Simulate a promotion:

incremental, cannibalization, margin

Sales volume does not indicate whether the promotion was successful

The sales peak during a promotion is easy to spot and looks good in a presentation. However, it doesn't answer the one question that matters to management: How much did we gain, in terms of margin, compared to a typical week? Part of that peak comes from customers who would have bought anyway, another part from similar items that sold less, and yet another part from purchases that were simply brought forward.

This is by no means a minor issue. Promotions now account for a larger share of retail sales than at any time in the past twenty years, and every percentage point of this poorly managed share directly impacts the margin.

We've already explained in detail how to measure the incremental impact of a promotion after the fact. Here, we're approaching the problem from the opposite angle: how to predict it before making a decision, and how to compare several options on a common basis.

What a Promotional Script Should Include

A scenario isn’t just “25% off item X in week 42.” It’s a promotion mechanism (instant discount, buy one get one half-price, bundle, loyalty discount), a time frame, a store network, a level of visibility (flyers, end-cap displays), and, above all, a projected income statement. For two scenarios to be comparable, this projection must always include the same line items:

Without the last four lines, we're comparing volumes. With them, we're comparing decisions.

Plan the base first, then the incremental part

It all starts with baseline sales. If they’re estimated incorrectly, so is everything else: an underestimated baseline makes any promotion appear profitable. This baseline must take into account seasonality, recent product trends, the calendar (vacations, holidays, paydays), the weather when it’s a factor, and past promotions to avoid “learning” a sales level inflated by previous campaigns.

Next comes the uplift—that is, the response of sales to the discount. It doesn’t depend solely on the discount rate: promotional tactics, in-store placement, and inclusion in flyers often matter just as much. A model that considers only price underestimates highly visible promotions and overestimates discreet discounts. This is precisely why it’s important to work with estimated price elasticities by product and by store group rather than using a national average.

Finally, we must take into account how customers respond to sales. A large proportion of them actively keep an eye out for them, which automatically leads to delayed purchases of non-perishable items.

Cannibalization, purchase deferral, and the halo effect: the three adjustments

Once the uplift has been estimated, three adjustments account for the difference between the displayed volume and the actual volume created.

Cannibalization

When a brand of pasta is marked down by 25%, some of its additional buyers would have purchased another brand of pasta from the same aisle. These sales are not gained; they simply shift to another product line, often to one with a lower margin. They are estimated using the cross-price elasticities between the promoted product and its substitutes. For a detailed explanation of the mechanics, see our article on cross-elasticity, cannibalization, and the halo effect.

Deferred Purchase

For products that can be stockpiled (laundry detergent, canned goods, beverages, personal care items), customers stock up their pantries. Sales rise during the promotional week but fall in the following two or three weeks. A realistic scenario anticipates this dip and factors it in.

The halo effect

Conversely, a promotion can lead to sales of other items: sauce with pasta, charcoal with grilled meats, or simply a full shopping cart for a customer who came in for the special offer. This benefit exists, but it is often cited without specific figures. It must be included in the scenario with a specific amount, not as a general justification.

A numerical example: two scenarios for the same product code

Let’s take a deliberately simple example (illustrative figures). An item sold for €3.00 yields a profit margin of €0.90 and typically sells 1,000 units per week under normal circumstances. Its closest substitutes yield an average profit margin of €0.80 per unit. We’ll compare two one-week immediate discounts.

On the standard dashboard, Scenario A “wins”: +160% in sales versus +80%. Once the adjustments are made, only half of that volume is actually generated, and each unit gained costs €1.23 in margin—more than the product’s normal margin. Scenario B generates half as much volume, but at a unit cost that is one-third lower.

Neither option is “right” in an absolute sense. If the goal is to attract customers in a strategic category, Option A may be justified. If the goal is to support the department’s margin, Option B is more reasonable. What matters is that this choice be made with full knowledge of the facts, understanding the costs involved, and, if applicable, factoring in the supplier’s contribution to financing the operation.

A margin-volume arbitrage that can be explained

A scenario is only useful if it can be discussed. When dealing with a sales manager, a buyer, or a supplier, simply saying “the model recommends B” isn’t enough. You need to be able to show where the numbers come from: how much of an impact seasonality had, which substitute products are losing sales and by how much, and why a dip is expected after the campaign. This is the same requirement as for a traditional pricing decision, which we discussed in “Explaining a Pricing Decision.”

Explainable arbitration is based on four elements:

• a stated objective for the operation (traffic, recruitment, profit margin, inventory turnover), which specifies which metric is used to distinguish between the scenarios;
• scenarios presented in the same format, so that the comparison does not depend on how each one is told;
• the factors behind the forecast made visible, not just the result;
• Requirements met: minimum margin, consistency with the price image, available inventory, and legal framework.

This last point is not merely theoretical in the mass-market retail sector. The cap on promotions, which has been extended through 2025, directly limits the range of possible scenarios and requires retailers to decide where to apply the deepest discounts.

In this context, two additional discount points on one item often mean two fewer points available elsewhere. Only by comparing net margins and incremental unit costs can one properly weigh the options between transactions competing for the same budget.

How Booper Connects Scenario Analysis, Forecasting, and Arbitrage

At Booper, we start with a simple observation: most promotional teams know how to do this analysis, but don’t have the time to repeat it for every campaign and every product. The tool must therefore generate this income statement automatically and present it in a clear, easy-to-read format.

• Forecasting: Sales forecasts take into account seasonality, past promotional data, and price elasticities, with several levels of ambition (conservative, balanced, aggressive) for each scenario.
• Correction: Sales shifted to or from neighboring SKUs are estimated and valued at their own margin, at the category level.
• Explain: An explanatory section lists the factors that influenced the forecast, so that the decision-making process can be based on concrete information.
• Decide: Each simulation shows the impact on sales, margins, and inventory, and then goes through your approval processes, with a record of the decision.

After the operation, the measured results are compared to the selected scenario. This allows for improved future forecasts and helps teams understand which strategies are truly effective for which products. For one of our clients in the food industry—a network of more than 1,700 retail locations—incorporating elasticities and cannibalization into simulations was part of the transition from reactive pricing to predictive pricing.

Five Questions to Ask Yourself Before Approving a Promotion

• What is the objective of the operation, and what metric is used to distinguish between the scenarios?
• How many would have been sold without the promotion during the same period?
• Which SKUs will result in lost sales, and what percentage of profit does that represent?
• What low is expected in the coming weeks?
• How much does each unit actually cost to acquire, and is that acceptable given the objective?

To place promotions within the context of a broader pricing strategy, our article on simulating the impact of pricing decisions applies the same logic to regular prices, while the article on markdowns and clearance sales applies it to end-of-season sales.

Frequently Asked Questions

Short answers to the most frequently asked questions about the promotional simulation.

Sources

NielsenIQ, 2025 Consumer Goods Market Outlook, report published in 2026. OpinionWay for Bonial, French consumers’ purchasing behaviors, reported by Républik Retail on June 26, 2025. Law of April 14, 2025, extending the regulatory framework for promotions until April 15, 2028 (DGCCRF, guidelines). Key figures and client results: Booper data published in *Informations Entreprise*, 2026. Example with figures: illustrative case constructed for the article, without client data. Last updated: September 29, 2026.