Insights lesenGespräch

Promotions

What is actually left?

Are Discount Promotions Profitable? Margin and Cannibalization

Promotion revenue almost always looks good. Two questions show whether a promotion actually made money: What is left over? And what would have been bought anyway?

14 min read

Key takeaways

  • Promotion revenue says little. What matters is the incremental contribution margin: what does the promotion deliver compared with the case without it?
  • A flat fixed-cost allocation makes successful promotions look bad: it charges them fixed costs that would be incurred without the promotion anyway. Less is left per order, but more in total.
  • Some promotion purchases would have happened anyway. A random holdout group measures this most reliably.
  • Even a 20 % discount can pay off – if the promotion delivers contribution margin that would not have existed without it.
  • Ranked by revenue and then by incremental contribution margin, promotions often end up in a completely different order.
Contents

The summer promotion went well. More orders, more revenue, plenty of redeemed coupons. Yet the real question usually stays open: was the promotion worth it?

That question is harder than it sounds. Promotion revenue only shows what was sold during the promotion. It does not show what is left after discounts and costs. And it does not show who would have bought anyway.

You are not alone in this. In consumer goods, 59 % of trade promotions lose money, and in the US it is as high as 72 %. McKinsey reports this based on a Nielsen analysis (McKinsey, 2019). The figure comes from brick-and-mortar retail. The pattern applies equally in online shops: many promotions look successful without being so. And some are more successful than the report shows.

59 %of trade promotions in consumer goods lose moneyMcKinsey 2019, citing Nielsen 2016
33 %of the incremental sales lift comes from brand switching – the rest from things like purchase acceleration and larger quantitiesvan Heerde et al., 2003 – studied on supermarket promotions

Three terms up front

To keep the calculation clean, this article distinguishes three measures:

  • Contribution margin: revenue after discount minus the costs incurred per order – cost of goods, shipping, packaging, payment.
  • Incremental contribution margin: what the promotion actually earns on top of the case without a promotion.
  • Promotion result: incremental contribution margin minus the costs that arise only because of the promotion, such as a mailing. It is the basis for any promo ROI.

All amounts in the examples are net, that is, excluding sales tax.

Why promotion revenue is misleading

A promotion report usually compares the promotion week with the previous week or the previous year. Two things are missing from that comparison.

First, the discount in the contribution margin. A 20 % discount reduces revenue by 20 %. Cost of goods, shipping, and fulfillment stay the same. The discount therefore hits the contribution margin in full.

Second, the purchases that would have happened anyway. Loyal customers pull their next purchase forward. Others would have ordered even without a coupon. A widely cited study shows that only about a third of the promotional lift there came from brand switching. The larger part came from effects such as purchase acceleration and larger purchase quantities (van Heerde, Gupta & Wittink, 2003). This is a finding from the categories studied, not a fixed split for every discount promotion. But it shows how large the share can be that is not truly new. Circana calls this “pull-forward”: the same money, just earlier and usually at a lower margin (Circana, 2026).

A promotion does not pay off because it generates revenue. It pays off when it generates incremental contribution margin.

What is left of a promotion order

The first question is: how much contribution margin is left per order? A simple step-by-step calculation is enough.

Cost of goods is the key item. Keep it at the same amount. A discount changes nothing about the purchase or production cost of the goods. Anyone expressing cost of goods as a percentage should therefore base it on the list price, not on discounted revenue.

A worked example with round numbers: the basket has a list value of €100 net. Cost of goods is €25. Shipping, packaging, payment, and service cost €12 per order.

Worked example: step-by-step calculation per order

Without discountWith 20 % discount
List value of the order (net)
€100
€100
− Discount
€0
−€20
= Net revenue
€100
€80
− Cost of goods (unchanged)
−€25
−€25
− Shipping, packaging, payment
−€12
−€12
= Contribution margin
€63
€43
Fictional example. The discount reduces revenue and contribution margin by €20 each. For revenue that is 20 %; for contribution margin it is almost a third.

