3 Ways Conventional BI Analytics Will Fail Retailers this Holiday Season

Print Friendly, PDF & Email

ArijitSenguptaPhotoIn this special guest feature, Arijit Sengupta, CEO of BeyondCore, discusses how retailers can use analytics to meet their customers’ needs and achieve greater results this holiday season. Before founding BeyondCore, Arijit held a variety of technical and management positions at Oracle and Microsoft. He has been granted a dozen patents in advanced analytics, business process as a service, operational risk, privacy and information security. Arijit has guest lectured at Stanford; spoken at conferences in a dozen countries; and was written about in The World Is Flat 3.0, New York Times, San Jose Mercury News, Harvard Business Review and The Economist. Arijit holds an MBA with distinction from the Harvard Business School and Bachelor degrees with distinction in Computer Science and Economics from Stanford University.

By most standards, for retailers who aren’t ready for the holiday season at this point, it is far too late. The “hot item” predictions, distribution plans and even the in-store displays have already been executed. By the time consumers start shopping in earnest, the products, displays and deals they see will have been in the works for over five months.

That would be fine if consumer taste, behavior and trends only changed every five months. In reality, we know that what’s hot can change in a matter of days. Trends don’t start with a dashboard—they start with people. And all it takes is a single tweet or posting by an influential style watcher or trend-setter and the entire season’s predictions, plans and promotions fly straight out the window. For retailers, that means months of wasted time, money and energy in planning and merchandising piles of overstock inventory and lost sales opportunity.

The fact is, regardless of the painstaking planning, trend-spotting and sophisticated analysis conducted in preparation for the holiday season, conventional BI analytics models fail retailers at this most critical time for three distinct reasons:

  • Lack of specificity. Retailers need to know which products will sell best in what locations, but most predictive models are very limited. They can consider only a few hundred factors, which might give you a clue about trends in various regions—such as clothing styles in the Northeast or which video game is all the rage in the Deep South. But, what if you want to be more specific? What if you could know which colors of cashmere cowl neck sweaters are most popular with women ages 24-36 in suburban Boston? Or, which M-rated first-person shooter game for Xbox will all the 18-24 year-olds near the University of Alabama want to see in their stocking this year? Conventional BI analytics can’t give you those answers to help optimize distribution and merchandizing. You need real-time analytics that can examine millions of questions, build hundreds of models and help you understand the subtle differences.
  • It’s based on stale data. Conventional prediction models take too long to manually craft and update. If your predictive modeling takes a month to complete and mine takes a day, I could decide based on the latest data possible, while you are deciding based on months-old data. And, since we know that consumer behavior is fickle, every extra day of data matters. When a model can be created, updated and acted upon faster—in days or hours, not weeks or months, you end up acting based on more current data.
  • They’re static. With so much riding on such a short window of sales, retailers need to be agile and responsive, even once the season begins. No matter how smart and accurate your planning might be, things can change quickly, and conventional BI systems are not equipped for the real-time data monitoring that enables mid-stream adjustments based on emerging factors. With smart pattern discovery, retailers could watch what’s happening with early morning Black Friday online sales on the East Coast and instantly start tuning West Coast promotions and deals to capitalize on opportunities later that same day. Is that hot item not selling quite like you thought it would in Dallas, Texas? Maybe a surprise lake effect snowstorm has shut everything down in Erie, Pa? With real-time analytics, you could see the impact of these factors on sales on an hourly, or even minute, basis to adjust promotions or shift your focus to online sales accordingly. With real-time analytics, you would know the specifics of what’s happening at the individual store level and be able to make adjustments, rather than examining what went wrong after Refund & Returns season has ended.

Despite their best efforts and most sophisticated systems, even the biggest retailers in the world will get it wrong this holiday season without real-time insight into how, when, why and where consumer buying behavior is changing. With only six short weeks to make or break the entire year’s revenue goals, there is still time to ramp up a real-time strategy before consumers start lining up and opening their wallets.


Download the insideBIGDATA Guide to Retail

Speak Your Mind