Definition
Category-level margin change attributable to cross-item demand substitution caused by AI-driven item-level price changes, isolating the margin transfer from one item to another within the same category.
The Measurement Gap
AI pricing engines optimize at the item level, adjusting prices to maximize margin or revenue for individual products. But customers substitute between items within a category; a price reduction on Item A steers demand away from Items B and C, which may carry higher margins. GMROI, the standard category performance metric, shows the net result of all pricing activity but cannot decompose how much of a category's margin change came from AI-driven substitution effects versus organic demand shifts. Gross margin rate tracks margin as a percentage of revenue but treats all sales within a category as interchangeable, masking whether high-margin items lost volume to price-promoted low-margin substitutes. Sales per square foot reflects total productivity but is blind to the internal margin composition changes caused by item-level price optimization. Markdown rate measures the prevalence of discounting but not whether markdowns on one item destroyed the margin of its substitutes. The specific failure mode is a pricing engine that reports positive results at the item level while destroying value at the category level through substitution effects that no existing metric measures.
Decision Supported:
Is the pricing algorithm's item-level optimization destroying category-level margin by steering customers from higher-margin items to lower-margin substitutes?
Formula
Plain language: When the AI changes the price on Item A, measure the change in sales volume and margin on Item A's substitutes (Items B, C, D in the same category). If Item A's price drop increased its margin contribution but decreased the margin of substitutes by a greater amount, the net category effect is negative.
Notation:
CICI(c) = [Delta_Margin_target(c) + Delta_Margin_substitutes(c)] / |Delta_Margin_target(c)|
Variables:
| Variable | Definition | Unit |
|---|---|---|
| CICI(c) | Cross-Item Cannibalization Index for category c | Signed ratio |
| Delta_Margin_target(c) | Margin change on the item(s) whose price was changed by the AI in category c | $ |
| Delta_Margin_substitutes(c) | Margin change on identified substitutes in category c during the same period | $ |
| c | Merchandising category or subcategory | Category |
Counterfactual requirement: Delta_Margin values must be computed against a control baseline, not raw pre/post comparison. Use one of: (a) difference-in-differences comparing price-changed items versus a matched control group of items in the same category where no AI price change occurred, or (b) cross-elasticity coefficients from the pricing engine's demand model to isolate substitution-driven margin transfer from organic demand shifts. Raw pre/post margin comparison without a control group will attribute seasonal and promotional effects to cannibalization, producing unreliable CICI values.
Proposed formula, pending validation pilots.