You're staring at a spreadsheet. Column A has prices. Column B has quantities sold. Your boss wants to know: if we raise prices 5%, what happens to revenue?
Most people guess. They shouldn't.
Price elasticity of demand isn't academic theory. And measuring it? It's the difference between a profitable price increase and a revenue disaster. That's where most businesses — even smart ones — get tripped up.
What Is Price Elasticity of Demand
At its core, price elasticity of demand measures how sensitive customers are to price changes. That's it. No jargon required.
When price goes up 10% and demand barely budges — maybe it drops 2% — you've got inelastic demand. People need your product. Think insulin, gasoline, or that one coffee shop on your commute with no competitors nearby.
When price goes up 10% and demand tanks 25%? They'll walk. Customers have options. That's elastic demand. Streaming subscriptions, generic snacks, most SaaS tools with competitors — these live in elastic territory But it adds up..
The formula looks simple:
Price Elasticity of Demand = % Change in Quantity Demanded ÷ % Change in Price
But the devil lives in how you calculate those percentages. And that's where people go wrong Most people skip this — try not to. Less friction, more output..
The midpoint method vs. the simple method
Here's what most textbooks don't make clear: the "simple" percentage change formula gives you different answers depending on direction.
Price rises from $10 to $12. Think about it: that's a 20% increase using the simple method (2/10). But if price falls from $12 to $10? This leads to that's a 16. Here's the thing — 7% decrease (2/12). So same absolute change. Different percentage. Different elasticity result Simple as that..
The midpoint method fixes this. You divide the change by the average of the two values:
% Change = (New - Old) ÷ ((New + Old) / 2)
So $10 to $12 becomes 2 ÷ 11 = 18.Because of that, 18%. And $12 to $10 becomes -2 ÷ 11 = -18.18%. Now, consistent. Symmetrical. This is the standard in serious economics work — and you should use it too.
Types of elasticity you'll actually encounter
Perfectly inelastic (0): Quantity doesn't change at all. Theoretical. Doesn't exist in real markets.
Inelastic (0 to -1): Price changes hurt volume less than they help revenue. Raise prices, make more money Surprisingly effective..
Unit elastic (-1): Revenue stays flat. Price up 10%, volume down 10%. Rare in practice.
Elastic (< -1): Price changes hammer volume. Raise prices, lose revenue. Cut prices, gain revenue (if margins allow) Most people skip this — try not to. That alone is useful..
Perfectly elastic (-∞): Any price increase kills all demand. Commodity markets with perfect substitutes. Also theoretical.
Most real products live between -0.5 and -3.0. Where yours sits determines your pricing power.
Why It Matters / Why People Care
Revenue optimization. That's the short answer Took long enough..
But let's be specific. If you don't know your elasticity, you're making pricing decisions blind. And pricing is the single most powerful profit lever most companies have But it adds up..
A 1% price increase on a product with -0.5 elasticity? Now, revenue goes up. Volume drops 0.5%, but the higher price on remaining units more than compensates. Here's the thing — a 1% price increase on a product with -2. And 0 elasticity? In practice, revenue tanks. On the flip side, volume drops 2%. You lose on both fronts.
Real stakes, real examples
Netflix learned this the hard way in 2011. They split DVD and streaming plans, effectively raising prices 60% for combined users. Stock dropped 77%. Even so, they lost 800,000 subscribers in a quarter. That said, elasticity turned out to be higher than they modeled. They reversed course The details matter here..
Meanwhile, Apple prices iPhones with near-total confidence in inelastic demand. Consider this: their ecosystem lock-in, brand loyalty, and lack of true substitutes for their target demographic means they can push prices up year after year. Volume dips slightly. Revenue soars Easy to understand, harder to ignore..
The difference? One knew their elasticity. The other guessed.
Beyond pricing: product decisions, promotions, forecasting
Elasticity informs more than price tags And that's really what it comes down to..
- Promotional planning: If elasticity is -1.8, a 15% discount drives 27% more volume. Worth it? Depends on margin. But now you can calculate it instead of guessing.
- Product line architecture: You want a mix. Some inelastic "cash cow" products. Some elastic "traffic drivers." Measuring elasticity tells you which is which.
- Forecasting: Input cost rising 8%? Elasticity tells you exactly how much volume you'll lose if you pass it through — or how much margin you'll sacrifice if you don't.
How to Measure Price Elasticity of Demand
This is the section you came for. There are three main approaches, each with trade-offs. Most businesses should use at least two.
1. Historical data analysis (regression modeling)
You have sales data. Even so, you have price data. Practically speaking, maybe you have competitor prices, seasonality indicators, marketing spend, economic indicators. Good. Use them.
The basic regression looks like:
ln(Quantity) = β₀ + β₁ × ln(Price) + β₂ × Controls + ε
The coefficient β₁ is your elasticity. That said, because it's log-log, a 1% change in price predicts a β₁% change in quantity. Day to day, clean. Interpretable That's the whole idea..
But — and this is critical — correlation isn't causation. Prices don't change randomly. They change because you changed them, often in response to demand shifts, competitor moves, or cost changes. Also, that's endogeneity. It biases your estimate.
How to handle endogeneity (the short version)
- Instrumental variables: Find something that affects price but not demand directly. Cost shocks (commodity prices, tariffs, exchange rates) are classic instruments. If coffee bean prices spike, your coffee shop raises prices — but the bean price doesn't directly affect how many lattes people want. That's a valid instrument.
- Natural experiments: Did a competitor open nearby? Did a tax change affect only certain regions? Did a supply chain disruption force a temporary price hike? These are gold. Use them.
- Control for everything you can: Seasonality, marketing, competitor prices, weather, unemployment, consumer confidence. The more controls, the less omitted variable bias.
Practical regression tips
- Use log-log specification for constant elasticity. Use linear if elasticity varies with price level (it often does).
- Cluster standard errors by store, region, or time period. Sales data is correlated within clusters.
- Test for structural breaks. Elasticity during COVID? Different. Elasticity during a recession? Different. Don't pool regimes blindly.
- Minimum data requirements: At least 50-100 price-change observations. More if you have many controls. Monthly data for 3 years? Maybe enough. Weekly data for 2 years? Better.
2. A/B testing (field experiments)
At its core, the gold standard. Randomly assign customers to different prices. Measure the difference. No endogeneity. Clean causality.
But it's not free Surprisingly effective..
Designing a valid price test
- Randomize at the right level. User-level randomization leaks (people talk, share screenshots). Session-level is safer. Geographic or store
