Baseline Magazine in one of my favorite reads (right next to InfoWorld, honest) for information about enterprise computing. Their articles are detailed and usually tell a story in an analytical way. This month's issue has a detailed article, with numerous sidebars, about MacDonald's decision to cancel Innovate its five year, $1B program to build a real-time system for monitoring everything about its 30,000 stores, right down to the temperature in the fry-cooker. They'd already spent $170M which is largely just money down the drain at this point. A few weeks ago I wrote about the real-time enterprise and this is a topic that is near to my heart---a move that I think is inevitable in business. So, what's the deal?

MacDonald's challenges are painfully obvious to anyone who's visited one of their restaurants lately. Many are out of date, the service levels have slipped, and the menu is rapidly falling out of step with what many want to eat. At most you can't even pay by credit card. To add insult to injury, the bathrooms aren't as clean as they used to be. So, how's a $1B real-time digital network going to solve those problems? That's the very question the new CEO asked when he took them helm last year. Consequently, he decided to make investments other places.

I guess I can't really blame him, but you've also got to wonder when the aging IT infrastructure is going to get its due. The Big Mac still does its books on a mainframe-based, custom built general ledger system conceived and built in the 1980s. Company executives, can't really get detailed information about sales in individual stores and what data they can get is usually a week old. This doesn't sound like a system for getting back in touch with your customers either. A couple of interesting quotes from the article:

[I]nstead of investing in Innovate over the next five years, Cantalupo [CEO] says McDonald's will invest in itself through the share repurchases and dividends. These measures might provide temporary relief for the beleaguered stock price but will do little to improve the quality of the food or the speed of service at its locations. But then technology has never fit easily on McDonald's menu. "Culturally, it was always a fight at McDonald's," Dill [the then CIO] says. "My first day on the job I remember meeting with then-CEO Fred Turner and he said 'Carl, I never want to fail to sell a hamburger because a computer is down.' McDonald's just wasn't comfortable with technology."

A few comments:

• The biggest problem with projects like this is the sheer size. MacDonalds spent $170 without even rolling a single thing out---just on pilots and testing. Eighty percent of big projects go awry. You have to find a way to do these in chunks or you're setting yourself up for failure. Even chunking this in terms of systems (update the general ledger first, work on POS systems next, etc.) would have help, in my opinion.
• Companies like MacDonalds are trying to solve multiple problems at once. At the same time, their size makes the problems enormous. Most companies don't face these same issues of scale. Smaller companies shouldn't apply MacDonald's lessons to themselves without allowing for scale.
• Web services provide a means of doing enterprise application integration (lowercase) in an iterative way. Connect up the things that matter most and then start on the next tier, and so on.
• Iterating to integration doesn't obviate the need for a plan. This is called an enterprise architecture. I'm pretty sure I'll have a lot more to say on enterprise architectures over the coming month.
• MacDonald's failure notwithstanding, I think that real-time enterprises are inevitable. Why? I call it the Fed-Ex principal. If you're competition is using Fex-Ex to move the mail and you're not, they get documents delivered faster. Soon everyone has to use Fed-Ex and even though no one is advantaged by it, no one can afford to not use Fed-Ex. The same thing will happen with real-time enterprises. Some companies can pull it off (witness Wal-Mart). This means that they are at a significant advantage over their competition. Their competition will either become real-time or die. Eventually everyone will be real-time and there will be no advantage, but no one can go back. This is, in part, the argument that Nicholas Carr was making.

If you care about enterprise systems, this article and the accompanying sidebars deserve careful study.