Voids are situations where a retailer has agreed to sell your product, but the product isn't actually on the shelf at the store. Identifying voids on your own typically requires a time-consuming and complex analysis. Crisp's Voids dashboards use a sophisticated machine learning model to identify possible voids early, so you can quickly take action to address issues and keep products on the shelves. Both Retailer and Distributor Voids dashboards are divided into three sections:

  • Overview: See how at risk voids trend over time and understand recent void levels and changes.
  • Drivers: Understand which regions, retailers/distributors, and chains are driving your voids.
  • Details: Dig in to granular information that helps you understand typical sales and which stores are identified as at risk.

Overview

The Overview section allows you to see overall void trends and totals. Use the following map of the screen to guide you.
voids_dashboard_001.png

Screen Shot 2022-01-04 at 2.16.26 PM.png At Risk Voids Over Time

This line graph displays your at risk voids over time, helping you recognize patterns or spot potential issues.

Screen Shot 2022-01-04 at 2.17.17 PM.png Overview tiles

  • Last Updated: The date this dashboard was last updated (updated weekly on Mondays).
  • Latest Week at Risk Voids: The count of product/store combinations that Crisp’s void algorithm has identified as potential voids.

Screen Shot 2022-01-04 at 2.18.19 PM.png Voids by Product

This table displays each product's void count. The void count is the count of product/store combinations (PODs) that Crisp’s void algorithm has identified as potential voids for the most recent complete week. You can use this visualization to understand which products have the greatest or least impact on total void counts.

Drivers
voids_dashboard_002.png

Screen Shot 2022-01-04 at 2.16.26 PM.png Voids by State (heat map)

This heat map shows void counts by state, to help you understand where voids are high and low. The void count is the count of product/store combinations (PODs) that Crisp’s void algorithm has identified for the most recent complete week. You can hover over a state to see void count information or select a state to filter the dashboard by that state.

Screen Shot 2022-01-04 at 2.17.17 PM.png Voids by additional metric (tables)

The tables following the heat map break down voids by geographic regions, retailers/distributors, and chains to help you gain additional insights into where voids are occuring. 

 

Voids Details

The Details table at the bottom of the dashboard provides store-level details including typical weeks between sales/shipments and weeks since the last sale. This can help you pinpoint specific stores where there may be void and more deeply evaluate if it has been an atypically long time since a store last had an order/sale.
Once you've applied filters to identify a list of stores with potential voids that you want to take action on or share with others, you can export the stores in this table by selecting the Tile actions menu at the top-right corner of the table, then selecting Download data.
reducing_voids.png

Void model details

Crisp's voids model works by using machine learning to predict voids using a mix of short-term and historical data, store context, and other variables. The technique our proprietary model uses to predict voids is called a gradient-boosted decision tree, which takes into account many variables that could contribute to a void to and decides the likelihood of a void at each point of distribution (POD). Likely voids then appear in our harmonized Voids dashboards.

While a human analyst might just look at a single variable, such as it having been a longer than typical amount of time since the last shipment/sale at a POD, the Crisp model takes into account many more variables and uniquely analyzes each POD for greater precision. Given the variety of inputs to the model, you may not always be able to look at voids or sales dashboards and pinpoint the exact reason the model identified a void. However, Crisp's model has been tested with real-world data and was very successful at identifying genuine voids. For more about how our voids model works, check out the Crisp blog post on voids