Dixons Carphone uses predictive technology to determine marketing strategy


One wouldn’t expect much innovation from a company with “car phone” in its name.

Dixons Carphone, however, is forging ahead with some of the biggest advances in retail business analytics. The name of the London-based company is a vestige of Carphone Warehouse, a major mobile phone retailer in the United Kingdom since the first cell towers went up. In 2014, large consumer electronics chain Dixons Retail merged with Carphone Warehouse and a new company with more than 1,000 European locations
was formed.

“You won’t find a ‘car phone’ in our stores but we keep the name because of its brand recognition,” says Anthony Morris, director of strategy and analytics for Dixons. “People know that ‘car phone’ is synonymous for ‘mobile,’ but there’s also ‘Dixons,’ which people in the U.K. know for being a leader in electronics sales.”

Another piece of the conglomerate is Currys, a chain purchased in the 1990s that sells home appliances and electronics throughout the country. Along with the company’s acquisition of PC World, the result is a firm that Morris says is kind of a combination Apple Store and Best Buy with a focus on creatively engaging its customers.
“We place a heavy emphasis on service in our stores,” Morris says, “and we’ve seen customers respond to our approach at helping them navigate a sometimes confusing consumer electronics world.”

“You need to know if what you’re doing is enriching customer insight and helping you understand what’s driving sales.”
— Anthony Morris, Dixons Carphone

Morris has been actively working on improving the company’s business analytics since his arrival in 2009. Coordinating analytical goals with business strategy is critical — and more importantly, the days of scanning spreadsheets for predictive patterns have gone the way of the typewriter and the abacus.

“You need to know if what you’re doing is enriching customer insight and helping you understand what’s driving sales,” he says. “That’s especially true in electronics, since consumers often put a lot of thought into those buying decisions.”

Removing variables

Morris’s department focuses on customer analytics, figuring out what kind of product mix or offers work for a market segment. It’s a critical function for a business trying to integrate various retail acquisitions. Morris used econometric models to simulate how particular items would do in a location, trying to see if a particular trial offer was effective for certain consumers.

The difficulty in getting accurate analytical data in retail is that there are so many variables. A product test at a handful of stores may be affected by rain at one and road construction at another. One store may be more crowded than another because of an unrelated promotion, which skews the sales figures.

For data professionals like Morris, the secret to successfully determine if a marketing promotion is working is weeding out the “noise” of unrelated variables to figure out the impact of a promotion.

Using his own algorithms and inputs on transaction data to calculate expected sales, the results Morris received were accurate; however, “they were long-winded and took time to develop,” he says. “We saw we needed something faster that could be acted upon very quickly.”

The company had done some work with Applied Predictive Technologies in the past and contacted APT to help streamline its analytics operation. “We were measuring lots of different product trials, and the challenge was how to get a clear response to whether a particular trial was working and why,” Morris says.

Using APT’s software, the company could remove unrelated factors affecting a trial the way one peels away an onion. “We could input the data and quickly see what aspects of a trial were working and which were not,” he says.

A unified system

A challenge for many businesses is that different departments will often have their own analytical experts calculating the same data. “It’s not uncommon to have executive teams working on similar issues and getting different answers based on their analysis,” says Rupert Naylor, senior vice president of European operations for APT.

“This can lead to confusion in the company as to what to do. It’s much easier if there’s a unified, reliable system you can depend on.”

It can also be affected by departments or executives that find it hard to trust an analytical view of a possible trial, preferring to base decisions on a “gut” feeling.

“You can’t put a value on experience, and many professionals will have a good sense of how a promotion will go,” Naylor says. “On the other hand, it would be better to wed that experience with hard data to show there’s a likelihood of success or failure.”

Naylor surveys retailers to ask how many of the marketing trials they’ve tested over the past year broke even. “I’m not asking if they didn’t meet ROI targets, I’m asking how many at least maintained corporate value,” he says.

“Among retailers, we find that about 44 percent of tested trials don’t break even, which means that going by your ‘gut’ to make these kinds of decisions is a good way to reduce your company’s value.”

Increased adoption

For Dixons, one of the big first tests of the APT system was to see how it would work with proposed product delivery times. Fast delivery is critical to any retailer, and free or extremely low-cost delivery is often expected by consumers. Information about delivery rates, types and slots was fed into the software, along with data about customer preferences.

“What it did was show us how to prioritize capacity at busy times,” Morris says, “and ensure that the greatest number of customers were getting the best value they could receive at the lowest delivery cost for us.”

With the integration of Carphone Warehouse and Currys stores into the Dixons operation, the company had to deal with combining various types of store footprints as well as employee training and business goals. “We could see information about what types of stores did best in certain areas and look at the impact of employee training to see how best to create a uniform, Dixons experience,” Morris says.

As use of APT began to become more widespread in company’s headquarters, Morris saw colleagues in other departments becoming increasingly interested in analytics. “In meetings people began using the term, ‘Let’s APT that information,’ using it as a verb,” he says. “Then I knew we were on the right track.”

Looking ahead, Dixons is checking into how to use the tool for different segments of customers and finding ways to differentiate its consumer base even more. “The more we can know about our customers and what motivates them, the better we can serve them and affect our business,” Morris says.

John Morell is a Los Angeles-based writer who has covered retail and business topics for a number of publications around the world.


Comments are closed.