Price elasticity of demand measures how strongly the quantity buyers purchase responds to a change in price. It is the percentage change in quantity demanded divided by the percentage change in price. A figure of -2 means a 1% price rise cuts sales by about 2%; a figure of -0.5 means sales fall by only half a percent. The measure exists because a price change has two opposing effects on revenue, more money per unit and fewer units sold, and elasticity says which one wins. Retailers use it to set prices, modellers to separate price effects from advertising, and ad exchanges to set reserve prices.

How the number is calculated

The formula is simple: elasticity equals the percentage change in quantity divided by the percentage change in price. The result is normally negative, and practitioners often quote its absolute value.

Take a hypothetical shop selling 1,000 units a month at $20. It raises the price to $22, a 10% increase. If sales fall to 850 units, a 15% drop, the elasticity is -1.5. Revenue falls from $20,000 to $18,700. If sales fall only to 950 units, a 5% drop, the elasticity is -0.5 and revenue rises to $20,900.

Those two outcomes define the categories. Demand is elastic when the absolute value is above 1: quantity moves proportionally more than price, so a price rise lowers revenue. It is inelastic below 1, where a price rise lifts revenue. At exactly 1, unit elastic, revenue stays flat.

The simple formula has a quirk. Measured from $20 up to $22, the first example gives -1.5. Measured from $22 back down to $20, the same two points give a 9.1% price cut and a 17.6% volume gain, an elasticity of about -1.94. Economists avoid the asymmetry with the arc or midpoint method, which divides each change by the average of the two values, giving about -1.70 in either direction. In practice, analysts estimate elasticity from many observations, typically by regressing the logarithm of units sold on the logarithm of price, so the price coefficient reads directly as an elasticity.

From Marshall to marketing science

The concept was set out by Alfred Marshall, the Cambridge economist, in Principles of Economics, published in 1890. In Book III, chapter IV, he wrote that "the elasticity (or responsiveness) of demand in a market is great or small according as the amount demanded increases much or little for a given fall in price, and diminishes much or little for a given rise in price." He also noted that "time is required to enable a rise in the price of a commodity to exert its full influence on consumption."

Abba Lerner linked the idea to market power in 1934. His index, the gap between price and marginal cost as a share of price, equals the inverse of elasticity for a profit-maximising seller. A firm facing an elasticity of -2 would, in theory, set a price at which margin is half the price, double its marginal cost; one facing -5 could keep only a fifth of the price as margin.

Marketing science supplied the benchmark numbers. Gerard Tellis's 1988 meta-analysis of 367 estimates across more than 220 brands or markets found a mean brand price elasticity of -1.76. Tammo Bijmolt, Harald van Heerde and Rik Pieters updated the exercise in the Journal of Marketing Research in 2005 and reported a mean of -2.62. By comparison, Raj Sethuraman, Tellis and Richard Briesch found in 2011 that the mean short-term advertising elasticity across 751 estimates was 0.12, which made price roughly 20 times more powerful in absolute terms, according to their paper.

What makes demand more or less elastic

Demand is more elastic when close substitutes exist, when the purchase is a large share of a budget, when the item is discretionary, and when buyers have time to adjust. Loyalty, switching costs and necessity push the other way.

Definition matters too. Demand for petrol in general is inelastic; demand for one station's petrol, opposite a cheaper rival, is highly elastic. That is why brand-level estimates exceed category-level ones, and why demand curves differ for luxury goods and commodities.

When 40% of Americans told Adtaxi they were switching to lower-cost brands, the survey pointed to rising price sensitivity, though stated intentions are not measured elasticities.

Where marketers use it

Pricing and promotions. Price tests, which show different prices to comparable groups, are the cleanest way to estimate elasticity. During the 2025 tariff period, Walmart's chief financial officer John David Rainey said the company was measuring price elasticity on affected items while managing quantity decisions. Discounts are judged against a reference price, shaped by rules on what a "was" price may claim.

Marketing mix modelling. In marketing mix modelling (MMM), price enters as a control variable so that sales lifts caused by discounts are not credited to media. Google added support for pricing and promotion variables to Meridian, its open-source MMM, in September 2025. Omitting price inflates media return on ad spend (ROAS) when promotions coincide with campaigns.

Algorithmic pricing. Dynamic pricing systems estimate local or segment-level elasticity and price from it. A class action filed on October 2, 2026 alleges that McDonald's pricing engine flags franchisees whose customers show "MEDIUM SENSITIVITY to Price"; McDonald's says its tools only make recommendations. In ticketing, SeatGeek reported 29% more tickets sold than projected at a 13% lower average price with seat-by-seat pricing, figures that are vendor-reported.

Elasticity inside the ad market

Advertising inventory has its own demand curves. On the sell side, a price floor trades volume for price. When Yahoo moved to model-driven floors, its exchange cleared 13.86% fewer impressions while effective CPM rose 16.57% in the third quarter of 2022, leaving exchange revenue up 0.41%. Read crudely, that is a volume response of about -0.84, close to unit elastic, which explains the thin revenue gain. Index Exchange took a different route, cutting its own fee on bids just below a floor; in its Guardian pilot it reported 45% more impressions and 4% more revenue.

On the buy side, the question is how much advertisers cut spending when cost per mille (CPM, the price per thousand impressions) or cost per click rises. Google's search business is the most documented case. Former Google executive Jerry Dischler testified in 2023 that the company raised reserve prices by about 5% on average and up to 10% on some queries to meet revenue targets. In September 2025 Judge Amit Mehta found that Google used "pricing knobs" in barely perceptible increments, behaviour that only works if advertiser demand is inelastic in the short run.

