Profit Center: Beyond the Magic
By E. Andrew Boyd
In my experience, most people come to their first analytics project in one of two ways. Either they have a great idea, or someone has it for them — typically, the person they report to. It’s not surprising. Great ideas for using analytics arise all the time.
“We’re picking prices out of thin air. Can’t we use historical data to do something more intelligent?”
“Our delivery drivers are sitting idle for half their shift. Can’t we help the dispatchers do a better job?”
“We keep records of every sales transaction on our Web site. Can’t we use this information to drive more sales?”
If you’ve managed an analytics project before, you already know how to go from great idea to great deliverable. But if you haven’t, there’s something important to keep in mind: analytics brings tremendous value, but it isn’t magic.
Why is this so crucial? Because the surest way for a project to fail is when expectations aren’t met. And it’s hard to meet expectations if they’re based on magic.
Forecasting provides a particularly good example. Forecasting projects often start when a company badly misses some sort of projection, like future interest rates or how many customers will buy a product. The decision-maker who uses the forecasts then tasks someone with improving them, and an analytics project is underway.
Forecasting is one of the great analytical tools used in business. Companies that rely on mathematical forecasting models do better than those that pull numbers out of the air. But with mounds of data at their disposal, decision-makers’ expectations are frequently out of touch with reality. They believe magic will lead to nearly perfect forecasts. The problem is exacerbated if third parties are involved — consultants or software vendors — since they have an economic incentive to sell magic.
Without a good foundation in reality, the project will fail painfully. However, if everyone involved reaches a consensus that analytical forecasting is a tool — a tool that can be worked and modified to create better, if not perfect, results — the organization is on its way to becoming analytically enabled.
Great Success Story
One of the great success stories with analytics is revenue management. As consumers we experience revenue management as frequent changes in the price of airline tickets (and hotel rooms, rental cars and many other items). It’s one of the most advanced analytics applications in the world, with leading airlines employing staffs of Ph.D.s to manage the price of their seat inventory.
One of the fundamental tasks in revenue management is determining how many people will purchase a ticket at a given price: “For flight 1234 departing Saturday, March 13, at 10:40
a.m. from Chicago to Orlando, how many tickets can I expect to sell at $347?” Of course, the actual number is governed by an underlying random process, so the forecasts are just that — forecasts. Better or worse forecasting methods may be employed, but none will be exactly right every time. That’s one of the few statements we can make with certainty about forecasting. The best forecast for the sum of two dice is seven, but snake eyes is always a possibility.
There are other limitations. Holidays and special events present challenges. When the Super Bowl’s in town, normal traffic patterns are altered. When an airline shifts from its winter to its summer schedule, forecasts can be shaky until enough new observations are gathered. For these and other reasons, trained analysts are employed to review and modify forecasts when necessary. The very best airlines have multifunctional teams that work closely with their analytical systems as an extension of those systems.
The important point is that airlines accept the limitations of their revenue management forecasts because they do better with these forecasts than without them. There’s no magic. It’s work — like tightening bolts on an engine or serving drinks to passengers. But it’s work with a big payoff, as Donald Burr, CEO of now defunct People Express Airlines learned too late. “We were a vibrant, profitable company from 1981 to 1985, and then we tipped right over into losing $50 million a month,” he said. “We were still the same company. What changed was [American Airlines’] ability to do widespread [revenue] management in every one of our markets” .
Getting beyond the magic can be difficult. It’s important to maintain enthusiasm about a project.
Analytics can do tremendous things, and there’s every reason for team members to be excited. But for a project to be successful, it’s necessary to balance enthusiasm with practical reality.
Games Provide Powerful Message
One of the best techniques for achieving this balance is through training in the form of specially designed analytical games. Games have the advantage of keeping participants actively involved while driving home essential points. The very best poker players lose many hands, but they know that over time they will almost certainly win, while poor players will almost certainly lose. It’s a powerful message about how analytics works and helps emphasize the three important points all team members must understand:
- Analytics brings real value. Analytics has proven its value many times over, and to remain competitive in the future a company must be willing to adopt an analytical approach to problem solving.
- Analytics won’t lead to perfect decisions every time. Making better decisions isn’t the same thing as making every decision perfectly. But the value of making consistently better decisions adds up.
- Analytics requires work. Like everything else in life, there are no magic buttons.
Andrew Boyd served as executive and chief scientist at an analyticsfirm for many years. He can be reached at email@example.com.
1. Cross, R. G., 1997, “Revenue Management: Hard-core tactics for market domination,” New York: Broadway Books, p. 125. (www.amazon.com/Revenue-Management-Robert-G-Cross/dp/0553067346)