Overview
Rate of Sale (ROS) calculations are a key component of Assortment Planning (AP) demand forecasting. ROS represents the Average Sales per Week per Store within a cluster, calculated and maintained at the cluster level.
Toolio automates ROS calculations via the ROS Model while allowing users to override these values as needed. This article covers:
How the ROS Model drives ROS.
How Toolio handles anomalies like Markdowns and Out of Stocks in ROS calculations.
How users can influence ROS calculations by ranking their choices.
How users can manually adjust ROS.
Rate of Sale (ROS) Model
Rate of Sale Model is used calculate rate of Gross Sales Units per week per store, which is then used in forecasting. You can define the time frame and filters (choices) that will be used so the system makes the calculation based on these defined criterion.
Please note that there are 2 distinct Choice attributes which correspond to ROS:
Calculated ROS
Type: System- Generated
This value is calculated at the Offering Level per cluster. It represents the average sales per location, per week, based on historical data from one or more choices. While a single average number is displayed in the "Calculated ROS" attribute, there may be a range of performances across the products and locations used in the ROS Model.
You can click into the calculated ROS cell to view the calculated ROS distribution and the rankings at the cluster level.
Adjusted ROS
Type: User input
User can adjust the Calculated ROS as they like. In the presence of a Calculated ROS, system uses the weighted distribution among the clusters and adjust the ROS according to the user input among the clusters. If no ROS Model was defined, then, each offering gets the same defined value for the defined Adjusted ROS.
You can read more about the full Choice Planning workflow + details around how the ROS values are utilized in conjunction with Sales Curves, Location Curves, and Cluster Groups in the full Choice Planning guide here.
Good Weeks in ROS calculations
Toolio automatically filters out 'Bad Weeks' (weeks with insufficient inventory) during ROS calculations. This ensures only 'Good Weeks' contribute to the calculation.
Good Week Definition: Any week where inventory was available to sell (BOP Units + Receipt Units + Transfer Units > 0).
Please note that this evaluation takes place at the Location x Week x Choice granularity during ROS Calculation
Details of Good Weeks in ROS
Details of Good Weeks in ROS
For example, let's say your ROS Calculation is configured to include 10 choices and utilizes a 52 week historical time frame. The system will evaluate and exclude any Bad Weeks at the Location x Week x Choice level during ROS calculation. Let's say for a given choice and location combination, 5 weeks in the 52 week time frame were deemed as Bad Weeks, then those weeks will be excluded from the ROS calculation entirely - meaning no sales will be considered from those weeks in the numerator, and the total Good Weeks for that 52 week period is reduced to only 47 in the denominator.
Thus, the ROS calculation for that choice x location would be:
(sum of GSU from the good weeks)/(number of good weeks)
in this case, (sum of GSU from the 47 good weeks)/47
In a case where Good Weeks are not enabled, the calculation would instead be:
(sum of GSU from all weeks)/(total number of weeks)
in this case, (sum of GSU from all 52 weeks)/52
⚙️ This behavior is controlled by feature flag: useGoodWeekToCalculateProductivities. If you would like this turned on, please reach out to your Customer Success Manager.
Please note that this behavior is universally configured - meaning that Good Weeks are always considered when enabled, or never considered when disabled.
ROS Distribution
To provide insights into the range of ROS calculations, Toolio presents the ROS distribution across six rankings by cluster:
Top: Highest-performing ROS within the cluster.
High: 75th percentile of calculated ROS.
Median: Middle value of calculated ROS.
Low: 25th percentile of calculated ROS.
Bottom: Lowest-performing ROS within the cluster.
Average: Average ROS across all results, aligning with the prior version of ROS calculations.
Users can view this distribution by double-clicking the "Calculated ROS" cell. A pop-out displays the rankings and weighted ROS, which aligns with the Calculated ROS shown on the line-building screen. By default, the Average is selected to match previous calculation methods.
The pop-out view presents the ROS across the rankings by cluster as well as a weighted ROS which will align with the Calculated ROS number on the line building screen above. Average is selected by default to align with the previous generation of ROS calculation.
Ranking Choices
Users can leverage the ROS range to rank products or maintain the default average ROS. Ranking allows users to assign a relative performance expectation to any new choice.
To rank products:
Add the "ROS Rank" attribute.
Select the rank (e.g., Top, High, Median) based on expected performance.
The Calculated ROS will align with the selected ranking.
For more information on using ROS Ranking, watch the video below.
To learn more about the use of Rate of Sale Ranking, please see the video below