It is quite a paradox that whilst planning is a future thinking and forward-looking exercise the performance of planning is always measured in terms of past and YTD performance.
As an example, Demand Planning teams spend a lot of time in analyzing their forecast performance in terms of different metrics such as MAPE (mean absolute percentage error) and Forecast Bias (tendency to over-forecast or under-forecast).
These metrics are great indicators of which areas the plan is off-track and not performing vs. the realized actuals.
Various views of these metrics are created:
Demand planners spend a lot of time in doing MAPE and Bias trend analysis, looking at exception areas such as high MAPE scenarios, upward trending MAPEs, consistent Bias in either over-forecast or under-forecast direction, etc.
In the overall organizational planning processes this aspect continues. For example, at the category level there is a keen view on Category level performance in terms of YTD Volume contributions vs. YTD Error Contributions. What has changed in the YTG (year to go) forecast (latest estimate vs. past latest estimate), etc.
As a natural result, a lot of the decisions and actions are based on the analysis of plan performance and errors of the past or as they are occurring.
Now, one might make the argument that the only way to test your predictions is to measure how they stood up against what really transpired. And I agree with that whole-heartedly. However, what that implicitly does is to subdue the “other question”.
While we should definitely ask “How has my plan performed in the past or until now (the present)”, it is as, and if not more, important to ask the question, “How will my plan perform in the future?”.
This simple change in line of thinking reveals what I feel is a big gap in the mind-set of planning today. What if planners were to pay as much attention to the following questions:
Planning processes often work with the implicit underlying assumption that the places that the plan has been going off/has gone off-track will be the places that it will go off-track in the future. However, that is a fallacious assumption and leads to significant blind spots.
The complexity and volatility of the operation (constant pricing changes, competitor actions, weather, consumer preferences etc.) makes it difficult to anticipate where the plan will go wrong in the future. Demand and Supply planners in CPG companies are always trying to catch up and react to new situations and exceptions. This results in imbalance in day-to-day operations and reactive decision making.
We see this every-day, in planners being inundated with exceptions to planning, drastic forecast changes and firefighting. It is not uncommon to see planning system alert modules generate 1000s of exception alerts on a weekly basis for global CPG companies.
While there are genuine exceptions that cannot be anticipated (e.g. Covid-19, etc.) many exceptions and alerts are a result of planning blind-spots.
It would be amiss to say that all planning teams operate like this. While we all instinctively understand that the “problem areas of the past” are seldom the “problem areas in the future”, it is difficult to put this into practice in a tangible way.
In recent times there has been emphasis on scenario planning and simulations to guide planning but only in the context of tactical and long-term planning (e.g. for S&OP, IBP, etc.).
In most conversations related to operational planning the past and present performance of the plan is discussed the most since it is the only tangible measure available to planners.
And this is where AI/ML has a very unique and high impact role to play. Anticipative, learning and context aware AI/ML can help planning teams anticipate the problems in the future by predicting the likely deviations from the plan. Planners can have quantitative, objective, early warnings as to where the plans might go off-track and more importantly why?
And, planning can move from the focus on the “performance-in-the-past” to the “performance-in-the-future”
Welcome to agile demand planning. Our engine blends intelligent cognitive automation and collaborative workflows to spot human bias, create transparency, and uncover previously hidden probabilities that will impact future inventory, promotions, and sales.