The monitoring step detects the product locations that are currently out-of-shelf. Therefore, this steps checks whether the current waiting time since the last sales transaction exceeds the maximum tolerable waiting time. This step applies the estimated model to the last sales, as determined in the estimation step. This results in the critical time (also called maximum tolerable waiting time) since the last sale. The calculation is based on:
The last sale
The estimation model per product location
Defined confidence threshold for monitoring
If the time between the last sales transaction and the detection time exceeds the critical time, the relevant product location is probably out-of-shelf, and an alert will be created.
The computed maximum tolerable waiting time (tL, max) is determined independently of the sales behavior of the product location. That means that the monitoring step can determine the maximum tolerable waiting time for all selling classes.
The monitoring step also calculates the probability of being out-of-shelf of a certain product in a location, during a certain transaction interval. The monitoring step considers regular intra-week and intra-day sales variations. Furthermore, the monitoring step can model the impact of price, discount, and promotions on sales intervals.
The monitoring step can consider listing information on product locations. This means that you can limit the items considered, for example to those that are currently listed, or to those that are listed and fulfill some other criteria. This is possible using the /OSA/LISTED_PRODUCT_DAO
(Retrieve Listed Product Data
) Business Add-In (BAdI). For more information, see .
Last sales transaction: Time of the last sale of a product in the location, in table /POSDW/TLOGF
.
Detection time: The start time of the current dispatcher run in the monitoring step.
This detection time will be compared with the time of the last sales transaction for the relevant product location, to determine the time interval since the last sale.
Estimation results: Parameter values of the estimation model that has been determined for the relevant product location, in the estimation processing step of the OSA calculation.
Monitoring is performed only for product locations where the following conditions are fulfilled:
In the estimation model for the product location, a quality of fit has been determined that is below the defined confidence threshold for monitoring.
In the previous monitoring run for the product location, a critical time was determined that has not been exceeded until now. This is done due to performance reasons.
You have defined the /OSA/MON_MIN_QOF
(Minimum Quality of fit (QOF)
) and the /OSA/MON_THRESHOLD
(Monitor Threshold
) parameter for monitoring in Customizing, under .
The quality of fit indicates how well the estimation model maps the sales transaction history. The /OSA/MON_MIN_QOF
parameter is a threshold value for the minimum quality of fit that is required to perform the monitoring step for a product.
The confidence level threshold indicates the sensitivity of an out-of-shelf alert. Based on the confidence level threshold parameters, the critical time is determined. With this monitoring threshold, you can configure the ratio between false alerts and true alerts depending on the intended purpose:
A high confidence level rather leads to a low false alert rate
A low confidence level rather leads to an early detection of an out-of-shelf situation
Note
By using different Customizing profiles, it is possible to define different standard confidence thresholds. Finally, store-specific or sub-department-specific standard confidence thresholds are possible. In addition, for each product location, you can define a specific confidence threshold in the table /OSA/PS_CONF
. If a specific confidence threshold is defined in this table for a product location, it will override the standard confidence threshold. You can maintain this table using the OSA/PRODUCT_STORE_CONF
(Retrieves Product Store Confidence Level
) BAdI.
You can run the monitoring processing as often as necessary. This can be several times a day during the opening hours of the location, and as often as the out-of-shelf alerts can be checked by the store personnel.
The resulting output data of the monitoring step is stored in the /OSA/MONITORING
table.
After this step, this table contains the following additional information:
Critical time: The maximum tolerable waiting time after which the product location should become critical, according to the defined confidence threshold for the monitoring step.
A sales transaction for the product location is expected at the end of the maximum tolerable waiting time, at the latest. After the maximum tolerable waiting time, the monitoring step needs to be re-run to determine the out-of-shelf probability.
Probability: The probability of a certain product in a location being out-of-shelf.
Lost transactions and lost revenue (lost turnover): During monitoring, the OSA algorithm has determined how much could be sold. Based on that, the OSA algorithm has calculated how much revenue has been lost due to the detected out-of-shelf situation.