Mining for riches among the debris of falling sales

Big data is a rich seam of information which can help retailers in difficult times, as Stephen Pritchard discovers


High-street retailers are struggling with falling sales and rising costs. But there is one area where the sector is suffering from an embarrassment of riches: customer data.

Each time someone scans an item at a till, swipes a loyalty card or searches an online catalogue, retailers can capture data, and start to build up a detailed picture of how well a product is selling, how profitable a store is, or even the spending patterns and preferences of an individual shopper.

This is helping the best retail companies to carve out new niches, develop new products and make healthy profits. For high-street stores especially, the need to compete with online rivals – often with lower prices – means emphasising convenience, quality service and a more personal shopping experience.

But retailers have to balance how they use customer information with privacy and also profitability. Some retailers have gathered data for many years, yet failed to use it to change the way they run their stores.

“Retailers have had ‘big data’ for a long time. The industry with the most data is retail,” says Chris Donnelly, global managing director of retail at consultants Accenture. “It almost looks like they have too much data. They have been collecting data because they know it’s important, but they don’t know what to do with it.”

In Japan there are grocery chains that change all their shelves three times a day, to capture trade from morning commuters, lunchtime shoppers and the after-work, top-up shop

Retailers, he says, have been storing more and more information, often collected through loyalty cards or sophisticated tills. But they have tended to run their buying operations and shops on conventional lines, buying goods for the season and marking down excess stock at the end, in the sales.

So-called “pure-play” online retailers are able to change prices almost instantly. This is based on sophisticated computer models and information they capture, not just from what their customers buy but, as importantly, what they browse, but do not buy. And consumers are becoming savvier, using websites and smartphones to compare prices and offers.

To fight back, the high street is turning to its information stores and “data mining” to find out what small groups, or even individual shoppers, want. If a customer holds a loyalty card or, increasingly, a smartphone app, it is easy for retailers to target them with enticing offers, even before they walk into a store. But to do so effectively needs a deeper understanding of what drives someone to shop in the first place.

Store groups are using their data to understand “shopping missions” to improve their targeting, says Duncan Ross, chief data scientist for the UK arm of Teradata.

A customer on a big weekly shop behaves differently to one doing a “top-up shop” during the week, he says. Knowing that helps the retailer make better offers – perhaps snacks rather than tinned goods – but also to make the shop environment more convenient or attractive.

Some retailers are even going as far as changing the layout of their stores or the lighting and music to suit different times of the day. In Japan, for example, there are grocery chains that change all their shelves three times a day, to capture trade from morning commuters, lunchtime shoppers and the after-work, top-up shop.

“The nirvana for the retailer is if a customer from Leeds or London goes to the seaside, they know whether they have bought sun cream,” says Eddie Short, a data analytics expert at consultants KPMG. Often this information comes from loyalty cards. “If on day two of their holiday they have bought their sun cream, they won’t offer it again.”

But retailers are turning to new sources of information to fine tune the customer experience, including information from social media sites, such as Facebook and Twitter, and data from shoppers’ mobile phones.

“As a consumer, you are surrounded by a context, which changes depending on what you are doing,” says Dave Coplin, chief envisaging officer for Microsoft UK. “If the retailer knew your mood or who you are with, they might think differently about the services they deliver to you. Where you are and the time of day, tells something about what you are doing. If you are searching for sushi on a weekday lunchtime, you might be looking for somewhere to eat. If it’s at home at the weekend, you might be looking for a recipe book.”

Retailers, though, have to use data in a way that benefits shoppers, without seeming creepy. And, despite years or, in some cases, decades of gathering data, there is still a lot that even the best retailers do not know.

“Data on what customers don’t spend with you may be as important as what they do,” says Christine Cross, chief retail adviser at PwC. “Understanding the drivers of customers’ different shopping patterns will allow you to structure services differently.” This, though, means tapping into new, and less certain, sources of data, especially social media.

“What we’re seeing is that the data sets are changing: they are getting bigger, with more channels and more interaction,” says Jay Henderson, a strategy director in the enterprise marketing management group at IBM.

Retailers can, for example, use information from Twitter to find out why customers did not shop with them and information from their online discussion forums to find out whether those who did enjoyed the experience.

But this, in turn, means stores need people with different skills; they have to have a good retail instinct, as well as analytical skills. “They have to be able to blend the technology and the science with the art,” he says.

PRIVACY

DON’T GET TOO PERSONAL

For retailers, the biggest challenge, when it comes to personalisation, is finding the right balance between effectiveness and consumer privacy.

Sometimes, data capture can have unintended consequences. A consumer might be comfortable with a café chain knowing she prefers cappuccino to latte, but less happy for a supermarket to deduce that she is pregnant.

“There is so much more data, so the challenge is to identify what is relevant. There’s almost a danger of collecting data for the sake of it,” warns Peter Bull, a retail expert at PA Consulting.

And, says Microsoft’s David Coplin, retailers have not always balanced a customer’s privacy with the benefits they gain from sharing data. This is especially important if retailers are to move to more individual, but also potentially more intrusive, forms of personalisation, such as downloading offers to a shopper’s smartphone as they walk into the store. “People don’t understand what happens to their data,” he says. “Customers need to know what they will get in return.”

CASE STUDY

EVERY LITTLE PIECE OF DATA HELPS

One of the greatest irritations for shoppers is visiting a store and finding that the product they want is out of stock. But demand for some items varies widely and often that demand is driven by factors outside the retailer’s control, such as the weather.

At Tesco, buyers work with data scientists to predict customer demand and to work out the best way to balance having fully stocked shelves with minimising costly waste. Take the upswing in demand for barbecue meat if there is a sunny weekend: retailers that do not plan ahead, can quickly run out of stock.

“Customer demand varies with weather,” explains Duncan Apthorp, Tesco’s supply chain systems development programme manager.

“We have four years of data and have stripped out all effects other than weather. If we have a 10 degree rise in temperature you see a 25 per cent rise in demand for barbecue food during the week, but a 42 per cent rise at the weekend. A hot sunny weekend can increase barbecue meat sales by 250 per cent. To meet this demand we have to work with our suppliers over a week in advance to give them the chance to meet the increased volumes.”

This, he says, is part of an ongoing process of moving away from the “educated guess” approach to demand planning, to a more scientific method, driven by data. And the supermarket chain also takes a rigorous approach to testing whether the data is supporting the right decisions. “New projects are introduced by testing in one store, then 200,” says Mr Apthorp. “This gives us confidence that it is the right thing to do, before introducing [it] into all our stores.”

Mr Apthorp expects more decisions to be based on data as using the evidence from the company’s vast databases is the best way to ensure customers can choose the products they want and the supermarket can supply them at a profit.

“We plan to give the customer the best customer experience and improving our supply chain efficiency is a key part of that,” he says.