Price elasticity by product, store, and cluster
.avif)
Price elasticity varies significantly from one store to another within the same chain. According to an NBER study, uniform pricing results in a 7 to 9 percent loss in profit compared to store-specific pricing.
The store cluster offers the best balance between analytical depth and operational manageability, while remaining consistent with the product lineup as perceived by the customer.
A single price elasticity per reference is a practical convenience, not an economic reality. The same product does not respond to price in the same way in a dense downtown area as it does in a suburban area facing three competitors. This guide explains why granularity—by product, store, or cluster—affects pricing decisions, and how to implement it without creating unmanageable complexity.
One SKU, multiple elasticities
Setting a single national price per SKU is convenient: one number, one decision, one export to stores. However, in most multi-store chains, this is also an approximation that erodes profit margins, because customers’ actual price sensitivity varies greatly from one store to another, depending on local competition, the socioeconomic profile of the catchment area, and foot traffic levels.
A uniform price is not a beginner's mistake: it is a widely used default choice, for reasons of operational simplicity and perceived consistency. The question is not whether it should be abandoned everywhere, but where it actually proves costly.
What the Research Reveals: The True Cost of a Flat Fee
A landmark study by Stefano DellaVigna (UC Berkeley) and Matthew Gentzkow (Stanford), published in the Quarterly Journal of Economics and released by the NBER, analyzed the pricing behavior of major U.S. retail chains. The finding: most of them charge a virtually uniform price across their entire network, even though actual price elasticity varies significantly from one store to another within the same chain.
This is the price elasticity gap measured between stores in the 10th and 90th percentiles within the same U.S. grocery chain—the same network, yet two very different realities in terms of price sensitivity (DellaVigna & Gentzkow, Quarterly Journal of Economics / NBER).
The numerical implication is clear: according to the same study, charging a uniform price rather than a flexible, locally optimized price reduces the chain’s profit by 7 to 9 percent on average, compared to a scenario of store-specific pricing.
in lost profit for chains practicing uniform pricing instead of store-differentiated pricing, according to estimates by DellaVigna & Gentzkow (NBER, Quarterly Journal of Economics).
A gap of this magnitude, applied to a network of hundreds or thousands of stores, represents a margin point that no supplier negotiation can easily recover.
Three levels of granularity, three distinct use cases
- The product. The elasticity of a SKU depends on its category, its positioning (KVI or long tail), and its role in the shopping cart; a loss leader does not have the same price sensitivity as a shelf-filler.
- The store. The same product responds differently based on local competition within a few minutes' walk, the neighborhood's socio-demographic profile, and the point of sale's traffic volume.
- The store cluster. Grouping stores with comparable characteristics (competition, customer base, store format) makes it possible to apply a common price elasticity without having to manage different prices for each store—a trade-off between precision and manageability.
In practice, the cluster is the level most commonly used by mature retailers: granular enough to capture most of the variation, yet aggregated enough to remain manageable by a team of people. See also cross-elasticity, cannibalization, and the halo effect—the other dimension to cross-reference with store-level granularity.
Building relevant store clusters
Choosing clustering variables
Direct competition (number and type of stores in the vicinity), the area's socioeconomic profile, and the store's format and size—not just administrative boundaries.
Validating clusters against actual sales data
Two stores that appear comparable on paper can exhibit vastly different purchasing behaviors once historical sales data is analyzed.
Limiting the number of clusters
A segmentation of 4 to 8 clusters remains manageable for a pricing team. Beyond that, operational complexity often outweighs the precision gains.
Periodically reassess
The arrival of a competitor or changes in a neighborhood can cause a store to shift clusters; segmentation is never set in stone.
The risks of poorly constructed granularity
Segmentation isn't free. Here are three specific risks to consider before introducing multiple price tiers:
- Unmanageable complexity. Having a different price per store, without a tool to standardize the process, amounts to multiplying the amount of manual work—the opposite of the intended goal.
- Customer-perceived inconsistency. A shopper comparing two stores of the same banner on their phone will often struggle to understand a price gap that seems unjustified to them, even if it is economically sound.
- Signal dilution. Too many clusters, based on too little data per store, result in statistically unreliable elasticities; it is better to have a broad, robust cluster than a fine-grained, noisy segmentation.
