Email
sinha.rk.dr@gmail.com
What should be the trigger for buffer changes?
We discussed in the last post that buffer sizes should be based on the forward demand prediction at the level where inventory buffers are kept, instead of the past consumption. We also saw how this would lead to better availability and lower inventory. The next question is… when should we revise the buffer sizes?
Most companies follow the conventional way of increasing buffers when inventory is consistently in the ‘Red’ zone for 1-2 RLTs. Similarly, buffers are reduced if the inventory is consistently in the ‘Green’ zone for 1-2 RLTs. While this methodology provides stability to the buffers, it does introduce ‘Action Latency’ where the requisite supply action is delayed and may be too late in a high acceleration or a high deceleration demand environment. For example, if the RLT is 3 days for a retailer with weekly cyclicity, we need to wait for 3-6 days before revising the buffers! Some companies use ‘Cumulative Red zone penetration’ methodology, which results in a somewhat lower amount of latency.
Since fresh demand data for most companies is available daily, the granular demand predictions can also be refreshed daily using ML techniques. Why not use this information to trigger buffer resizing?
Actual implementations which have used this methodology have seen high levels of availability even during sharp demand fluctuations, with a much faster response to market demand shifts.