When @WalmartLabs (NYSE:WMT) on Monday (June 10) announced that it had purchased Inkiru—a realtime predictive intelligence firm—it said something unusual. Walmart's statement certainly didn't want to leave any buzzword left out and it even managed to work in today's buzziest of buzzwords—Big Data—five times in the 306-word statement. But the unusual part of the statement was this sentence: "The similarities between Inkiru and @WalmartLabs are uncanny, with both having an innovative spirit and the ability to leverage big data to improve the customer experience."
Typically, acquisitions stress how the two companies are polar opposites, how their strengths and weaknesses complement each other, making the synergy of the merger a beautiful thing to behold. Stressing similarities raises the question of why the acquirer needed the acquire. And it's the second reason Walmart cited—please forgive me for not addressing their mutual "innovative spirit" on a full stomach—that is intriguing. Namely, that both have the ability to leverage big data to improve customer experiences.
Indeed, 'tis true. Walmart has been very effective at rapid analysis of its in-store shoppers and their activity, especially as it folds mobile into the environment. But online, though, has a very different experience. (If Walmart.com would stop assuming my local Walmart is 800 miles away, and if it would stop forgetting MyStore even though I have personally set it more than nine times since New Year's Eve, I'd likely feel differently.) Walmart's struggles are certainly nothing unusual, as just about all of major chains have discovered that even though online delivers far more information about every shopper action, the ability to use that information and to make correct predictions from it is a lot harder than it looks.
Enter Inkiru, which has developed quite a reputation in predictive analytics. Then again, Kosmix had just as strong a reputation in social media when Walmart purchased it. If you ever want to figure out the upper-limit of scalability—to determine when a perfectly reasonable model will blow up when deployed—the world's largest retailer is the perfect laboratory. In theory, scale is where analytics become the most impressive. A model that might work when applied to 500 shoppers will likely work more accurately when applied to 50,000 or 500,000 shoppers. But there comes a ceiling where the numbers can become too big, too unwieldy. Is that the Walmart curse?
Walmart describes Inkiru's approach as "an active learning system that combines real-time predictive intelligence, big data analytics and a customizable decision engine to inform and streamline business decisions." That all sounds great, and it's the perfect direction for Walmart. But how far will Walmart take this? Will it finally use mapping data to determine when a shopper is likely visiting Target or Macy's and then promote items appropriately?
Walmart already described an ideal mobile app, one that would autopopulate a shopping list based on shopping history and reasonable projections. That was just a goal, though. The question is: Will it allow Inkiru's team to turn it into reality?