When businesses try to predict their future demand, they often begin with a mathematics based statistical forecast. Once this is established, some users are asked to provide overrides that can make this statistical forecast better. Typically, this is asked of users who have a client facing role. For example, sales representatives and sales management are often asked for their input. However, depending on the organization, others are asked for their input as well. This type of forecast which is based on inputs from users is called collaborative forecasting.

A question that often comes up in my interaction with clients relates to the need to do collaborative forecasting. Since a statistical forecasting system is often in place, the users sometimes questions the benefits from this collaborative input. This problem is often exacerbated by the perception that very often the user’s input is making the forecast worse.

Let me first state that research has shown that the collaborative input from knowledgeable users makes the forecast better. For example, in a study done on weather forecasts, human input improved the forecast by 20-30% based on which computer forecast was used as the baseline. (Source: Slide 5 ) So, there is proof that this input can create value. In a business setting, 10-15% forecast accuracy improvement has been reported.

Assuming that, an unbiased statistical forecast is available to create a baseline statistical forecast. Here are 5 reasons why a business should strongly consider asking their users to provide collaborative input as these make the forecast better by adding information to the process.

  1. Users can bring qualitative information into the forecasting process.
  2. For example, users might know about new products leading to cannibalization of demand for old products, promotions being run by their customers or perhaps about a planned shutdown.
  3. Different users might have different information. For example, a CSR might have information on an impending shutdown about which sales person might have no information. Or the CSRs might have more knowledge in the near term whereas the sales rep might have more information in the medium term.
  4. Users can see through the numbers: The need for this is best demonstrated by this often cited example of green cars from a particular company. Owing to lackluster sales, the sales department decided to get rid of the excessive stock by offering attractive special deals. The deals created the desired effect and the cars started moving. A statistical forecasting engine relying solely on the numbers would tend to increase the forecast. By comparison, a user would understand the underlying reason for the increased sale and would resist the urge to increase the forecast.
  5. Whenever significant portions of the demand is project or contract driven, statistical forecast will have limited efficacy by way of an accurate forecast. A forecast provided by a knowledgeable user is the best way to a reasonable forecast in such a case.
  6. Forecast is all about future sales, which requires commitment to the numbers on the part of those who do the selling. Collaboration helps create the necessary buy-in from the sales team and creates the necessary commitment. Nobody feels committed to reaching a statistical forecasting based target. Collaboration often results in agreement about the forecast and therefore commitment.

For more information on this topic please feel free to download our Collaborative Forecasting whitepaper.

Like this blog? Follow us on LinkedIn or Twitter and we will send you notifications on all future blogs.