New channels show the opposite pattern: ChatGPT ad CPMs fell from about $60 at launch towards $25 within months as minimum commitments fell.

Connected TV (CTV) ad tiers face elasticity on both sides: subscribers react to the monthly fee, advertisers to CPMs. Netflix raised its US ad-supported plan from $6.99 to $7.99 in January 2025, its first increase on that tier, and to $8.99 in March 2026, according to Fox Television Stations. Low churn after each rise suggests inelastic subscriber demand.

Limits and disputes

Endogeneity is the core statistical problem. Prices are not set at random: retailers raise them when demand is strong and cut them when it is weak. A naive regression of sales on price then mixes the demand curve with the seller's response, usually understating elasticity. Instrumental variables, such as cost shocks, and randomised tests are the standard remedies.

Confounding is the practical one. Discounts arrive with feature placement, email pushes and paid media, so the measured response bundles several causes. Promotions also pull sales forward, so a deep discount followed by a dip overstates the long-run response.

Short run versus long run cuts both ways. Many goods become more elastic as buyers find substitutes, and repeated discounting can train buyers to wait, a pattern visible when 57% of US households said they planned to wait for deals before the 2026 holidays.

Aggregation distorts estimates: Tellis found that data aggregated above the weekly level biased them towards zero.

The sharpest dispute is ethical. Estimating each customer's elasticity enables personalised pricing. The Federal Trade Commission (FTC) sent orders to eight firms over "surveillance pricing" in July 2024, and on August 19, 2026 proposed treating undisclosed personalised pricing as likely unfair or deceptive. Chairman Andrew Ferguson said consumers expect a price that is "not the retailer's estimate of how much they are willing to pay."

Not the same as

Pricing power is the ability to raise prices without losing much volume. Low elasticity is evidence of it, but pricing power is a property of a firm's market position; elasticity is a measured response at a given price.

Income elasticity measures how demand responds to changes in buyers' income, not price. Cross-price elasticity measures how demand for one product responds to a change in another product's price; positive values indicate substitutes, negative values complements. Cross-price effects matter in retail media, where a promotion on one brand can cannibalise another.

The demand curve is the full schedule of quantities at each price. Elasticity describes its slope in proportional terms at a particular point, and changes along it.

Recent developments

Judge Mehta's remedies ruling requires Google to disclose material search auction changes, and PPC Land's analysis on October 10, 2026 traced how a bidding change from August 17 pushed CPCs up on budget-capped campaigns. From October 12, 2026, Google will automatically pull website discounts into Search and Performance Max ads for some advertisers. The McDonald's case and the FTC's draft policy have made estimating customers' price sensitivity a legal question as well as an analytical one.

Timeline

  • 1890: Alfred Marshall defines elasticity of demand in Principles of Economics
  • 1934: Abba Lerner publishes his index linking price-cost margins to elasticity
  • 1988: Gerard Tellis's meta-analysis reports a mean brand price elasticity of -1.76
  • 2005: Bijmolt, van Heerde and Pieters report a mean of -2.62 in the Journal of Marketing Research
  • 2011: Sethuraman, Tellis and Briesch put mean short-term advertising elasticity at 0.12
  • September 2016: NBER paper uses Uber surge pricing to estimate demand elasticities and consumer surplus
  • Third quarter of 2022: Yahoo's model-driven floors cut cleared impressions 13.86% while eCPM rises 16.57%
  • September 2023: Jerry Dischler testifies that Google raised search reserve prices by about 5% on average
  • July 23, 2024: FTC issues orders to eight firms over surveillance pricing
  • January 2025: Netflix raises its US ad tier price for the first time, to $7.99
  • September 2, 2025: Judge Mehta finds Google used pricing knobs and orders auction disclosure
  • September 30, 2025: Google adds pricing and promotion variables to Meridian
  • October 23, 2025: Index Exchange announces dynamic take rates
  • March 2026: Netflix raises its US ad tier to $8.99
  • August 19, 2026: FTC votes 2-0 to propose a policy on undisclosed personalised pricing
  • October 2, 2026: Class action filed over McDonald's menu pricing engine; SeatGeek announces TourIQ
  • October 12, 2026: Google starts auto-applying website promotions to some Search and Performance Max campaigns

Summary

Who: Economists who defined and measured the concept, from Alfred Marshall to marketing scientists such as Gerard Tellis; retailers, brands and ticket sellers that set prices from it; publishers, exchanges and platforms such as Google, Index Exchange and Netflix whose ad and subscription prices depend on it; and regulators, including the FTC, examining how it is estimated for individuals.

What: The ratio of the percentage change in quantity demanded to the percentage change in price, classed as elastic above 1 in absolute value, inelastic below 1 and unit elastic at 1, and estimated through price tests, regressions and models.

When: Defined in 1890, quantified for brands in meta-analyses from 1988 onwards, and in 2025 and 2026 central to court findings on Google's search pricing, Netflix ad tier increases and the FTC's proposals on personalised pricing.

Where: In retail and e-commerce pricing systems, marketing mix models, programmatic auctions and floor settings, search advertising, streaming subscriptions and US courts.

Why: Elasticity tells a seller whether a price change will raise or lower revenue, separates price effects from advertising effects in measurement, and governs how far a platform can lift ad prices. Its estimates are fragile, sensitive to endogeneity, promotions and time horizon, and increasingly contested when computed for individual customers.