I'm in what's a very foggy Chicago at the moment, counting down the hours to when we'll be kicking off our SAS pre-conference workshop at INFORMS. I'm very fortunate to be here this year, because I've always been a big believer in using optimization to drive returns.
The greatest thing about optimization is that intuitively, everybody understands how important it is. It's something we do every day, whether it's trying to avoid sitting in traffic by taking the back roads, fitting in a bit of exercise over lunch to have more time in the evening, or spending hours of obsessive research trying to save $10 on new camera gear.
OK, maybe that last one's just me, but the point remains: optimization is one of the few areas of analytics that's really intuitive. It's about making things go further, finding better answers, and making the most of what are often limited resources. Fundamentally, it's about generating returns. Anyone who had an allowance as a kid learned very quickly the importance of stretching it as far as it would go! What isn't necessarily as intuitive is how easy it is to take that approach and apply it to drive profitability, let alone how much of an impact it can have. For example, one of our customers is an intercontinental airline concerned about the amount of fuel they were using. That in itself isn't surprising, as pretty much everyone's discovered over the last few years, fuel's expensive. What was surprising (at least to me) was the sheer amount they had to spend to keep their planes in the air. Without flinching, they were talking about spending billions of dollars just on their international legs! With numbers that large, it wasn't a huge leap to realize that the potential savings were huge.
In itself, optimizing the amount of fuel they were using didn't seem like an overly complex problem: - given accurate weather (difficult to do on some of their longer legs) and load forecasts, it was simple algebra to work out how much fuel they'd need to purchase to avoid being fined by the FAA and, ideally, get the plane to where it was supposed to go. Unfortunately, as with many things in life, things weren't that simple.
To give them the ability to plan with some degree of certainty, they'd created a variety of commercial agreements with various ports to lock in pre-agreed fuel prices and minimum purchase amounts, both of which varied by port. In isolation, that's still fairly simple; what made things really complex was that every extra bit of fuel they had over the minimum needed for the flight actually added weight to the plane, increasing the fuel they'd need to purchase!
Overlay a relatively complex international flight plan with short-notice cancellations and re-routing due to mechanical issues and weather patterns, and all of a sudden what looked fairly simple problem became very confusing! It's in situations like this that analytical optimization shines.
In the past, no matter how hard they'd tried to optimize their purchasing process, it simply didn't compare the cost reductions they were able to identify using operations research. Once they tried optimization, with relatively little work, they were talking about hundreds of millions of dollars of savings! And, possibly even more importantly, they were actually able to do sceneario-testing interactively, comparing optimal answers based on different scenarios and using that to influence their purchasing and negotiation strategies.
More than anything else, this is why I'm so looking forward to INFORMS next week: it's a rare opportunity to learn from some of the best and brightest in the field. Optimization and operations research can be applied anywhere, from workforce planning to direct marketing to pricing and almost inevitably, it reduces highly complex problems into simple answers. One the best things about working at SAS is being able to cross-pollinate great ideas into new domains - knowing what's possible is often half the battle.