The myth of price elasticity

A large retail chain, with a portfolio of more than 60,000 SKUs, conducted an analysis of its sales over the last three years. The goal was to measure the price elasticity of each of its products in order to make price optimization decisions. They calculated how much the sales volume of each item had changed due to the price changes experienced during the evaluated period.

To their surprise, the information obtained was of very little use; 98% of the products showed no relationship between volume variations and price changes. A minimal part of the portfolio showed some level of correlation between price and units sold, but this information was insufficient to make decisions for the entire portfolio. What did this company do wrong? Is it a utopia to try to measure price elasticity to optimize the product portfolio?

The measurement of price elasticity
We must begin by acknowledging that price elasticity does exist. The problem is that measuring it is very difficult, as it is necessary to isolate the other variables that influence the units sold: competitors’ prices, stockouts, distribution intensity, shelf placement, own advertising and competitors’ advertising, etc.

Even if a model is built that includes all these variables, it would be very unlikely to achieve an accurate forecast, because one would have to estimate the level each of those variables would have in the future. In other words, it would require a true “crystal ball.”

The true role of price elasticity
Does it then make no sense to try to measure elasticity to make price optimization decisions? The answer is: it depends. It depends on what the elasticity data is intended to be used for. If elasticity is to be used as the determinant of the price, that indeed makes no sense. Saying “since this product has low elasticity, we can raise its price” is quite risky. There is no guarantee that the elasticity measured in a specific period, under specific conditions, will repeat itself in the future. Even if a product’s elasticity was low in the past, it does not mean it will remain low in the future.

But if the elasticity data is to be used not to determine the optimal price, but to make an approximate projection of a fair price calculated through perceived value, then it does make sense to estimate the approximate elasticity of a given product. Under this approach, elasticity is not used to obtain the price; instead, the price is determined first by comparing the price/value relationships of the products available on the market, and then the impact on units generated by the suggested price change is estimated using elasticity. Thus, price elasticity is not the protagonist, but a secondary actor in the price optimization process.

In summary...
The company described at the beginning stopped making decisions based on the elasticity data it had measured. It focused on implementing a process to quantify the value customers perceived when buying the products in its stores compared to the same products in competitors’ stores. Later, it used costs and estimated elasticity to project the financial results it would obtain after adjusting prices.

​Although the forecasts are not 100% exact, they are accurate in their order of magnitude compared to what was originally projected. And this retail chain understood that basing price optimization on elasticity is not just a myth—it is a serious mistake.