"Everybody has a plan until they get punched in the face" - Mike TysonAs much as you plan, as much data you collect and analyze and as much as you predict, you never can be 100 percent correct. Heck, you’ll probably be correct just a small fraction of the time. As more out-of-your-control variables become part of your business, the less you can predict the future. Instead of trying to, why not have a business and production model that can respond in real time, asked Tony Haile at the Future of Web Apps conference in Las Vegas. Haile is the general manager of Chartbeat, a real-time Web monitoring application. Don't pour all your money and effort into trying to create better prediction models. Make the problem of an uncertain future not matter, Haile said quoting Taiichi Ohno, the father of the Toyota Production System. As an example, Haile used the clothing industry, which has a six-month production cycle from concept to retail. The entire industry tries to predict what people will wear in the future. That’s a costly process (research and analytics), and risky (still have lots of misses, which adds even more cost). Conversely, clothing manufacturer Zara realized the way to surmount this was to simply stop predicting so it could employ a faster production timeline. It reduced its process from six months to 15 days. The company no longer has to spend money on predictions, nor does it have to gamble. They simply look at what people are wearing, and produce that. Monitoring real-time data is a great idea, but you need to have a production plan to actually take advantage of it. Real time data isn’t worth anything if you can’t respond in real time, observes Haile. Don’t predict. Simply have a better system to respond.