you’re not going to last very long!’” He laughs, apparently imagining a conveyor belt loaded with junk food. “What we do at Dunnhumby is the same, just on a larger scale.”
Using Clubcard data, Dunnhumby can tell that you have a new baby, or that your children have left home. It can judge your social class and knows whether you are a good cook. It also gives Tesco clues about what it could sell more of - and to whom.
Dunnhumby takes the information registered on Clubcards each time those 13 million families come into Tesco for their weekly shop, and turns it into five billion pieces of data. Each separate product bought has its own set of attributes. A ready meal can have up to 45 “values” ascribed to it: is it expensive, or cheap? Tesco-branded, or made by Birds Eye? An “ethnic” recipe, or a traditional British dish? Clubcard is the Big Brother of the shopping world.
This information is also stored in a vast search engine that can be used by suppliers trying to launch products. Dunnhumby makes about ₤30m a year selling Tesco data to more than 200 consumer-goods companies, such as Procter & Gamble, Unilever and Coca-Cola. Within hours of launching a product or introducing a promotion into a local Tesco store, brand managers can track who is buying their products or responding to their promotions. Are they empty-nesters or young mums, lawyers or factory workers? “If you understand who is buying and how they are buying, you can make better decisions,” Hayward says. “The joy of our sample is that it is so large, and because Tesco is so representative of the country it is the best source of insight a supplier can get.”
When Dunnhumby first linked up with Tesco in the autumn of 1994, it was a tiny, five-year-old company, turning over a mere ₤750,000 a year. Husband-and-wife team Clive Humby and Edwina Dunn had come up with the idea of starting the business in the mid-1980s, while working for market analysts CACI.
Using census data to better understand different areas and neighbourhoods in the UK, Humby helped companies make decisions on where to open a shop, or how they should tailor a door-to-door postal advertising drop to a particular area. “The basic idea then was ‘you are where you live’,” explains Humby, who holds the rather grand title of “chief information architect” of the Clubcard scheme, and takes great pride in his background as a trained mathematician and “data-miner” extraordinaire. “The argument was that if you live in a street full of flats, you are one sort of person, if you live in leafy suburbs, you are another sort of person,” he says, settling into the leather chair in his (much bigger) office down the hallway from Hayward.
Humby’s bright idea was that you could sketch more accurate pictures of people based not only on where they lived, but also on what they bought. He quit CACI with plans to go back to university. But then CACI fired Dunn.
“So we thought, ‘Oh well, let’s develop the idea for Dunnhumby.’ I thought, ‘Why don’t we work out how we can apply the same principles to customers instead of local areas?’ We would go to an organisation and say: ‘You have all this data locked up in your transactional system. If you can get that data, we’ll apply some magic to it and help you find out more things.’”
Dunnhumby signed up the likes of Cable & Wireless and BMW, acting as a “classic sort of marketing-services business”, finding clients and working out what they could do for them. While Humby and Dunn built their business, Tesco was simultaneously looking at ways of better understanding its shoppers, and was beginning to look at loyalty schemes. In 1993, Grant Harrison, a plucky young New Zealander, was asked to work on a trial programme to see if it was worth Tesco introducing a loyalty card. He spent months collecting shoppers’ data at tills in a handful of trial stores, then sending reward vouchers to customers’ homes.
It quickly became clear that loyalty cards were effective both at getting hold of customer data and encouraging shoppers back into stores. What Harrison needed now was someone to turn transactional data into information on individual customers’ needs and preferences.
“I went to this conference and there was Clive Humby talking about processes and saying how data-mining was not just a science, but an art,” Harrison remembers. “After the conference, Clive sent round a note saying it was nice to see me.” He starts laughing at this point. “It was silly because we hadn’t met. I sent him a reply to that effect and said, ‘That’s a poor form of direct marketing.’ Then Simon Hay [now chief executive of Dunnhumby in the US] called me, and we got on, and I brought them in.”
Humby, Hay and Harrison were given a simple brief by Tim Mason, Tesco’s marketing director at the time. “Convince the board that Clubcard adds value to the business,” he said. They had to win over Lord MacLaurin, Tesco’s chairman and the man who had scrapped Green Shield stamps, the old loyalty scheme, when he wrested power from founder Jack Cohen in the 1970s.
The trials ran throughout 1994. “We had lots of data about how people shop,” explains Humby. “You know, ‘Did you buy cheese? Did you buy milk?’ But we were finding things that really woke up lots of people. People assumed you shopped at one supermarket for everything. But we could show that people were not doing that; they were going to different places. Relatively few people don’t buy bread or meat. So if they are not buying it at Tesco, they must be going somewhere else. These holes were big opportunities.”
