Mary Hayes Weier of InformationWeek recently published a great article, Coke’s RFID-Based Dispensers Redefine Business Intelligence.

Coca-Cola has introduced a new RFID-enabled drink dispenser called Freestyle. It provides mass customization with 30 flavor cartridges (sodas, teas, flavored waters, etc.) that can be mixed to make more than 100 different drinks (out of the more than 3,000 that Coca-Cola sells around the world).

Starting this summer, the new dispensers will be available in fast-food joints across California, Georgia, and Utah. Customers can select a brand, such as Sprite, and are then offered several variations (Peach Coke? Grape Sprite?).

The RFID-enabled dispensers keep track of consumption trends (flavors, quantities, times…) and transmit that information each night over a wireless network to point-of-sale management software from SAP, and then to Coke’s SAP BW data warehouse in Atlanta (Coca-Cola has standardized on a common SAP backbone across 45 countries, and uses a mix of Microsoft and BusinessObjects front-end BI products.)

The data will be used for operational activities (automatically scheduling refills, stopping distribution if there’s a product recall of a particular cartridge, etc.) and for analytics: assessing how new drinks are doing in the market, identifying differences in regional tastes, and helping fast-food outlets decide which drinks to serve.

Coca-Cola is hoping to use its new analytic power to speed up and improve test marketing processes, with the ability to try out new flavors with almost real-time feedback. That information can then be quickly shared throughout their worldwide distribution channels.

I’ve tried to imagine what a dispenser analytics dashboard might look like, using Xcelsius:

Disclaimers: (1) This is purely speculative, and not endorsed by Coca-Cola in any way; (2) This is for fun, NOT an example of dashboard best practice. If you’re interested, see Stephen Few’s blog about vendors abusing Xcelsius in this way

It’s also worth noting that Coca-Cola has used Xcelsius in the past, for example as part of the launch of Coke Zero in Australia:

Links

• Coke’s RFID-Based Dispensers Redefine Business Intelligence
• Coca-Cola Goes Freestyle
• Coca-Cola’s Business Intelligence Strategy
• The Coca-Cola Company Invites Consumers to Experience “Freestyle”
• Coca-Cola readying 100-flavor soda fountains

Later today, the Tour de France will be rolling into Paris to complete the final leg on the Champs-Elysées, so it seemed a good time to share this story of business intelligence and bicycles.

The city of Paris recently rolled out their latest big initiative in making Paris a more livable, green city: 10,000 bicycles, distributed across 750 “stations” across Paris, as part of a program called “Velib” — from “vélo” (bicycle) and “liberté” (freedom).

The bicycles are designed for city use — solid, but comfortable, easy to adjust, and supplied with a basket, lights, and lock. To take a bike, you just swipe your credit card, grab a bike from one of the locked stands, and return it any station in the network. The prices are very reasonable: 1 euro for a day pass, 5 for a week, and just 29 euros for the year (I’m personally a huge fan already, and I’m eagerly awaiting my yearly pass).

The city doesn’t actually run the program, they outsource it to JC Decaux, a big outdoor advertising specialist, in exchange for an exclusive contract for the city’s advertising billboards. And JC Decaux is a big user of BI software, using it to analyze the profitability of different advertising placement strategies.

The Velib project has been an unqualified success, with over 150,000 users in the first four days. JC Decaux’s operational systems provide real-time information, so that you can go to the web site and click on a station to find out how many bikes are available (although unfortunately it’s not yet easy to get the map to appear on a mobile phone, where it would be the most useful).

BI is going to be key to the optimal running of the organization (and presumably to whether JC Decaux can make their desired profit on the service).

One example is keeping maintenance costs down — the bicycles have to be serviced on a regular basis — there have some teething problems with the stations themselves. In the first week or so,  to 5% of the stations have had some sort of problem, including occasions when users returned their bicycles, but the station refused to register the return, and continued to charge usage fees.

Another big problem is going to be optimizing the usage: currently, at any given time of the day, some of the stations have no bicycles, and some are full (so you can’t return a bicycle to that stand, and have to find another one — from personal experience, this is very annoying..).

There are teams of trucks that move bicycles around Paris to make sure they’re in the right places — but as you can imagine, this is an incredibly complex operation to optimize, since it’s based on the time of the year, the weather, the state of traffic, whether the Tour de France is in town, etc.

I believe this is currently being done in a relatively manual, operational way (i.e. when the stand is empty, move bicycles to it), but BI will be the key to optimizing the system, with proactive distribution in advance of problems, rather than after the fact. After a solid history has been built up over time, JC Decaux should be able quickly react to any differences between planned and actual operations. Disneyland Paris have already taken a similar approach, with great success.

The Velib project has been so successful that the number of bicycles and stations is set to double in the next few years (and extended to the near suburbs), and other cities — notably London — are strongly thinking of following Paris’s lead.

Photo Credits:
http://www.flickr.com/photos/malias/821755467/
http://www.flickr.com/photos/luc/917511203/