Without a discount, €63 of contribution margin is left. With a 20 % discount, €43 is left. That a discounted order brings in less is obvious. But it does not yet say anything about whether the promotion is worthwhile.

The real question is: does the promotion bring in enough additional orders to make up for the lower amount per order? In the example, the threshold is about 47 % more orders. Whether that is achievable depends on factors such as target group, discount effect, and basket size. Only additional orders count here, not orders pulled forward.

Why a flat fixed-cost allocation makes successful promotions look bad

Many reports spread fixed costs as a percentage of revenue: team, software, agency, warehouse, administration. This is convenient, but it gives a false picture for promotions. Every order then carries its share, including every additional one. A successful promotion with high volume and revenue is therefore charged an especially large amount of costs. Yet the existing fixed costs do not rise because of the promotion.

A fictional example: a shop has €30,000 in fixed costs per month. In a normal month, 1,000 orders come in at full price, each with €63 contribution margin. In a promotion month with a 20 % discount, 2,000 orders come in, each with €43 contribution margin. Half of them would have come even without the promotion.

Worked example: flat fixed-cost allocation versus contribution margin accounting

Flat fixed-cost allocation

Fixed costs allocated at 30 % of revenue

Normal monthPromo monthOrders1,0002,000Revenue€100,000€160,000Contribution margin€63,000€86,000− Allocation 30 %−€30,000−€48,000= Result€33,000€38,000per order€33€19
Promotion seemingly adds +€5,000
Contribution margin accounting

Fixed costs as they are actually incurred

Normal monthPromo monthOrders1,0002,000Revenue€100,000€160,000Contribution margin€63,000€86,000− Fixed costs−€30,000−€30,000= Result€33,000€56,000per order€33€28
Promotion actually adds +€23,000
Fictional example: €30,000 in fixed costs per month, €63 contribution margin per order at full price, €43 with a 20 % discount, all amounts net. The allocation charges the promo month €48,000 in fixed costs – €18,000 more than exist. The flat allocation underestimates the promotion by exactly that amount.

The flat allocation spreads fixed costs at 30 % of revenue, as in the normal month. On €160,000 in revenue, that is €48,000, or €18,000 more than the fixed costs that exist at all. Per order, €19 is left instead of €33. The promotion looks weak, and its plus of €5,000 seems hardly worth the effort.

Contribution margin accounting deducts the fixed costs only once, because they are only incurred once. The monthly result rises from €33,000 to €56,000. The incremental contribution margin of the promotion is €23,000. Without it, this amount would never have existed.

The amount per order is smaller. The overall result is larger. For the question “promotion or no promotion?”, contribution margin is therefore the right basis. This is not a criticism of full costing as such. It remains important, but for a different question: whether the business as a whole is sustainable. Done properly, it allocates costs by cause rather than as a flat share of revenue. The problem is the flat allocation when it is used to decide on individual promotions.

One exception applies. What matters is not whether a cost type is generally called fixed or variable. What matters is whether, and in what amount, it is additionally incurred because of the promotion. If a large promotion requires temporary warehouse staff or more customer service capacity, these extra costs belong in the calculation.

Assign marketing costs by cause

Marketing costs in particular often distort the picture. Many reports include them as a fixed share of revenue. Each promotion then carries a portion of the agency retainer, newsletter software, marketing team, and image campaign. But the promotion did not cause these costs. They are incurred even if the promotion never takes place.

The result is the same as with fixed costs: high-revenue promotions are charged a lot of costs that are not theirs and look weak. Small promotions, by contrast, look strong. In the worst case, a good promotion is cancelled. The costs remain anyway; they are simply spread across the rest.

You should assign only what would not have been incurred without the promotion:

  • Printing and postage for mailings or inserts
  • additional media budget for exactly this promotion
  • Free gifts, free samples, and promotion commissions
  • external costs that arise only for this promotion

Everything else belongs in the overall marketing calculation, not in the evaluation of the individual promotion. If these costs are allocated anyway, exactly the distortion from the example above arises.