They tried to plug those holes with tempting titbits for customers. “We did things like offer someone who didn’t shop in a certain category ₤2 off if they tried it,” Humby recalls. The trials started in three stores, then expanded to six stores, then nine, then 14. “We showed that it could make a big difference. We could say, ‘Well, this investment is necessary because it gives you this pay-back;’ it was very tangible.”
Tesco refuses to say how much the rise in sales were worth in those early days (it is so proprietorial about Clubcard information that Tim Mason, who is now chief executive of Tesco USA, has insisted on approving every word in the second edition of Scoring Points, by Terry Hunt, Clive Humby and Tim Philips, Humby’s book on Clubcard). But Grant Harrison is more candid. “People were coming into the store more often, and spending more. Take Amersham [a prosperous town in Buckinghamshire]: the impact was substantial, a 12 per cent increase. And that store was already making ₤700,000-₤800,000 a week, so 12 per cent is a massive amount.”
By November, the retailer’s management team wanted to see the findings; the trio was summoned to a board meeting. Humby smiles ruefully when he is asked to recall that day. “I was so nervous, I’ve blocked out a lot of the details,” he says. “It was the first time we presented to the board. We told them what we had found. When we’d finished there was this deadly silence - I don’t know whether it lasted 30 seconds, but it felt like 30 minutes, as I recall! There was this hush and my heart dropped a little bit.”
The silence was broken by Lord MacLaurin, uttering his now legendary line: “What scares me about this is that you know more about my customers after three months than I know after 30 years.”
Humby and his colleagues stood in silence as chatter broke out across the room. For half an hour the directors discussed what had happened across those 14 stores during the autumn’s trials, and what it could mean for the rest of the business.
By the end of the meeting, the board had committed to Clubcard. For the next three months Humby, Hay, Harrison and other members of the tiny Clubcard team worked in secrecy on the launch. They were ensconced in an empty office near Tesco’s headquarters - dubbed the Bunker - and told to keep a low profile. The rollout was so secret that hardly any of Tesco’s staff knew about the plans, let alone its competitors. The official line was that Harrison was working on a home-shopping project.
“It didn’t surprise me, where it went,” Humby says of that day in the boardroom and the rush to launch. “I didn’t come out thinking I was going to be rich; that wasn’t my motivation. I like the numbers; I still like the numbers. Today I would much prefer to do some analysis of what is happening in Freefrom [Tesco’s wheat, dairy and gluten-free range] than worry about a corporate deal. I knew Clubcard could make us, and make us something special.”
But Humby and Dunn have become rich - very rich - earning millions of pounds by selling stakes in their business to Tesco. They sold 31 per cent to Tesco in May this year, taking the grocer’s stake from 53 to 84 per cent. Tesco will have paid tens of millions of pounds to increase its shareholding, given that Dunnhumby’s turnover is close to ₤90m a year and its pre-tax profits are expected to grow from ₤8.4m last year to more than ₤13m this time around.
Work for Tesco makes up between 20 and 30 per cent of that sum, while other suppliers are paying about ₤30m a year for Dunnhumby data and analysis. Humby reckons that is the tip of the iceberg: “I want to get to a $1bn business.” He doesn’t blink.
And what about Tesco? Tim Mason of Tesco USA agrees to meet me for an hour - he is based in Los Angeles, but is making a flying visit to the UK - to talk about Clubcard’s impact. Mason’s view is that it has driven Tesco’s current success.
“Clubcard has brought about a step-change in the size of the company,” he says. “We started off being able to understand the departments that people shopped in and the frequency at which they shopped, and we worked with those two variables.
“We are actually analysing what you do in a grocery store. The success of Dunnhumby’s approach lay in not just saying ‘you are Oxbridge-educated, and born in the north, therefore you behave in these ways’, but instead saying who you are is driven by what you buy - so we group people by products - and by how you behave. So we will look at people who shop in an Extra once a month, or a Tesco Express five times a week. It is an extremely good, thoughtful tool for analysing food shopping.”
At the simplest level, Tesco divides its customers into “convenience” shoppers (sub-divided into “time-poor, food-rich” and “can’t cook, won’t cook”) and “price-sensitive” ones (”stretching the budget” and “cheapest I can find”). There is also the 29 per cent who are more discerning and opt for “finer foods” (whether “natural chefs” or “cooking from scratch”). The “mainstream” customers (who buy lots of “kids’ stuff”, or “commonplace brands”) are the mid-market group. “Less affluent” shoppers make up 27 per cent of Tesco’s demographics; there are sub-categories - “traditional” and “price sensitive” - in that segment too.
