Earlier this week, a colleague brought to my attention a company called Planalytics. Actually, at a supply chain conference several years ago, I attened a presentation by a manufacturer that was attempting to use Planalytics Business Weather Intelligence in their supply chain planning processes. At the time, weather-driven demand planning seemed interesting, but not ready for prime time. However, if you go to Planalytics’ website today, you’ll find several recent customer presentations, many of them from retailers who are using long range weather forecasts to help them make better merchandising and buying decisions. (I’d need a lot of convincing before using long range weather forecasts in this manner).
There are also some presentations by manufacturers, who are more apt to use weather information for either demand planning or replenishment execution. The presentation by Randy Kitchin, National Supply Chain Manager for Nestle Waters North America (“Nestle”), was the most interesting to me. I will summarize some of the key points here, but I recommend you watch the presentation for all the details.
Nestle is using the Planalytics data for long term demand and supply planning. The company calculates that for every percentage point of improved demand accuracy for monthly forecasts they could save over one million dollars (Nestle does over $4 billion in business annually in North America). If they could improve the weekly forecast by that same margin, they could double the savings. In the past, Nestle had been growing so fast that statistical forecasting was not accurate. But in recent years, as the pace of mergers slowed and their growth settled into single digits, statistical forecasting became more feasible. The weather data helps Nestle clean up their baseline statistical forecast. Water sales are highly correlated to hot weather. In some years, the weather can be warmer than usual, in other years colder. Planalytics claims that the weather in a particular region only repeats itself 25 percent of the time. If you are dealing with historical data from a small number of years, you will not have a good baseline forecast. So, the Planalytics data helps Nestle to “deweatherize” their forecast.
Nestle also uses weather data for short term replenishment decisions. The company prefers to carry very low inventory (water is a low-margin product that is heavy and costly to store and transport) and having regional short-term weather forecasts helps them avoid this. Nestle receives a report that lists every distribution center they deliver to, along with how much additional or fewer product each DC will need based on the weather forecast. The company then communicates these findings to customers and they work together to change short-term order allocations, with some warehouses for a particular customer taking more inventory than they had initially ordered and other locations taking less. Further, if there is apt to be a large surge of work in a particular Nestle warehouse, the forward visibility allows Nestle to better schedule their warehouse workers.
While hot weather can lead to regional demand spikes of 20 to 30 percent, the threat of a hurricane can lead to spikes of 500-600 percent. However, by the time the National Weather Service forecasts the path of a hurricane, the cost of transportation has soared. Weather forecasters at Planalytics can predict a hurricane’s path faster than the National Weather Service, which allows Nestle to lock up trucking capacity earlier at a more reasonable price.
For particular product groups, it’s not surprising that weather can greatly impact demand. The impact will differ by product group and region. A cloudless, 50 degree day in Naples, Florida in April will lead folks to stay at home. In Boston, a similar day in April will lead people to rush outside to enjoy the weather, which would lead to a surge in water sales and perhaps spring clothing. The customer case studies on the website lead me to believe that Planalytics has taken giant steps forward in making weather data actionable.