“I skate to where the puck is going to be, not to where it has been.” For forecasters, the ultimate dream is to be able to read the game like ice hockey legend Wayne Gretzky did. With big, dynamic data entering the arena, that uncanny ability—or “gift” if you like—to read the demand game comes within reach. Find out how to improve your forecasting techniques below:
Professional ice hockey is a difficult job—and so is forecasting. Still, companies have to constantly review forecasting needs and efforts made to see where they can improve their process and, eventually, decrease forecast errors. Then again, the advantages of good demand planning and forecasting are numerous and well-known: lower safety stocks, less slow-moving and dead stock, better inventory rotation and so on.
In the past, a lot of companies used to stick to rather simple and classic forecasting methods, for instance, trying to predict the future based on what had happened in the past. In the modern business context, this way of forecasting is increasingly coming under pressure.
“We’re witnessing constantly shortening product life cycles, rapidly changing customer behavior, new trends, new sales channels—to name just a few examples—for which we often don’t have data.”
What’s wrong with your current forecasting techniques?
- You only use data from your own company: In classic forecasting, we see a strong focus on intrinsic forecasting methods. These are methods in which a company’s own data—mostly historical data such as order details—are used. It works like this:
- A base forecast is generated through statistical models (simple or more complex ones).
- The forecast is enriched with information that comes from outside the statistical models, e.g. promotion planning.
- Your forecast demand is rather static: Demand management processes are often rather static. Forecast demand is seldom monitored (and even more rarely adjusted) over time, but usually, as time goes on, the more information you will have about the future. Even in cases where freshly gathered information indicates that a forecast will turn out to be clearly wrong, very few companies will consider this new information and adjust the short-term forecast. Think of an unexpected upswing in orders, a substantial future order that has already consumed the forecast, information on an additional promotion by the sales department, etc. Most demand forecasting and planning systems are not good enough to notice these demand signals. Consequently, you tend to keep focusing on the static forecast outcomes based on past data. Or in other words, once we think of a number, we hardly ever change it afterward.
- Your attitude is too much “leaning back”: Supply chain latency is often accepted and rarely questioned. This means we are willing to wait until a customer order comes in instead of trying to capture information earlier via other signals. So, today, most companies and their supply chains respond to an estimated demand (which does fulfill incoming orders), but they don’t really sense the current demand due to the focus on past data and a long-term static forecasting process.
Do you think Wayne Gretzky would have listened to a coach who just wanted him to play as his team had always done?
- Gretzky anticipated where the puck was going and foreshadowed where the play might take him. You may not be able to pride yourself on the instinct Gretzky had for his sport, but as a forecaster, you have access to an enormous amount of data that allows you to read and anticipate the demand play just as well. The big challenge is, however, to scale and process the massively growing volume of unstructured data. A lot of information is left untouched despite the fact it can help to identify new business drivers that are already affecting customer demand.
- Gretzky reached maximum performance and production.
Playing almost every other day and scoring a total of 894 career goals, Gretzky was a relentless athlete who never leaned back. Adopting this mindset as a forecaster means you have to translate those insights into actionable information over the short, medium and long term.
Detecting demand information at an earlier stage and taking this into account in the forecast is what we call demand sensing. If you master the appropriate methodologies and technologies, this might be a big step toward increasing forecast accuracy.
Want to learn to read the forecasting game in the Gretzky way? We might be able to indicate where the puck is going to be.
In our next blog, we will take you there and introduce you to the world of demand sensing and demand shaping.