Thursday, September 09, 2004

Automatic customer recognition, with fries

"Good afternoon, Mr. Yakamoto. How did you like that three-pack of tank tops you bought last time you were in?" So says a smiling holographic Gap employee on a giant screen, greeting Tom Cruise as he enters the store. The movie Minority Report is set in 2054, but personalized marketing at store level is just around the corner. And fast-food chains are leading the way.

The starting point is supply chain management on a micro-scale. If you want to maximize profits in fast food, you need to anticipate orders before they happen. You can do that with modeling -- massaging all of your historical sales data to produce a predictive model of what is going to be ordered at what minute of the day. You can fine-tune that model by watching in real-time the volume and nature of traffic approaching your store. According to this report in InformationWeek, some fast-food outlets have implemented a system that gives food-preparers a few minutes advance notice of a likely order. It uses cameras to recognize the type of vehicle entering the drive-through, infers from past data what type of order is about to be placed, and notifies workers to get busy preparing it. A mini-van is likely to order more burgers than a sedan -- it's that simple.

Now that's a clunky and imprecise system compared with the eye-scanning person-recognition system of Minority Report. But it apparently has dramatically cut average wait times, has reduced waste significantly, and has cut average training time for new hires from three months to just over a week!

Imagine the impact on the bottom line of a system that recognizes individual customers and knows their ordering habits. The existing system could be easily updated to read license plates instead of just recognizing generic vehicle types. But that's low-tech. How about a system similar to toll-road smart-tags that each vehicle or each walk-in customer would carry, transmitting at a distance a unique identifier that pulls from the database of past transactions today's likely order? Perhaps a frequent customer card like those issued by Starbucks, but with a smart RFID tag embedded?

This is not far-fetched Spielberg sci-fi. It happens already in the virtual world of e-commerce. There are not many people who do not have a computer overloaded with cookies from every site they have visited. Those cookies serve as personal identifiers, and make your visit to or any other site more personalized. Privacy advocates warn of potential abuses of the cookie system, and there are many, but consumers don't seem to care. Surveys have shown that we are willing to risk privacy for the convenience of personalization online.

So how much resistance will there be to the concept in the real world of brick-and-mortar retailing? Would I really object if, on my next visit to Blockbuster, I was greeted by a virtual host who not only recommended a movie that I would probably enjoy, based on my history of past rentals, but ran a trailer for me? Oh, I forgot -- Blockbuster has gone online, too, and that service is already virtually available...

No comments: