How do you understand price elasticity?

Price elasticity is the answer to one question: if we move our price, how much volume do we gain or lose? It is measured against a reference price, the price the customer is actually comparing yours with, not against your own price history. That reference is usually the price they last paid you, or the price of the nearest comparable competitor. Good practice estimates elasticity from granular history against that reference. Better practice tests it with regular, controlled price experiments in small parts of the customer base.

What price elasticity is

Elasticity is the percentage change in volume divided by the percentage change in price. A product with an elasticity of 2 loses about 2% of volume for every 1% of price added. One with an elasticity of 0.3 loses about 0.3%. Below 1 is called inelastic, above 1 elastic.

The number decides whether a price move pays. A price cut only makes money if the extra volume covers the margin given away on every unit. A business earning a 30% contribution margin that cuts price by 5% now earns 25% on each sale. It needs 20% more volume just to stand still. If the market will not deliver that, the cut loses money however busy it looks. The same arithmetic runs the other way: a 5% increase on a 30% margin can lose about 14% of volume before it costs anything. Knowing which side of that line the market sits on is the point of elasticity.

Two elasticities, not one

Elasticity is loose shorthand for two different things, and mixing them up produces bad decisions.

Primary demand elasticity is about the absolute value of the product to the customer. At this price, do they buy the category at all, or buy more of it? It is limited by switching barriers and by how much the customer needs what you sell. It plays out over the long term.

Cross elasticity, or share allocation, is about the relative value of your offer inside a competitive set. At this price gap, do they buy from you or from the alternative? It is limited by availability and awareness, and it plays out fast.

Most of the elasticity a business can act on is the second kind, which is why the reference price matters so much.

The reference price is the whole game

A customer does not respond to your price. They respond to the gap between your price and whatever they are silently comparing it with. That anchor is the reference price, and it comes in two common forms.

The price they last paid you. For repeat purchases, renewals and contracts, the customer's reference is their own history. The signal they read is the change since last time. Customers tolerate small, regular moves far better than occasional large ones. In one general-insurance portfolio, renewal behaviour was driven by two things: the gap to competitors, and the size of the increase since last year. Small annual increases held renewal rates better than holding price for three years and then catching up.

The nearest comparable competitor's price. For new customers, tenders and anything the buyer shops around for, the reference is the best alternative they can see. In that same portfolio, new customers were about five times more price-sensitive than renewing ones: an elasticity of 2 to 3 against 0.3 to 0.5. That is one business, not a rule, but the direction is common. New customers had no history with the business, so the competitor's price was the only anchor they had.

If you regress your own price against your own volume, you have measured neither of these. You have measured how volume moved while your price happened to move. So did the season, the promotions, the stock-outs and your competitors. That is not elasticity. It is noise with a trend line through it.

Estimating elasticity from history

Granular history is the cheapest source of elasticity you have. Most businesses have never used it properly. The method is four steps.

1. State the reference price for each segment. Last price paid, or nearest comparable competitor. If you do not know the competitor's price, that gap in your competitor price record is the first finding.
2. Build the price gap over time. For each product, segment and period, the difference between your price and the reference, and how that gap changed.
3. Line up volume against changes in the gap, not against your price alone. Use the most granular cut you can defend: customer or store, week or month, product or pack.
4. Take out what is not price. Promotions, season, stock-outs, range changes, competitor moves and mix all move volume. Flag them and control for them, or exclude the periods they dominate.

Two warnings. History only reveals elasticity where prices actually moved; a price unchanged for three years has no elasticity in the data. And prices usually moved for a reason, often because demand was already changing, which contaminates the estimate. History gives a directional answer and a shortlist of where to test. It rarely gives the number.

Testing elasticity properly

Experiments are the only clean source of elasticity, because you choose the price move and the customers who see it. Three designs cover most situations.

Elasticity tests. Vary price up and down from standard, typically by 5% to 10%, for a small slice of customers. A tenth to a fifth of a segment is usual. The slice is chosen to give several points on the curve at the least disturbance. One rule cannot be broken: a customer must see the same price in every channel, or the test measures channel leakage instead of elasticity.

Champion and challenger. Keep the standard offer as champion. Run one or two challengers a small step away, a percent or so up and down. After a fixed period the better performer becomes the new champion and the next step is taken. The moves are small, so a challenger can carry a large share of volume safely. The process can run itself, and over time it traces the curve.

Test cells. For bigger moves or new offers, run a handful of designed variants in specific regions, stores, channels or time windows. Monitor them tightly so a bad cell can be unwound in days. These give fewer points on the curve but answer bigger questions.

Where there is no history and no scope to test, survey research stands in: purchase intent at a ladder of prices, or the acceptable price range. Useful for a new product, a weak substitute for an existing one.

The usual ways a test goes wrong are mundane. Stock-outs in one cell. A competitor's move in the middle of the window. Cells too small to separate signal from noise. Each is avoidable with a control group and a calendar.

Elasticity is a distribution, not a number

The average elasticity for a business is nearly useless, because almost nobody is average. One example: in a portfolio estimated customer by customer, about 30% of customers had an elasticity below two-thirds of the average, and about 15% sat above one and a half times it. The shape differs by business; the lesson does not. The profit was in the tails.

That leads to the simplest use of elasticity there is. Where customers are inelastic and margin is thin, raise the price. Where they are elastic and margin is fat, lower it. A business can often do both at once, a few percent each way, and keep the same number of customers at a higher total margin. That is the first thing a segmented estimate tells you.

Where a business stands on elasticity

The practice "how price elasticity is understood" sits in the data, analytics and performance dimension of a stages-of-excellence pricing assessment. For a business that negotiates prices, the five stages read:

• Lagging: Nobody can say what volume a 5% price move would cost or win; the argument is settled by whoever fears the customer most.
• Basic: Elasticity is a regression of our own price against our own volume, and it says whatever the last two years happened to say.
• Competent: Elasticity is estimated from granular history against a stated reference price (what the customer last paid us, or the nearest comparable competitor's price), by segment.
• Advanced: Price experiments run regularly in sub-segments (region, channel, customer cohort, timing) with controls, and the results refresh the elasticity estimates.
• Leading: Elasticity by segment is a maintained model built from experiments and history, and every list or discount move is sized against it before it is made.

Most businesses sit at the second stage and believe they sit at the third. The insight is the reference price. If nobody can say what the customer is comparing your price with, the estimate is a regression, whatever it is called.

Frequently asked questions

Do we need a data scientist? Not for a first estimate. A spreadsheet with price, reference price, volume and flags for promotions and stock-outs, by product and month, shows the shape. The statistics get harder when you want a number to bet a list-price move on. That is what the experiments are for.

What if we cannot see competitor prices? Use last price paid as the reference and start a competitor price record now. A monthly log of list prices for twenty products gives you a reference within a year.

Is elasticity the same as willingness to pay? No. Willingness to pay is the most a customer would pay before walking away. Elasticity is how many customers walk, or buy more, for each step of price. The first is a ceiling, the second is a slope.

How often should the estimates be refreshed? Whenever the reference price moves. A competitor repricing changes the gap, and the old elasticity no longer applies.