Spare parts forecasting accuracy is driven by the validity of the premise and the data that is used in the analysis, not the sophistication of the modeling
Making predictions is hard, especially if they are about the future!
This is a variant of a quote that is attributed to the physicist Niels Bohr, baseball manager ‘Yogi” Berra, writer Mark Twain, and various others. While the origin of this quote may be in dispute, surely the issue that it raises is not. What happened in the past is not necessarily an indicator of the future (now I sound like a financial services ad!)
In the world of spare parts management, spare parts forecasting is often like a game of Whack-a-mole. For those that don’t know (or don’t care to admit they remember), Whack-a-mole was an arcade game, developed in the 1970’s, where a mole would pop up through one of several holes and the object was to ‘whack’ the mole with a hammer.
Promoting cruelty to animals? Perhaps.
Unnecessarily violent? Sure.
A model of the way that many companies seem to approach spare parts inventory management? Absolutely!
Think of the mole as demand for a spare part and the player like the hapless storekeeper. In this context the game is an analogy where demand (the mole) is unpredictable and all the storekeeper can do is react to what comes up next. Sound familiar?
The problem, of course, is that essentially all spare parts inventory stocking problems are really spare parts forecasting problems. This is because the very essence of spare parts inventory management is determining the most appropriate level of inventory to hold, to service the expected future demand for that inventory, based on the expected supply constraints. Thus, all inventory management requires a forecast of both demand AND supply in order to establish the buffer that needs to be held to match these two factors.
When most demand is based on (apparently) random failure events, as with MRO and spare parts, then demand, almost by definition, is impossible to forecast. A bit like that little mole sticking its head out of the hole. So, how do you make spare parts forecasting work?
Forecasting Methods for Spare Parts
All spare parts forecasting methods can be grouped into one of two classes: either, extrapolation of historical data or causal and predictive models
Extrapolation of historical data is a fancy way of saying that you take whatever happens in the past and assume that it will happen again in the future. The approach for doing this can vary from the simplistic to the highly sophisticated but importantly all historical data methods are based on the premise that the future can be predicted by looking at the past. Something that we know to rarely be true.
The methods employed are typically quantitative and because of this they can appear to be rigorous. However, the accuracy achieved is driven by both the validity of the fundamental premise, that is, can the future of that part be predicted by looking at the past, and the quality of the data that is used in the analysis, not the sophistication of the modelling.
On the other hand, causal or predictive methods can be either quantitative or qualitative.
A quantitative approach might rely on forecasts of future planned activities (such as planned maintenance) and the expected usage of the part in each activity.
A qualitative approach might rely on the opinions of people involved in using and procuring the parts. For example, let’s say that a maintenance team member wants to have a spare part stocked and the decision on how many to stock is based on their opinion of how many to stock. This is a qualitative predictive method and is not an appropriate way to determine stock levels. Unfortunately, the natural conservatism of maintenance personnel typically results in significant overstocking.
It is this aspect of causal or predictive methods that often makes people think that all such approaches are less accurate than historic data driven approaches. However, this is not the case because, with spare parts inventory, causal approaches and the use of forward-looking information is more appropriate for deciding future inventory holdings than relying on the extrapolation of history. The key is to ensure that the basis of decision making is more robust and consensus-driven than just ‘today’s opinion’.
Which Forecasting Method to Use?
The choice of which forecasting method to use should be determined by the situation and the information the forecaster can access.
However, as a general rule of thumb, if solid causal information is available it should be used in preference to relying on historical data.
The following rules are helpful in determining which forecasting method to use:
- When the forecaster has access to clear information regarding factors that will influence demand, this information should always be used to generate the forecast. Predictive information should take priority over extrapolation of history.
- When the forecaster does not have access to useful causal information, and demand for the product is consistent and stable (rarely the case with MRO and spare parts), moving average is a simple method that should produce acceptable results.
- When the forecaster does not have access to causal information and demand for the product is known to exhibit clear patterns relating to growth, decline or seasonal behavior it will be appropriate to use a more complex technique that specifically addresses the known demand profile.
- When the forecaster has no causal information and the demand is thought to be unpredictable and the item is considered to be critical, then it is best to openly acknowledge this situation and apply a consensus-driven, risk based approach.
Note however that the forecast alone does not determine the holding levels, it is only one factor to consider.
An everyday example of the last rule is the holding of a spare tire. In this case, and thinking about a puncture as the potential failure, it is clear that there is no causal information and that the past is no indicator of the future. That is, the fact that a puncture was experienced last month is no indicator that another will be experienced this month. The decision to hold that spare tire is therefore entirely dependent upon the impact of a puncture and the other options available at the time. These options may include: buying expensive ‘run flat’ tires so that you are not stranded, holding a can of puncture repair ‘goo’ (effectively, a different type of spare), accepting the inconvenience and waiting on a repair, or holding a spare tire. This decision is not based on any history or causal information but rather an assessment of the impact of a failure with the expectation that a failure (puncture) is probable.
By understanding that an inventory stocking problem is essentially a forecasting problem and that all forecasting methods have their limitations, it is easy to see that good spare parts inventory management relies on an understanding of both these limitations and the application of the techniques in practical terms, not just in theory. This is why training is the most important task you can undertake to improve your spare parts inventory management outcomes. Without this, you may as well be playing whack-a-mole.
SparePartsKnowHow.com members can also download a white paper on spare parts forecasting at the Operations section of the Resource Library.
You might also be interested in:
An article titled The Management of Spare Parts Inventory is Different – Here’s 8 Ways
A video on forecasting spare parts requirements.
For information on our Pro Level membership please visit our Pro Level page.
Posted by: Phillip Slater