Elasticity from a past promotion
Derives price elasticity from your sales before and during a promotion, shows a range from two observations, the sales growth needed to pay back, the result of the promotion in profit and the size of the dip after it.
At a glance
- What question does it answer?
- How much demand changes with a discount, whether the last promotion paid back and how many sales were stolen from the future.
- What data is needed?
- Prices before and during the promotion and weekly sales before and during it.
- How to read the result?
- Elasticity is negative when the price falls; the larger the absolute value the more sensitive the demand. Below 1 in absolute terms revenue falls; above 1 revenue grows and profit depends on margin. Compare actual sales growth with the growth needed to pay back and look at the result after the dip. Use the absolute value of this figure in “Weekly markdown ladder” and “Clearance scenarios”: they use the same linear model. Example on the default values: Elasticity from the fact — −2.75×; Range from two observations — −2.75× … −1.75×; Sales growth during the promotion — 55%; Weekly revenue: promotion against base — RUB 48,000.
Why and how to reason
Elasticity answers: by how many percent will sales change if the price changes by one percent. It is taken from fact: sales during a promotion are compared with the base before it. Elasticity = (percent growth in sales) / (percent change in price). It is a negative number because price goes down and sales go up.
But a promotion also gives a “dip” after it: some buyers bought in advance what they would have bought later. So promotion profit is viewed together with the dip.
Example. Price 5,000 to 4,000 (-20%), sales 40 to 62 a week (+55%): elasticity 0.55 / (-0.2) = -2.75. Margin before promotion 3,000, discount 1,000, required growth 3,000 / 2,000 = +50%, achieved +55%: promotion profit +RUB 8 k over two weeks. But after it sales are 33 instead of 40: shortfall 14 units x 3,000 = RUB 42 k. Net -RUB 34 k: the promotion did not pay back. A second observation (RUB 4,500, 47 units) gives -1.75.
When to use it
It comes in handy after a promotion to get an honest estimate of the reaction to price and use it in discount planning. Suitable if the base before the promotion is clean, without other events. Assumption: other factors (season, weather, advertising) did not change. If they did, set a seasonal adjustment. Elasticity from one promotion is noisy: compare several observations. Compare promotions over periods with equal seasonality.
How to read the result
Compare actual growth with the required one: 55 against 50% means the promotion formally paid back. But the dip after it turned the plus into a minus (-RUB 34 k). The range across two observations shows uncertainty: -2.75 and -1.75 give different forecasts. For other calculations use the absolute value of elasticity, and conservatively take the smaller one. Decision: run promotions whose result including the dip is positive and cut those that merely move sales in time. Use the elasticity you found in the markdown ladder and promotion calculation. It differs by category, so do not transfer one figure to everything. Remember it holds for a narrow price range, not for any depth.
Common mistakes
The usual mistakes: computing elasticity from a price change coinciding with a season; taking promotion sales without allowing for advertising; not measuring the dip; applying a promotion’s elasticity to a permanent price cut. Also: confusing the sign of elasticity. Also: taking a promotion without a base, choosing a random week for comparison.
What data is needed?
- Prices before and during the promotion and weekly sales before and during it.
- Weeks of the promotion and the seasonal adjustment if any.
- Optional: unit cost, sales after the promotion and a second observation.
Inputs
| Field | Unit | Value in the example | Meaning |
|---|---|---|---|
| Price before the promotion | ₽ | 5,000 | Prices excl. VAT or both incl. VAT: the point is one basis. |
| Promotion price | ₽ | 4,000 | The shelf price during the promotion, on the same basis (incl. or excl. VAT) as the price before it. |
| Sales before the promotion | pcs a week | 40 | Average weekly sales over 2–4 normal weeks before the promotion, without other promotions. |
| Sales during the promotion | pcs a week | 62 | Average weekly sales during the promotion. |
| Weeks of the promotion | weeks | 2 | How many weeks the promotion lasted. |
| Seasonal adjustment (optional) | % | 0 | How sales would have changed without the promotion over this period (season, advertising). Empty: zero. |
| Unit cost on the same basis as the prices | ₽ | 2,000 | With or without VAT, like both prices. |
| Sales after the promotion | pcs a week | 33 | Average weekly sales after the promotion: show the dip if there was one. Empty: the dip is not calculated. |
| Weeks after the promotion | weeks | 2 | How many weeks after the promotion you observed sales. Empty: the dip is not calculated. |
| Second observation: price | ₽ | 4,500 | Another promotion or discount level, to show a range. |
| Second observation: sales | pcs a week | 47 | Weekly sales at the second observation (another price). Together with the first they give an elasticity range. Empty: no range. |
How it is calculated
| Step | Formula |
|---|---|
| Elasticity | (promotion sales ÷ base − 1) ÷ (promotion price ÷ price before − 1) |
| Sales growth needed | margin ÷ (margin − discount in money), where margin = price before − cost |
| Dip | (base − sales after) × weeks after; loss = dip × margin before the promotion |
Worked numbers
Default values from the app; in the interactive version you enter your own.
- Elasticity from the fact
- −2.75×Simple elasticity from the base: change in demand per 1% change in price, negative when the price falls. Use its absolute value in “Weekly markdown ladder” and “Clearance scenarios”: they use the same linear model, sales = base × (1 + elasticity × discount), which reproduces the observed growth on these data.
- Range from two observations
- −2.75× … −1.75×Do not rely on one figure: use several promotions.
- Sales growth during the promotion
- 55 %
- Weekly revenue: promotion against base
- 48,000 ₽Revenue grows if the absolute elasticity is above 1.
- Sales growth needed to pay back
- 50 %The discount pays in profit if sales grew at least this much.
- Promotion profit for the period, before the dip
- 8,000 ₽
- Dip after the promotion
- 14 pcs · 32 % of the promotion’s extra salesSales missed after the promotion: some customers bought earlier (pull-forward demand).
- Promotion result after the dip
- −34,000 ₽
- To enter in other calculations (absolute value)
- 2.75×The “Discount ladder”, “Clearance scenarios” and price calculations ask for elasticity without the minus sign.
- points
- regression line
Data as a table
| Price, ₽ | Sales per week | |
|---|---|---|
| 5,000 | 40 | before |
| 4,000 | 62 | promo |
The promotion paid within the period but turned negative after the dip: part of the gain is pull-forward demand.
Absolute elasticity is above 1: revenue grows and profit depends on margin (see the sales growth needed).
One model for the whole platform: linear, demand = base × (1 + elasticity × discount). It is exact for the discount you measured and approximate for much deeper ones: for discounts deeper than 40% use a promotion of similar depth. A logarithmic (arc) estimate from the same data gives a different figure and fits a different, power-law model: do not mix them.
Without removing seasonality and advertising the estimate is overstated; one promotion gives an order of magnitude, not a coefficient for the whole brand. Demand also depends on stock and size completeness.
Nuances and limits
- Do not compare a promotion week with a normal one without adjusting for season and advertising.
- Do not use one elasticity for the whole brand.
- Do not forget the dip after the promotion: pull-forward demand.