How a pricing tool manages this granularity
A mature Pricing Optimization Software tool does not calculate a single elasticity value per SKU: it provides a multi-level view, capable of drilling down from the national level to the cluster level—and even to the store level for the most strategic SKUs—while maintaining centralized governance over rules and limits. It is this structure—a consolidated view at the top and granularity available at the bottom—that makes it possible to leverage variations in elasticity without losing overall control.
At Barbotteau Group, a multi-brand retail company in the French Caribbean, the pricing maturity journey documented with Booper illustrates this principle: there, price is treated as one decision-making factor among many, not as the sole lever; the approach combines local elasticity factors with operations research to transform forecasts into decisions, without a “black box” or a one-size-fits-all price imposed on market realities that vary significantly from one retail location to another.
Before segmenting your prices by store
- Do I have sufficient sales data per store to estimate reliable elasticity, or only a few weeks of history?
- Are my clusters validated against actual sales, or just based on geographical intuition?
- Can I explain to a category manager why two stores have different prices?
- Do I have a tool capable of industrializing this granularity, or does it rely on yet another spreadsheet?
Measuring Cluster Elasticity Without Relying on Chance: Controlled Price Tests
Calculating elasticity by cluster is not enough if the measurement method itself is flawed. The controlled price test remains the most reliable way to validate a figure before using it to guide decisions: a control group of stores keeps the price unchanged, while a test group receives the new price, over a sufficient period—typically four to eight weeks—to smooth out short-term seasonal fluctuations such as a holiday or unusual weather.
- Exclude from the test stores that are under renovation, have recurring stockouts, or have recently opened, and whose sales do not reflect normal behavior.
- Include at least eight to ten comparable retail locations per cluster tested: if the number is lower, the measured difference may reflect statistical noise rather than a true price effect.
- Neutralize any known competing promotional campaigns during the test window, or at least document them so they can be excluded from the post-test analysis.
Relate cluster elasticity to the depth of the breakpoint
Elasticity measured by cluster is not meant to be just another number on a dashboard: it directly informs another decision—namely, the depth of the markdown to apply at the end of a product’s life cycle. A cluster with low elasticity—often a loyal customer base that is less price-sensitive—can withstand a later and shallower markdown without compromising inventory turnover. Conversely, a cluster with high price elasticity—which is generally more exposed to direct competition—requires anticipating markdowns and acting more quickly to avoid being left with unsold inventory at the end of the period. This relationship between price elasticity and the markdown schedule is detailed in our article on the timing and depth of markdowns.
Additional Questions
What is the minimum number of stores required in a cluster to ensure reliable elasticity?
There is no universal threshold, but with fewer than eight to ten comparable stores, the measured difference between the control group and the test group becomes difficult to distinguish from simple natural variation in sales. A cluster that is too small gives a false impression of precision: the figure appears accurate when it is not statistically sound. This constraint is directly related to the topic of pricing segmentation, which must take this minimum size into account to remain usable.
How often should a cluster’s elasticity be retested?
An annual review is the minimum requirement, but certain events warrant a retest before the deadline: the arrival of a new competitor in the catchment area, a significant change in the customer base, or a shift in local purchasing power. A cluster built two or three years ago based on data that has since become obsolete leads to pricing decisions that seem consistent on paper but no longer reflect actual customer behavior.
FAQ
The questions we are most frequently asked before getting started.
Rarely done in practice: Most mature retailers use clusters of stores with comparable characteristics rather than setting prices on a store-by-store basis, in order to keep the system manageable without increasing the amount of manual work.
This article points out that actual elasticity varies greatly within a single retail chain: according to DellaVigna & Gentzkow (UC Berkeley / Stanford, NBER), the measured difference between stores in the 10th and 90th percentiles of the same chain ranges from -2.28 to -2.98—a variation too large to be ignored, but too nuanced to be managed on a store-by-store basis without an AI-driven price elasticity model.
Clustering is the approach chosen by mature organizations: granular enough to capture most of the variation, yet aggregated enough to remain manageable by a human pricing team—typically across 4 to 8 clusters.
Stocking individual stores is reserved for the most strategic SKUs, using a system capable of scaling this level of granularity; otherwise, the operational complexity quickly outweighs the gain in precision.
A store cluster consists of retail locations with comparable characteristics (local competition, the area's socioeconomic profile, store format and size) to which a common price elasticity and common pricing rules are applied.