By knowing these customers, Tesco can guess what they might like. Mason says that Clubcard data informed a series of strategic decisions, such as the move into smaller-store formats and the launch of the internet shopping site. It also shaped the development and sale of Tesco mobile phones, pet insurance and the Finest food range.
Humby gives another example of Clubcard’s many uses (he clearly loves it, too, as he grins from ear to ear while describing it). “One of the biggest successes in terms of analysis was watching how people shopped in the wine section. We could see that people were trading up to stuff Tesco didn’t stock. At Christmas, people wanted to buy ‘posh’ wine; those who usually bought cheap wine went from spending ₤2.99 a bottle to ₤5.99 a bottle - but where were the people who should have been trading up from ₤5.99 to ₤7.99? They were in Oddbins [the specialist wine retailer] - because Tesco didn’t have a full enough range.”
Dunnhumby can do all this because millions of people have agreed to have their shopping habits scrutinised in return for discounts on their groceries. One in three households in the UK has a Clubcard. Joining is easy. You go into one of Tesco’s 1,900 stores and pick up an application form. It asks for your name, address, telephone number and basic details about your household. It also asks if you are diabetic, teetotal or a vegetarian.
In return, Tesco gives you one point for every ₤1 you spend in its shops, petrol forecourts and on its internet site. You’ll get extra points for using a Tesco mobile phone, for signing up to Tesco broadband, or using a Tesco credit card. Build up enough points and you earn money off future purchases (Tesco handed out about ₤250m worth of discounts last year). Four times a year, Clubcard holders receive a letter with money-off vouchers, along with a range of other targeted offers.
All supermarkets have the basic information of what they sold at what time of day and for how much. But what they do not have is the extra layering that Clubcard - and Dunnhumby - applies.
Every time a checkout assistant swipes your Clubcard at a Tesco till, your shopping data drops into the Dunnhumby database. Every purchase is endowed with additional attributes. And added to that is longitudinal data: Dunnhumby has 40 to 50 customer variables that it weaves into purchasing patterns. Who is the “can’t cook, won’t cook” and how often are they buying ready meals? Does that person like new ranges or respond well to promotions? Each time you use your Clubcard another layer of information is added to your purchasing profile.
On top of that, Dunnhumby adds in more information from a software program called Zodiac. This trawls websites and other sources such as census data to build up socio-economic profiles of areas and customers. “Take healthy eating,” says Humby. “There is a high correlation with areas where there are more mums who don’t work. We know that if you live in an area where you are more likely to be a mum at home, you are more likely to cook different meals from someone who lives in an area where both parents tend to work.”
This layering process goes on in the silicon depths of Dunnhumby’s 40-terabyte database. The information is entered in four sections: items, promotions, attributes and customers. It is filtered and cleaned up, then pushed on into the preparation stage, sorting the data around customers and stores, baskets and products. Here is where the great electronic brain works on identifying patterns: who, say, visited during the week, how many times they visited, and what part of the store they bought from. Were they older people, or were they part of young families? Who went for promotions, and what did they like?
On one level, Dunnhumby is looking for large-scale information, watching how shopping habits change in order to anticipate what customer groups might want in the future - such as more organic lines, or more expensive wines. But at another level, it is putting together information on individuals in preparation for its quarterly Clubcard mailings with money-off vouchers and offers. It is a huge operation: the mailings account for more than 6 per cent of the UK’s annual post. Mason says the days immediately after the mailings bring in huge numbers of customers, and big uplifts in sales. “They are big weeks,” he says.
Inevitably, Dunnhumby’s piecing together of information means it has at its fingertips the ability to paint a very full picture of an individual Clubcard user - more than someone might expect when they sign up. But this is the sort of comment Humby hates. He is keen to stress that Dunnhumby’s mapping is about “collections of people” rather than individuals - although all Clubcard holders are individually identified in the database by their name as well as their card number. “We analyse in aggregate. We don’t build profiles by saying what Martin Hayward does as an individual. But we do know what items he purchased, and our job is to make patterns from that information.”
Still, if it wanted, Dunnhumby has the power to drill into everything you have bought in the past decade. Add the other information that the company ties in - from the electoral roll, the Office for National Statistics, the Land Registry and so on - and it probably has more frequently updated personal information about named individuals in the UK than any other organisation.
Humby reiterates that identities are protected, and that no third-party users have access to personal names and addresses. Tim Mason agrees: “It concerns me when people write about Clubcard being intrusive and frightening, because you know the Big Brother angle is a compelling story. The important point is that we look at people’s grocery-shopping behaviour rather than their behaviour as individuals, and we try to run better stores and give them better offers as a result of that understanding. I think we do that and I think that is why we are successful.”
