A price test must do more than detect conversion change.
A price test must do more than detect conversion change. It estimates causal demand effect, translates results into contribution and controls potential loss during the experiment. Price range, randomisation unit, group leakage, competition and price memory determine validity. Use an amplitude sufficient to learn while limiting customer and profit exposure while elasticity is uncertain.
Decision and method.
Test candidate price on a limited traffic fraction when contribution remains near the current threshold in adverse elasticity and the test can be isolated. Choose reference, candidates, marginal cost, discounts, taxes and service costs; write plausible elasticities and retention effects. Randomise at a level that prevents treatment switching and impossible comparisons; exclude or measure contamination. Simulate volume and contribution for every price and elasticity; set maximum exposure, size, duration and stop rule before launch. Deploy only when incremental contribution and guardrails cross thresholds; document the actual tested range.
Worked example.
Current price €100, variable cost €55, volume 800, contribution €36,000. Candidate is €105; plausible isoelastic elasticities −0.8, −1.4, −2.2. Simulated contributions at €105 are €38,469, €37,359, €35,929. Worst scenario loses €71 over the full period. On a 10% traffic pilot, corresponding exposure is about €7 before fixed test cost. €105 merits a pilot because potential is positive and direct risk limited, but deployment needs an interval compatible with contribution at least equal to reference plus retention guardrails.
Checks and limits.
Price treatment remains stable for randomised unit; contribution includes variable cost, returns, service and retention; maximum exposure is calculated before launch; deployment rule uses interval and guardrails. Experiment can alter long-term price expectations. Isoelastic demand is a local assumption, not law. Competitor reactions rarely appear in a small pilot.
Resources and sources.
Download price-test risk map. Related: size a price test, estimate elasticity, simulate price, volume and contribution. Sources: Danaher (2002); Fisher, Gallino and Li (2018); Wang et al. (2025).