Promotional Dependency Erosion Rate (PDER)

GMI-RET-021Retail OperationsIntegration-Required

Definition

Rate at which AI promotion targeting erodes customers' willingness to buy at full price, measuring whether the targeting engine is training its best customers to wait for discounts.

The Measurement Gap

AI marketing systems learn which customers respond to promotions and target them with increasing frequency. The targeting works: promoted customers buy more and ROAS looks strong. What ROAS does not capture is the behavioral shift underneath. Customers who receive promotions frequently begin to expect them. They learn to wait for deals. Their full-price purchase rate declines. This is promotional dependency erosion: the AI's targeting creates the price sensitivity it then exploits. The marketing team sees rising promotional conversion and increases budget. The merchandising team sees declining full-price sell-through and assumes the products are the problem. Neither connects the two observations because no metric links promotion exposure to full-price purchasing behavior over time. PDER bridges this gap by tracking full-price purchase behavior as a function of promotion exposure, revealing whether short-term revenue comes at the expense of long-term margin.

Decision Supported:

Is AI promotional targeting creating customers who will not buy at full price, and is the promotion strategy generating short-term revenue at the expense of long-term margin?

Formula

Plain language: Segment customers by AI-promotion exposure level (high, medium, low based on offers received in the past 90 days). For each segment, measure the percentage of purchases made at full price. Compare across segments and track over time. If full-price purchase rate declines as promotion exposure increases, the AI is training customers to wait for deals.

Notation:

PDER = Delta_FullPrice_rate(high_exposure) - Delta_FullPrice_rate(low_exposure)

Variables:

Variable Definition Unit
PDER Promotional Dependency Erosion Rate Percentage points
Delta_FullPrice_rate(segment) Change in full-price purchase percentage for the segment between current and previous measurement period pp
high_exposure Customers receiving above the 75th percentile of AI-selected promotions Segment
low_exposure Customers receiving below the 25th percentile of AI-selected promotions Segment
Full-price purchase Transaction with no promotion, coupon, or discount applied Binary flag

Causal isolation requirement: Self-selection bias is the primary methodological threat. Customers who receive more promotions may already be more price-sensitive, which confounds the analysis. PDER is only valid when computed using one of: (a) a randomized holdout group where a subset of promotion-eligible customers is excluded from AI targeting for the measurement period, or (b) propensity score matching that controls for prior purchase behavior, price sensitivity signals, and demographic covariates. Without one of these controls, PDER measures correlation between promotion exposure and full-price behavior, not causation. Observational PDER can be reported as a diagnostic signal, but promotion strategy decisions require the experimental or matched version.

Proposed formula, pending validation pilots.

At a Glance

Unit
Rate
Trend Direction
Lower
Reporting Frequency
Quarterly
Category
Workforce and Marketing AI Economics
← Multi-AI Coherence Score (MACS)Task Queue Achievability Index (TQAI) →

Assess Your Organization

The AI-GPS Assessment evaluates your AI governance maturity and recommends KPIs like this one for your implementation roadmap.

Take the Free AI-GPS Assessment