And Tesco is fiercely protective of Clubcard information. During my research, I got my father to ask Tesco for his Clubcard record. A cardholder for 12 years, he is the perfect example of a long-term customer who has changed his shopping habits, having gone from having two of his three children still at home to being in an “empty nest” household of only himself and my mother.
Two letters, a photocopy of his passport, a letter showing proof of address and a ₤10 administrative charge later, Tesco dispatched my father’s data. At 16 pages long, it is bare on detail. It lists all his store visits since May 2006, and the amount he has spent. It also details the points he has earned since August 2004. It states, correctly, that he is over 60 - but claims that he has two children aged between 10 and 19 living in his household. It says he is in the “FO” life stage. I ask Tesco what that means. But it won’t comment, and asks my father to call instead. He duly does so, and we discover that he is in the Family Older life bracket. Tesco has not updated his information since he joined the scheme in 1994. I call Hayward. He is rather gleeful. “You see - it is not about personal information as such. That information is captured when you apply. It is just a start point; we don’t want to hassle people for new information. What is crucial is that we send you a mailing with offers that relate to what you have bought.”
While it doesn’t pass on personal information, Dunnhumby’s big growth area is selling on information to fast-moving consumer-goods companies. Once Dunnhumby has pro-cessed Clubcard data, it publishes a 10 per cent sample in a system called Shop and charges suppliers up to ₤50,000 a year for access to it. This may seem a lot - but Shop is, after all, the largest customer sample of its kind in the UK, taking in a million shoppers from all social strata.
Shop allows suppliers to search products and promotions to see how they are performing in different stores. They can also see how launches of new ranges or product lines are working out, and examine local and regional trends by selecting store clusters within the database. Humby cites the main advantages as “the volume of data and how quickly trends can be spotted”.
But it is also crucial to suppliers’ relationship with Tesco. One supplier, who wished to remain anonymous, said it is extremely hard to pitch to Tesco buyers these days without using Dunnhumby data - since this is the information that the buyers are used to basing their decisions on. “They like suppliers to talk to them in a consistent manner that they understand. Is it an industry standard? Not yet, but Tesco would like it to be.”
Tesco is a crucial client and suppliers are nervous of upsetting their biggest buyer. Procter & Gamble does not want to talk about Clubcard. Neither does Unilever. Muller, the yogurt maker, says it will, but its managing director pulls out of the call just an hour before it is to take place, despite being sent a full list of questions ahead of the appointment.
One supplier that is willing to talk is Dairy Crest. Its chief executive, Drummond Hall, agrees to discuss how Dunnhumby data helps his business, and we meet in his office near Esher, in the Surrey stockbroker belt. Hall was an early adopter of Shop. “It has become absolutely integral,” he says. “We are sharing common, real-time data with Tesco. We use it to make sure we have the right ranges in store and the right shelf space, and we use it in discussion with Tesco themselves.
“The final piece of the jigsaw, and an important one, is promotional activities, such as couponing at the till or through the quarterly mailings. You can target customers. Without getting too detailed, you can do a cross-correlation of who is most likely to be using your brand in Tesco.
“Take Clover [dairy spread], a slightly upscale product for a slightly older customer. Dunnhumby can give us an analysis of which people are using, say, Anchor Spreadable; they are more likely to use Clover, so I can try to get them to change brands by sending them a coupon.”
Hall says the data can also help with product launches. Keen to drill home that Dairy Crest is more than a milk producer, he cites Cathedral City cheddar and his launch of a branded mild cheese.
“The mild sector is about 20 per cent of the cheddar market, and 30 to 35 per cent of the volume. Brand penetration is under eight per cent. We take the view that there is probably a market there. So are people buying mild because they don’t like brands and want a cheaper product, or are they buying it for taste reasons, for younger children, say? The Dunnhumby data show us that mild-cheese shoppers are every bit as brand-prone as mature-cheese shoppers. It also shows that mild shoppers are not buying it just because it is cheaper, they are buying it all the time; they buy only a limited amount of mature cheese. This is a profile that says to us we have a real opportunity.”
A couple of months later, following the launch of Cathedral City Mild, I call him to see how it went. He says a disproportionate number of families are choosing the brand. “If you take young families on an index of 100, Cathedral City Mild is hitting 140. So a lot more people than you would expect in this group are buying it.”
But do those housewives (or husbands) care that their every cheese-purchasing decision is being tracked? “I don’t think customers are particularly aware that they are being analysed,” says Hall, who has worked in the consumer-goods industry all his career. “But what we are talking about here is only a refinement of what has been going on ever since Mr Procter and Mr Gamble started marketing. It is simply a refinement of what we were always trying to do. Get to market quicker, improve our success. We will only be able to do that if we meet consumer requirements. So why should consumers be worried about it?”
