How should companies improve their sales forecast accuracy?

Most companies struggle with improving the accuracy of their sales forecast. They try different forecasting software, using different time series models. Some companies even look for an experienced demand planner from a leading FMCG company. Does it really help? My experience with several companies in this space indicates that such actions don’t result in any substantial improvement. The accuracy in most cases goes up by just 1-2%.

What should we do?

The solution lies in revamping the complete process of sales forecasting. Do it at the granular level to catch differing patterns. Use automated AI/ML based self-learning models instead of plain time series ones. Increase the frequency of predictions. Use as many demand drivers as you can think of and let the algorithms select the relevant ones.

Once we incorporate these changes, we actually move from demand forecasting to demand sensing!