What could you do with 21 billion lines of customer purchase data?
If your store participates in the Nectar loyalty rewards program in the U.K., you could analyze promotions and product launches, optimize the layout of your store, discover availability problems and micro-target promotions to your cardholders -– about 12 million households, or half the population of the U.K.
Those 21 billion lines of data represent every transaction made on a Nectar loyalty card in the last two years. Overlaying point-of-purchase data with information from the cardholder allows powerful analysis of purchasing patterns, according to Michael Poyser, solutions director for LMG Insight & Communication, which runs the Nectar program.
“It’s not a data mining tool where you’re just pulling out numbers,” Poyser said. LMG’s SelfServe retail analytics tool is structured to answer business questions by running queries against that enormous transaction database, and then reporting (and usually within two minutes).
The queries are run in LMG’s London-based data warehouse, which is run by Kognitio. Queries are sent and reports are returned online. Data is gathered daily and loaded into the data warehouse weekly. It could be done almost in real-time, he said, “but that’s almost two much information … when you’re changing an assortment, it doesn’t have to be real-time.”
LMG was bought by the Montreal-based Groupe Aeroplan Ltd. in 2007, and the company launched the service for Aeroplan member businesses and their suppliers in Canada in late November. In the U.K., LMG launched its business with Sainsbury’s Supermarkets Ltd., the No. 3 grocery retailer in the U.K., and more than 50 consumer packaged goods companies that supply it.
It’s a much more powerful tool than simply parsing point-of-sale data, Poyser said. Being able to track consumer behaviour over time offers the opportunity, for example, to do promotional analysis.
Take a new product promotion: point-of-sale data can only tell you about purchasing spike, Poyser said. Analyzing against the loyalty card data can tell a retailer or a supplier if the customer was new to the category, what product they switched from, and whether the customer continued to buy the new product or switched back.
They also have a way of talking to customers, by providing direct mail offers that are relevant to those customers’ buying patterns, Poyser said.
Using such volumes of data for analysis isn’t a gold mine for retailers and their suppliers, said Sebastien Ruest, vice-president of services research for IDC Canada Ltd. “It’s a platinum mine,” he said.
In the past, that volume of data wasn’t cross-tabulated. Processing power has made this more realistic. Ruest said when he did data mining for Quebec’s medicare program a decade ago, queries against 10 million records would take two days to process.
He said, “Now, you have the ability to follow the patterns of the individual and make micro-targeted offers. That’s really the value of the information.”
It also allows more out-of-the-box segmentation, like what was bought by “left-handed individuals with brown hair,” he joked. Point-of-sale data alone can’t connect cause and effect.
“I think that’s what business intelligence aspires to, the ability to correlate all data points,” Ruest said.
