Zara – Using ‘Big Data’ to create business value

“It’s a major mistake to theorize before having data.” Sherlock Holmes (Sir Arthur Conan Doyle)

… especially as the advent of the so-called “Big Data” makes the question of data scarcity of the past. Capturing data and their transformation into business insights as a core element of strategy has long helped Spanish retailer Zara increase productivity, improve decision-making and gain competitive advantage. As a result, it overtook Gap as the world’s largest clothing retailer in 2008.

Zara has been a poster child for supply chain expertise because of its ability to spot trends as they emerge and to deliver new items to stores quickly to meet the needs of fashion-conscious customers. In an industry where standard lead time – designing, producing and delivering new garments – is about nine months, Zara is leading the way with as little as two to three weeks. However, the driver behind this efficient supply chain is the use of data and analysis for accurate forecasting and decision making. It is enabled through processes and systems built to gather data, analytics, front-line tools and people to create business value. Zara’s main differentiating uses of analysis are to: –

– institutionalize the collection and use of real-time statistical market data. Zara’s cross-functional design team pores daily sales and inventory reports to see what sells and what doesn’t, and continuously updates their views on the market. Bi-weekly orders from store managers provide additional real-time information on what might be sold;

– supplement the statistical market data with the finished grain market data. Authorized retailers regularly send word-of-mouth feedback on customer requests and preferences – anything from “the length of this skirt is too long” to “our customers don’t like the fabric of this dress”. Managers can also suggest changes to an existing style or suggest brand new articles or designs. The advantage of store insights is marked by the example of a number of slim-fit clothing that did not sell. The feedback from the stores was that women loved what the thin fit clothes looked like, but could not fit into their usual sizes when trying on the clothes. Zara remembered the items and replaced the labels with the next sizes up, and sales exploded;

– create an adaptive and informal planning process. It integrates into the company’s flexible supply chain as it maintains strong ties with its 1,400 external suppliers, working closely with its designers and marketers. Based on market data, Zara is experimenting with a wide range of offers in small batches. If they prove to be a hit, production increases in response to local conditions while maintaining lean inventories and a low level of markdowns;

– disseminate information widely throughout the organization. Designers, pattern makers, marketing executives and merchandisers as well as anyone else involved in the production are on a single open office floor. This allows for frequent discussions, serendipitous meetings and visual inspection. The entire team can diagnose the overall market, see how their work fits into the big picture, and spot opportunities that might otherwise fall between the cracks in organizational silos;

– Build a simple and effective information technology system that is accessible to everyone. Zara’s internal IT reflects the way the organization works. It is silo-free as well as available to suppliers and suppliers who report that it is easy to use and quick to respond; and

– Build a culture of data usage to learn new things and discover the right answers. Data analysis is the basis of Zara’s model and its use for decision making is encouraged as bad decisions are not severely punished. The error rate for Zara’s new products is reported to be only 1% against an industry average of 10%.

A few years ago, Zara entered the virtual ground for e-commerce in the US, Europe and Japan. With this move, it entered the next generation of real-time decision-making and marketing analytics: tracking the behavior of individual customers from Internet click-streams, updating their preferences, and modeling their likely behavior in real-time in addition to monitoring social networking conversations and location-specific smartphone interactions.