How to Achieve 100% Customer Service at Close to Minimum Inventories

by Klaus Spicher: Contact:? [email protected]????

Introduction

Customer Service represents the key driver for planning operations. So, the common process is, market analysis (understanding the customer needs), design of Marketing Activities, risk analysis and forecasting demand, which represents the key input for planning operations. Forecast quality determines the efficiency of planning down the road along production to material supply. Inventories are the shock-absorber for maintaining a high service level. FVA represents a useful tool for improving forecasts.

Outbound Forecasting Approach

From the commercial point of view inventories should be as low as possible – in line with the service strategy - while salespeople insist on readiness of deliveries whatever the customers ask for. Reducing inventories is based on improving forecast quality. But improving forecasts is a never-ending story with limited results due to growing (even political) uncertainties, volatility of demand and changes in customer/consumer behavior, etc. Therefore, it’s time for new ideas.

NEW: Inbound Supply Approach

It will be shown that in spite of significant inventory reduction, 100% customer service will be achievable. In most cases Safety Stock (SS) is not needed but can be added and high forecast quality is not needed as well. This result is possible by identifying the right inbound “Supply Function” (SF). The SF specifies the timing and quantities to be delivered into the warehouse. The method works even if the original data include significant outliers <see Example below>. For extreme peak-values different tuning options of SF are given for ensuring 100% service. Tuning tools are Safety Stock, Number of planned inbound deliveries per year and a SF-Scaling-Option (T1) and the number of deliveries per year. The precise peak-values requiring additional tuning cannot be specified at the moment <further research necessary>. – So, the SF represents the key input for planning operations - and in some sense – replaces the forecast of the Outbound Forecast Approach.

The example given here is based on monthly data. Two years of data are required for in-sample analysis. – One-year history and one-year of forecasting and actual data for demonstrating the quality of results.

For demonstration, the method will be applied for 2 different forecast-scenarios. FC 1 is nearly perfect while FC 2 represents a modified Na?ve Forecast based on truncated history. It does not matter, which forecast ‘method’ will be applied.

The approach of the analysis applied is trivial – running the traditional stock calculation on a monthly basis [Opening Stock (OS) - (Forecast / Actuals) = Final Stock (FS)] along the FC-Year. The result is measured as “Average Annual Stock” (AAS). So, for step 1, the Supply Function [SF(FC)] for the forecasts has to be determined. Step 2 applies the procedure on Actuals, getting SF(Actuals).

In spite of the volatility / peaks of T/S 242 no manual “tuning” for extreme outliers is necessary. That means: SS = 0, SF-Scaling-Option = T1 =1 and Customer service equals 100%. This can be seen from the Exhibits. 3 & 4 (above) because the Opening Stock (OS) <red line> exceeds the forecast (dotted lines) along the year for both forecasts.

The selected Supply Functions for both forecasts are given in Table 2.

In case of FC = Actual the Supply Functions “produce” customer Service = 100%; i.e. no Stock-Out. The AAS = Average Annual Stock with 7 inbound deliveries shows AAS (FC 1) = 7,05% of total annual forecasts, while for AAS (FC 2) = 6,00% with 9 inbound deliveries. At the first glance the results are surprising because of the huge difference in forecast quality (see MAPE’s in Exhibits 3 and 4). The resulting inventories for both forecasts are rather equal.

Takeaway

For any FC of a T/S a set of Supply Functions exist, enabling 100% customer Service at low levels of inventories, in case AAS-values of 6% to 7% for this example are regarded as “low”.

Step 2 will reveal what will happen replacing the FC-values by Actuals. Intuitively, it might be assumed that, the better forecast (FC 1) will result in lower inventories. Right?

Now the actual values of the FC-Year will be used for the same analysis as before with the two different FC-Functions. The values of FC 1 are stepwise replaced by the Actuals - resulting in no stock-outs along the year (again Service = 100%). The same result shows up applying the procedure with the values of FC 2.

The graphical displays will be omitted, because they are similar to Exhibits 3 and 4 – just replacing the forecast values (dotted lines) by the Actuals.

The resulting Supply Functions are identical and given in Table 3.

The result of equal Supply Functions is no surprise, because in both cases the same Actuals are applied. The AAS (Actuals) = 8,24%. (Tuning brings the AAS down to 5,42%.)

How to calculate the Supply Function SF:

Summary of Results

(1)? The objective fulfilled – Inventory management w/o depending on FC-Quality.

(2)? The Inbound Supply Approach allows estimating the inventory level close to minimum – but still providing 100% Service.

(3)? Practically the Inventory AAS-Values can be seen as a (theoretical) benchmarks.

(4)? Inventory AAS-Values are independent of the forecast. So, improving the forecast is not relevant. Methods like FVA just support the Outbound Forecasting Approach and are for the Inbound Supply Approach irrelevant.

(5)? The focus of Planning Operations moves to Supply.

(6)? The Tuning-Methods for reducing Inventory AAS have to be analyzed in depth. In general, further analytical research is needed.

?

I welcome any comments, especially your job experiences and constructive critical comments! You can reach me here at [email protected]

????????????????????????????????????????????????????????????????