Welcome to grokstats.com, a home for human-understandable stats. Grokstats mixes dependable analytics with what business decision makers know for sure, to produce practical profit capture.

For example … extracting knowledge from historical data.

Intuitively, it makes sense that not all products (Stock Keeping Units (SKUs)) in a grocery aisle perform identically. Looking at category sales, the logical possibilities are that a SKU can:

1. Increase category sales
2. Do nothing to category sales
3. Decrease category sales (or cannibalize demand)

Using a simple model to analyze category sales history, when each product is in the aisle, shows a well-behaved pattern of individual category-contribution by each SKU.

At the left of the diagram are the large category contributing Products/SKUs, and at the right are cannibal Products/SKUs. Category contribution is measured with multiple linear regression on historical sales data by store. Where did this analysis come from? The business decision maker made a wish.

“I wish I knew the weekly contribution of each product/SKU in store.”

Business decision maker wishes are common-sense models. And this simple model made a 30% ROS improvement possible by dropping Products/SKUs that decrease category sales.

And, common-sense allowed the analysis to go a little further. Taking costs into account. For example, the costs of shipping, carrying inventory, forecasting each Product’s/SKU’s sales, etc. can be taken together to establish a “hurdle rate” to compute whether a SKU is above, at, or below the break-even category sales contribution.

Dropping below break/even SKUs further boosted return on sales available to this client, from 30% to 40%. If you are in retail for profit, knowing the answers to business decision maker wishes can point out where you are wasting your efforts. Giving you a platform to stop fighting sales quotas, and start growing profitability.

Part 2 of this blog post is available at Applying Human-Understandable Stats