How to Calibrate Price Elasticity Models Using Competitor Reaction Patterns

Most pricing teams estimate price elasticity using their own historical sales data. But in competitive markets, your price changes don't happen in a vacuum—they trigger competitor responses that distort the true demand signal. Ignoring this feedback loop leads to elasticity estimates that are either too aggressive (if competitors match your cuts) or too conservative (if they undercut you).

‍

Why Own-History Elasticity Fails in Competitive Markets

When you lower price by 5% and volume jumps 12%, a simple model suggests -2.4 elasticity. But if three competitors matched your cut within 48 hours, the volume lift came from market expansion, not just your price advantage. Conversely, if you raise price and competitors hold, you lose disproportionate share—making demand look more elastic than it truly is. Both scenarios bias your model.

‍

Build a Competitor-Adjusted Elasticity Framework

• Tag every price change with competitor response data. For each SKU-week, record: your price delta, the number of competitors who matched, countered, or ignored within a 7-day window.
• Segment elasticity by response type. Calculate separate elasticity coefficients for: (a) no competitor reaction, (b) full match, (c) partial match, (d) undercut. You'll typically find elasticity is 30-50% lower when competitors match.
• Weight observations by competitive intensity. In categories with 5+ active repricers, downweight solo-move observations—they're rare and unrepresentative.

‍

Operationalize with a Rolling Calibration Window

Don't treat elasticity as static. Re-estimate quarterly using a 12-month rolling window, but stratify by competitive regime. Use clustering (k-means on competitor match rate, price volatility, assortment overlap) to define 3-4 regimes per category. Each regime gets its own elasticity curve.

‍

Validate Against Holdout Periods

Reserve the most recent 8 weeks as a test set. Apply your regime-specific elasticity to predict volume for actual price changes during that period. Compare MAPE against a single-coefficient model. Teams using this approach typically see 15-25% forecast error reduction.

‍

Feed Results Back into Repricing Rules

Once you have regime-specific elasticity, embed it in your repricing logic: when competitor match probability exceeds 70%, use the lower elasticity coefficient to avoid over-discounting. When match probability is below 30%, use the higher coefficient to capture share aggressively.

‍

The result: pricing decisions that account for the competitive game theory, not just your own demand curve. That's how you stop leaving margin on the table—or losing share to reactions you didn't model.