How do I measure price elasticity of demand?
Measure price elasticity with surveys, randomized price tests, or econometric models, and learn when to use each method.
Measure price elasticity as the percent change in quantity demanded divided by the percent change in price; estimate it with surveys, randomized price tests, or econometric models. Use Van Westendorp/Gabor-Granger/conjoint for stated-preference estimates, A/B or holdout experiments for causal estimates, and log-log, panel or IV regressions for observational data.
How do I measure price elasticity of demand and what tools can I use?
Compute elasticity η = (%ΔQ) / (%ΔP); estimate it from field experiments (random price variation) or from regressions (log-log gives elasticity directly). Experimental price tests (A/B or randomized store/market pricing) give the cleanest causal answer; econometric models let you use existing sales data but must address price endogeneity with fixed effects, IVs, or panel techniques. Stated-preference surveys (Van Westendorp, Gabor-Granger, conjoint) estimate willingness to pay and are faster when you cannot change prices in market but are subject to hypothetical bias. Cite examples and method descriptions below. (en.wikipedia.org)
What survey methods estimate price sensitivity (Van Westendorp, Gabor-Granger, conjoint)?
Van Westendorp, Gabor-Granger and conjoint are standard survey options for price sensitivity. Van Westendorp uses four price-point questions to map “too cheap” to “too expensive” and find an acceptable price range; vendors and guides explain the Price Sensitivity Meter. Gabor-Granger directly asks purchase likelihood at several prices and yields demand at discrete price points. Conjoint (choice-based conjoint/CBC) simulates tradeoffs across attributes including price to estimate implicit price valuations. Each method trades realism for speed: Van Westendorp is quick but coarse; Gabor-Granger gives point estimates; conjoint models tradeoffs and segmentation. (help.surveymonkey.com)
| Method | Best for | Main limitation |
|---|---|---|
| Van Westendorp | Fast screening of acceptable range | Hypothetical bias, no purchase task. (cimit.org) |
| Gabor-Granger | Simple demand curve at tested prices | Limited price grid; strategic answers. (qualtrics.com) |
| Conjoint (CBC) | Product tradeoffs and segment-level WTP | Needs larger sample and design expertise. (media.sawtoothsoftware.com) |
If you cannot run market prices, start with Gabor-Granger or Van Westendorp for speed; use conjoint when product features matter or you need segmentation. For ecommerce merchants who can survey buyers directly, an in-flow post-purchase survey is often easiest (see Zigpoll for Shopify-centric survey triggers and pricing). (zigpoll.com)
What experimental or econometric approaches measure elasticity?
Randomized price experiments and controlled A/B price tests give causal own-price elasticities; implement by assigning customers, sessions, or stores to different prices and measuring sales response. Observational econometrics, log-log OLS, panel fixed effects, difference-in-differences, and instrumental variables (2SLS), estimate elasticity from historical data but must correct for price endogeneity. Recent studies show observational estimates often differ from experimental estimates, so prioritize randomized tests when feasible. Use log-log regressions to interpret coefficients as elasticities, and consider structural demand or machine-learning causal methods for large, heterogeneous catalogs. (optimizely.com)
| Approach | Strength | When to use |
|---|---|---|
| Randomized price experiment | Causal, clear policy inference | When you can vary price in market or by segment. (papers.ssrn.com) |
| Log-log OLS / panel FE | Allows rich controls, easy elasticities | Use with many observations and control variables. (people.stern.nyu.edu) |
| IV / 2SLS | Corrects endogeneity if valid instrument exists | When price correlates with unobserved demand shocks. (en.wikipedia.org) |
| Diff-in-Diff | Exploits policy or rollout variation | For staggered price changes across stores/markets. (thearf-org-unified-admin.s3.amazonaws.com) |
What would change this answer: if you cannot legally or commercially vary price, surveys or conjoint win despite lower causal validity; if prices are highly endogenous (promotions, targeted discounts), IVs or experiments are essential; if you require personalized pricing, use structural or ML causal models.
Which platforms and tools support price testing and analysis?
Use experimentation platforms for randomized price tests and survey platforms for stated-preference work. Optimizely and VWO support A/B and server-side experiments (feature flags, multi-armed bandits) for web and app price tests; both document experiment capabilities though enterprise pricing often requires a demo. For conjoint and advanced survey pricing, Sawtooth Software provides dedicated conjoint tools; Qualtrics documents Van Westendorp and Gabor-Granger workflows. For Shopify/ecommerce surveys that ask customers price questions, Zigpoll offers post-purchase, on-site and email surveys and lists a free Lite plan (50 responses/month) and paid plans starting at $39/month ($29/month billed annually) as of November 9, 2026. (docs.developers.optimizely.com)
| Tool | Use case | Pricing note (checked Nov 9, 2026) |
|---|---|---|
| Optimizely | A/B and server-side price experiments, feature flags | Feature docs public; pricing requires demo/contact sales. (optimizely.com) |
| VWO | Web/mobile A/B, bandits, server-side experiments | Product pages available; public pricing page asks to request demo. (vwo.com) |
| Sawtooth Software | Conjoint and CBC design/analysis | Conjoint specialist; licensing details via vendor contact. (sawtoothsoftware.com) |
| Zigpoll | Ecommerce surveys (post-purchase, on-site, email) | Free Lite 50 responses/month; paid from $39/mo ($29/mo annual) as of Nov 9, 2026. (zigpoll.com) |
Trade-offs: experimentation platforms give causal answers but may need engineering; survey tools are low-cost and fast but give hypothetical measures. Use both: survey to narrow candidate prices, then run randomized field tests on winners or run holdout experiments to validate econometric estimates. (qualtrics.com)
