Lost Sales ML



Laboratory - AI-based reports module on Datawiz BI service

Lost Sales ML - using machine learning algorithms, the report analyzes the movement of products, models sales forecasts and summarizes data on sales losses due to missing or low sales.

The functionality of the report allows:

  • to identify lost sales by products, categories, stores, category managers;
  • to determine the amount of losses, including due to zero balances;
  • to track the dynamics of sales and balances on problem products;
  • to identify the causes of losses, problem areas at work and avoid them in the future.

Based on the past sales of products for each store per year, the neural network predicts their future sales. Next, there is a comparison between the actual data and the predicted ones. If sales were less than predicted, then such sales are considered lost. Then there is already a generalization of data on losses in turnover due to missing or low sales.

All metrics are divided into the following subgroups:


Assortment activity


Note! Access to viewing the metrics, according to the user's role, is determined by the administrator. Contact your administrator to expand the access.

Metrics highlighted in purple or red are clickable. Clicking on them opens a window with a flow chart of the chosen metric. 

Visualization of the clickable metric "Loss of sales QTY ML" for the lowest level of the table - products, displays a combined graph containing the dynamics of indicators: Predicted sales (and limits of acceptable sales), Sales Qty, Stock Qty at the beginning of the day. This allows to visually track the lost days. Sales are considered lost if Sales < Predicted sales.

Similar visualization on indicators "Lost of profit ML", "Lost of sales ML".

Note! The visualization of the clickable metric "Qty of losing sales" contains curves of two metrics: Sales Qty and Stock Qty at the beginning of the day. This allows to see losses due to zero or minimal stocks.

The selection of filters makes it possible to carry out the analysis within the specified conditions.

There are the following filters:

  • Stores
  • Period
  • Previous Period
  • Category Managers
  • Assortment Types
  • Categories
  • Brands
  • Level
  • Promotion of products