Listen to this article
Consider one consumer filling a supermarket trolley with food, and another, sitting on a sofa, browsing electronics via an iPad price comparison app. They are the two extremes of contemporary shopping – and in more ways than one.
Beyond the bricks-versus-clicks divide, they also embody the contrary forces that scientists are harnessing by mining digital information to shape pricing decisions. These moves to unlock the power of big data are increasingly redrawing the relationship between retailers and consumers.
Kroger, a US supermarket, uses sophisticated tools to give shoppers personalised discounts, while Decide.com, an online data repository, forecasts how a range of retailers are likely to change the prices of certain products in the future.
These two examples of big data in action– made possible by computer programmes called algorithms that discern patterns and predict behaviour from the numbers – have implications that run deeper than the chance for consumers to save, or waste, dollars.
Kroger’s underlying goal is to make shoppers more loyal. Decide is encouraging them to flit around in search of whoever has the lowest price.
The future of retail will depend partly on which kind of initiative wins out: does big data make shoppers faithful or fickle?
“Within retail, pricing is one of the areas furthest along in the application of big data,” says Kurt Kendall of Kurt Salmon, a retail consultancy.
Decide talks about transparency empowering consumers in an era of minute-by-minute price changes, which are driven by algorithms that were pioneered by stock market traders and first deployed in ecommerce by Amazon.
The result is prices that increasingly behave like real-time expressions of supply and demand – and of retailers’ ploys to outwit each other, and sometimes shoppers.
The volume and velocity of data can be overwhelming and Mike Fridgen, Decide’s chief executive, says: “We’re the people working on behalf of consumers. We’re arming consumers with the tools to fight back. The data scientists are on their side.”
Via a smartphone app that can be used in a mall, to the chagrin of some traditional retailers, Decide can advise consumers to wait to buy a certain fridge, for example, because it expects the price online to drop by $500 in the next two weeks.
Its forecasts cover appliances, electronics and housewares and it attaches a level of certainty to each one – the average is 77 per cent.
More than simply extrapolating from the past, its algorithms make predictions by analysing a host of factors, including the evolution of new product models, the relationship between different types of goods, the erratic decision-making of human beings and the jousting of rival stores.
Over the Black Friday shopping period in the last 10 days of November, Decide watched retailers including Amazon, Best Buy, Walmart and Target slash prices to match each other’s “door buster” discounts then push them back up bumpily days later (see graphic).
“Any [retailer] who’s counted on the opacity of information is going to be the first disrupted,” says Michael Paulson, Decide’s vice-president of product and marketing.
For a $30 annual fee Decide’s service includes product recommendations and a price guarantee to back up its predictions, but Robert Wollan of Accenture, a consultancy, says that for now it is probably only for “extreme price shoppers”.
“The question is will that become more mainstream,” he adds. Accenture surveys show that while consumers are “switching” more often between different retailers, they are driven more by the quality of service – good or bad – than price.
Kroger’s bet is that customers don’t want to be fickle flitters. They want trusted retailers that know them well enough to give them deals they like without the hassle.
That is where the big data comes in. Kroger began mining it after it set up a loyalty card programme in 2003 with Dunnhumby, a London-based data analytics group that helped create Tesco’s pioneering Clubcard and also works with Macy’s.
The US’s second-biggest grocer with $90bn in annual sales, Kroger saw in the data that focusing discounts on the top-selling brands – where it was often willing to lose money to lure in new shoppers – was not the most effective strategy.
“Loss leaders were disproportionately going to customers that were not your best customers,” says Ted Sarosy, Kroger’s vice-president of loyalty.
So it cut back on across-the-board discounts. Instead it sends its most price-sensitive customers about 18 sets of personalised discounts a year, arriving via both old-fashioned mail and its loyalty programme’s smartphone app.
They are crafted by analysing a raft of data: transaction histories at Kroger; other bricks-and-mortar spending, which is tracked by researchers such as Nielsen; browsing and purchasing patterns online; and even television viewing habits.
In cereals, for example, that means fewer discounts on the Kellogg’s Cornflakes popular with higher-income shoppers, and more on cheaper own-brands favoured by lower-income families, such as Kroger Marshmallow Treasures.
Kroger says its discounts add up to $2bn of savings a year. Even the supermarket’s best customers spend only half of all their retail dollars at the store, says Mr Sarosy, and his goal is to increase that figure rather than woo new shoppers.
“We have more chance to get share of wallet from them than from others visiting infrequently,” he says.
Kroger is running a risk: its cultivation of loyalty could hit a hitch if friends shopping together are annoyed to find that they pay different prices for the same product.
No matter how big the data, its power to nurture faithful or fickle consumers, with trolleys and iPads alike, will depend on little moments of discovery like that.
Get alerts on Big Data when a new story is published