A US online retailer increased revenue and profit in a controlled field experiment after researchers used pricing tests to estimate how customers respond to its prices and those of competitors.
The result comes from a single retailer and should not be treated as a forecast for other businesses. It shows why knowing a competitor’s price is only part of a pricing decision: the size of a customer’s response can change depending on whether an offer is cheaper or more expensive than an alternative.
Carnegie Mellon University’s Zhaohui Jiang and the University of Michigan’s Jun Li describe the work in a 2025 paper in Operations Research. They developed a method for estimating demand and tested its pricing recommendations in a separate experiment covering more than 10,000 products.
What the researchers tested
The researchers first introduced small randomized price variations to estimate how demand changed. Their model considered a retailer’s own price, competitors’ prices and the possibility that customer behavior differs when the retailer has a price advantage.
They then used the estimates to recommend prices subject to constraints and tested that method in another controlled field experiment. In that experiment, revenue increased by more than 15% and profit by more than 10%, the authors reported.
Those figures describe the tested retailer and comparison in the study. They do not mean every product improved by that amount or that other retailers can expect similar gains. The paper does not provide a general guarantee for automatic pricing systems.
Why matching a competitor is not always enough
Price elasticity describes how strongly demand responds when a price changes. A business with highly price-sensitive customers can lose many sales after a price increase, while another may retain more customers because its delivery, service or product selection is valued.
The same response need not hold above and below a rival’s price. Customers may react sharply when a product becomes the cheaper option, or they may still choose a more expensive retailer because they trust it or prefer its return policy.
For a business, the practical question is how much additional demand a lower price produces and whether the extra sales compensate for a smaller margin per item. Our guide to profit margins explains why revenue and profit can move differently.
How smaller businesses can use the idea
A smaller seller may not have the volume, data or technical team to reproduce an experiment across thousands of products. It can still frame a narrower question: how does a reasonable price change affect sales and contribution for a defined product group?
A useful test needs a meaningful comparison and an agreed duration. Managers should record competing offers, stock availability, advertising, seasonal demand and other changes that could influence the outcome. Random assignment, where feasible, can help separate the price effect from unrelated events.
Businesses should also check whether lower prices attract new customers or simply reduce revenue from buyers who would have purchased anyway. Shipping, fulfillment and returns affect the final economics.
Constraints and customer trust
Frequent automated changes can confuse customers or make a retailer appear unpredictable. Any pricing system also needs to respect applicable laws and avoid misleading claims about discounts or price histories.
The researchers tested a method under the conditions of their partner retailer. Different products, competitors and buyers may call for different rules. The value of the study is its demonstration that carefully designed experiments can inform pricing decisions that would otherwise rely on assumptions about what customers will do.