This article details the four-stepconstruction method: selecting clustering variables beyond just administrative geography; validating the clusters using actual sales data rather than intuition; limiting the number of clusters; and periodically reassessing the segmentation.
The cluster falls between two extremes: a single national price, which ignores actual variations in price elasticity, and store-by-store pricing, which is unmanageable without a dedicated tool. For this reason, it is the approach most commonly used by mature retailers.
Its reliability depends directly on the quality of the data used to validateit: two stores that appear comparable on paper may exhibit very different purchasing patterns once their sales histories are compared, which is why a rigorous method for measuring price elasticity is so important.
No. It remains relevant for certain categories that are less sensitive to location, or when the brand consistency perceived by the customer takes precedence over marginal profit gains—a trade-off similar to the one described in our guide to managing retail price-image using the right KPIs. The question isn’t whether to ban it everywhere, but rather to determine where it’s costly.
This article quantifies that cost precisely: according to DellaVigna & Gentzkow, charging a uniform price rather than using store-specific pricing reduces a chain’s profit by 7 to 9 percent on average—a difference that, when applied to a network of several hundred stores, represents a margin point that no supplier negotiation can easily make up for.
A uniform price remains, however, a legitimate default choice for reasons of operational simplicity: it is a trade-off between potential profit margin and perceived inconsistency on the part of the customer, who does not always understand a price difference—which seems unjustified to them—between two stores of the same chain.
The challenge, therefore, is to target differentiation where the difference in elasticity is most significant, rather than applying it uniformly or rejecting it outright.
There is no universal number, but this article recommends staying within a range of 4 to 8 clusters for most organizations: beyond 8 to 10, operational complexity generally outweighs the gain in accuracy.
This limit is not arbitrary: too many clusters based on too little data per store result in statistically unreliable elasticities—a phenomenon this article refers to as “signal dilution.” It is better to have a broad, robust cluster than a fine-grained, noisy segmentation, based on a clear definition of how price elasticity is calculated.
The right number also depends on a pricing team's ability to explain each price difference to a category manager or internally: overly granular segmentation quickly becomes impossible to justify and maintain over time.
Periodic reassessment is just as important as the initial assessment: the arrival of a competitor or changes in a neighborhood can cause a store to switch clusters; segmentation is never set in stone.
By analyzing several variables: direct competition in the vicinity (number and type of stores), tracked through a competitor price monitoring strategy; the socioeconomic profile of the catchment area; and the store format or size—not just administrative boundaries.
This article emphasizes an often-overlooked step: validating the grouping based on actual sales history, not just on self-reported criteria. Two stores that appear comparable on paper may exhibit very different purchasing patterns once the data is compared.
This empirical validation is what distinguishes a truly usable cluster from a mere intuitive geographic grouping: it ensures thatthe common elasticity applied to the cluster reflects market reality, not an approximation.
The approach is never set in stone: segmentation must be reassessed periodically, because local competitive dynamics change faster than the administrative boundaries of a neighborhood.
We need a solution that groups stores into clusters (by region, store type, and local competition), calculates price elasticities by cluster, and enables the management of controlled price variances. A single national price is rarely optimal, but neither is a different price for each store. See geopricing.
Sources: Stefano DellaVigna (UC Berkeley) & Matthew Gentzkow (Stanford), “Uniform Pricing in U.S. Retail Chains,” Quarterly Journal of Economics / NBER Working Paper No. 23996, nber.org · Booper × Barbotteau Group webinar (case study published by Booper).
Lowering a price almost always leads to higher sales; that’s never the issue. The real question is whether the additional volume generates enough margin to offset the margin lost on each unit already sold. The answer depends on two figures that rarely align: the product’s markup rate and its actual price elasticity.
A price change results in both cannibalization (of substitute products) and the halo effect (on complementary products). Ignoring this mechanism is equivalent to measuring only half of the actual impact of a pricing decision.
On average, 22% of the increase in sales of a product on promotion comes from a simple shift from other SKUs in the same product line.
Price elasticity varies significantly from one store to another within the same chain. According to an NBER study, uniform pricing results in a 7 to 9 percent loss in profit compared to store-specific pricing.
The store cluster offers the best balance between analytical depth and operational manageability, while remaining consistent with the product lineup as perceived by the customer.