Our management systems are based on technical data, often maintained by dedicated teams. Rigor and discipline are required to maintain this data on an ongoing basis. If you have participated in an ERP implementation, you know how much attention is paid to the subject — which sometimes drifts a bit afterwards.
But what is sometimes hard to admit is that these data are essentially estimates, like most of the elements we manipulate to drive our supply chains. Forecasts are estimates. Requirements dates and delivery dates are estimates. Bills of material are estimates (actual material consumption will differ). Inventories are estimates… Um, not inventories, right? Although…
Supply, production, and transport times are — definitely — estimates. They are sometimes contractually defined, with suppliers for example, but they remain approximations.
The events of the last few months — Covid, electronic components, geopolitical developments — have shown even more acutely how much uncertainty there is in lead times.
At the same time, lead times are crucial: how can we ensure the availability of our products and maintain a high service rate if we don’t know how quickly we can replenish?
The “Roughly Right” Philosophy
When I was a student, my physics teacher used to tell us: focus on orders of magnitude rather than on precision!
To make timely decisions and arrive at tenable availability dates when we work to order, we need reasonable, pragmatic lead times in our systems. That allows us to size our replenishment loops and stocks more or less correctly.
Periodically, this might mean measuring our actual lead times, calculating their mean and standard deviations, and comparing them to the lead times we’d set. However, the Demand Driven model’s “roughly right” philosophy suggests that organizations should not use these figures to automatically calculate and adjust lead times in the ERP, because it adds noise and instability. When you measure lead times, as when you calculate average daily usage, you should average your data over a period of time, and clean them of exceptional events.
In our experience, this maintenance should occur as part of the DDS&OP process, for example on a monthly or quarterly basis, to ensure that the operating model remains realistic.
Here are some additional tips for mastering lead times for purchased items, manufactured articles, and distributed items.
1. Purchased items: Tame chaotic supply lead times
Procurement professionals have come to realize lately that lead times are partly empirical estimates. They reflect both the organization and the capacity of suppliers, and they can be affected by shortage and allocation situations. It is not uncommon today, for example, to see lead times of one or even two years for electronic components, when not so long ago these same components were available within 3 or 4 months.
Purchasing organizations increasingly have to commit to long-term volumes, based on forecasts that are known to be wrong, and especially one or two years out. This is the law of the current market for several commodities, and geopolitical instability means that we will not get out of these constraints in the short term. These volume commitments are intended to be derived from DDS&OP projections, with scenario evaluation and validation of the level of risk taken.
DDMRP buffers should be sized according to the call off horizon, which is generally shorter than the commitment horizon. They allow you to control the flow of supplies at the rate of real demand in the coming days/weeks, and to adjust execution. When you design your model, make the distinction and ask yourself the question of the relevant lead time to size the buffers on your sites.
2. Manufactured articles: Squeeze frozen horizons
For manufactured items, the lead times used are the result of a planning horizon and load smoothing processes, rather than the actual time it takes to complete the production operations. Experience shows that the more you freeze, the more you generate instability in production schedules and the need to reschedule frequently. A firm horizon is the enemy of stability. To determine the relevant lead times, take a pragmatic and gradual approach to reducing firm horizons, by challenging the status quo, positioning decouplings, making component availability more reliable, and moving from weekly to daily planning buckets.
It is this planning horizon that determines how much time will elapse between a replenishment signal is triggered and the completion of the goods.
In a complete operational model, these lead times will be tested in capacity and will integrate the time buffers established to protect against operational variability.
Distributed items: 3. Transport time is of the essence
For distributed items, the transportation time determines the lead time to be taken into account, i.e., the time between the initiation of a transfer order and its receipt in stock at destination.
Lead Time vs. Frequency
When sizing inventory or committing to customer dates, two factors come into play: the lead times mentioned above, and the frequency of supply/manufacturing/shipping. There is often confusion between the two.
On a DDMRP stock buffer, the lead time will size the yellow zone, and to a lesser extent the red and possibly green zone. The frequency will mainly affect the green zone, and to a lesser extent the yellow and red zones.
Continuous Improvement in a Demand Driven World
The Demand Driven approach is very structuring. The first step is to visualize the company’s flows, to take two steps back and to ask ourselves these basic questions: what lead times to use, what frequencies, are our firm horizons adequate, etc. Once the model is in place, the performance reports will allow us to measure its execution over time, and to ensure that we are constantly maintaining a relevant model.