Localized Replenishment–Step 7: What NOT to do: Safety Stock

Bookmark and Share

THE PROFIT LAB // 10 Steps to Finding Profit in Localized Replenishment

7. PLEASE MIND THE GAP //

Step #7: What NOT to do (Part 2) – Safety Stock

While ideally replenishment is driven by the perfect forecast, as we discussed earlier in the series there is no such thing as the perfect forecast. This is particularly true in localized replenishment where we must execute at very low levels of detail. Among the most difficult products to forecast are those that have high amounts of variability or volatility. These conditions are inconsistent and challenge even the most sophisticated forecasting methods.

Safety stocksUnderstanding this situation, replenishment systems have endeavored to compensate for it using something commonly referred to as safety stock. This is basically an acknowledgement that there are gaps between the forecast and actual sales. Safety stock is commonly calculated based on some measure of demand volatility, the implicit assumption being that volatile demand implies difficult-to-measure demand, which in turn implies high forecast error. The problem is that demand volatility is itself constantly in flux, making it difficult to reliably calculate the ideal safety stock.

What can you do now?

If you have no measure of volatility in your current process, you can start to make progress by looking at rate-of-sale. Larger locations and/or products that have a higher rate-of-sale tend to be easier to forecast and therefore require lower relative levels of safety stock. As an example, suppose you are currently holding 4 weeks of supply in all locations for a given product. You can likely do better holding only 3 weeks of supply in the higher-volume locations and 5 weeks in the lower-volume locations given that the higher-volume locations are likely to have more consistent week-to-week sales.

If you already have a safety stock measure, you then need to look for ways to make it more accurate. Reanalyzing for different seasonal periods or timeframes in product life can begin to accommodate this. Any opportunity to calculate safety stock at a more detailed level can help as well. Different location levels or lower level of product (rather than using averages) can enable you to get better results. The effort can be tedious, but – especially for high margin products – it can be quite fruitful.

What should you consider in the future?

Some solution providers have invested the efforts of their science teams to generate processes such as “safety stock optimization” to try to reduce the gap left by volatile and variable behavior. While better than manual processes, these are generally a signal that traditional approaches are still the foundation of the replenishment logic and that newer thinking has yet to be applied to the solution space being covered.

More forward thinking approaches use processing capabilities of today’s technology in addition to new scientific approaches which understand the facts of what different recommendations resulted in to measure their productivity rather than trying to estimate forecast error as a foundation. Better systems measure historical facts against the goals that we’ve been discussing throughout this series. If an item is a money item, did the system recommended position result in achieving the goal of highest possible profitability? If not, what level would have? How, when, and why is that? The best systems ask and answer these questions, learning from the results and constantly improving the answers and subsequent performance.

 Learn more about systems that constantly improve and learn here >>

Write a Comment