Measuring promotion cannibalization

The second question is: which purchases would have happened anyway? This share is called the cannibalization rate.

A comparison with the previous year is not enough for this. Season, assortment, and customer base change. The measurement only becomes reliable with a control group. There are two ways to do this.

Best: a random holdout group

Before the promotion, customers are randomly split. Most receive the promotion, a small share does not. Afterward, you compare both groups as a whole: revenue and contribution margin per customer, not just the buyers. The difference is the true incremental effect. This is easy to implement for mailings and newsletters.

If no test is possible: approximation using comparable customers

Some promotions are visible to everyone, such as a discount directly in the shop. In that case, an approximation using comparable customers remains:

  1. Divide customers into groups. For example by RFM segment, purchase history, or days since the last purchase.
  2. Compare promotion customers with customers in the same group who did not take part in the promotion. They show how revenue would have developed without the promotion.
  3. Set a fixed observation period. For example from the start of the promotion to one repurchase cycle after it ends. If customers tend to wait for promotions, it begins a few weeks earlier. All following figures refer to exactly this period.
  4. Estimate the expected revenue. How did the comparison group’s revenue develop compared with the same period last year? Apply this trend to the promotion customers’ revenue from last year. This gives their expected revenue in the observation period had there been no promotion.
  5. Estimate the rate. First determine the incremental revenue: the promotion customers’ actual total revenue in the observation period minus expected revenue. Total revenue includes promotion orders and all other orders. The cannibalization rate is the part of promotion revenue that is not incremental revenue: 1 minus incremental revenue divided by promotion revenue. All revenue figures need the same definition, for example net, after returns, and across all channels.

A short worked example: the promotion customers had €100,000 in revenue in the prior-year period. The comparison group grew by 5 %. Expected revenue is therefore €105,000. In reality, the promotion customers bought €165,000 in the same period: €120,000 with the promotion, €45,000 without. Incremental revenue is 165,000 − 105,000 = €60,000. The rate is 1 − 60,000 ÷ 120,000 = 50 %. Half of the promotion revenue would therefore have come anyway.

The observation period deliberately includes the promotion itself. Only then are both effects counted: customers who would have bought during the promotion period anyway and simply take the discount, and purchases pulled forward from before or after. Looking only at the time outside the promotion shows just the pull-forward effects and underestimates cannibalization.

Important: this formula delivers an estimate, not an exact cannibalization rate. It holds only if the period and the revenue definition are defined consistently. In addition, purchases shift in time, for example from the weeks after the promotion into the promotion. That is why the observation period must include the time after the promotion.

There is also a second weakness: people who buy during a promotion have chosen to do so themselves. Promotion buyers often buy more frequently anyway, even within the same group. The result should therefore be reported as a range, not as a single value. Calculate with two groupings and draw bootstrap samples. That yields three honest verdicts: cannibalization demonstrated, not demonstrated, or more of an incremental boost.

The right time window

With both approaches, the time window must extend beyond the promotion. Only then does it become clear whether purchases were merely pulled forward. How long depends on the purchase rhythm of the assortment. As a rule of thumb: at least one typical repurchase cycle after the promotion ends.

When the ranking flips

Only both questions together yield the incremental contribution margin. A fictional example using the values above shows why this matters.

Promotion A gives a 20 % discount. It brings 1,000 orders and €80,000 in revenue. Only 300 of these customers would have bought anyway. The 700 additional orders bring €43 each. The 300 regular buyers were given €20 each. Incremental contribution margin: 30,100 minus 6,000 equals €24,100.

Promotion B gives only a 5 % discount. It brings 1,500 orders and €142,500 in revenue. But 1,200 customers would have bought anyway. The 300 additional orders bring €58 each, and €5 each is given away. Incremental contribution margin: 17,400 minus 6,000 equals €11,400.

B ahead of Aby revenue: €142,500 versus €80,000Fictional worked example
A ahead of Bby incremental contribution margin: €24,100 versus €11,400Fictional worked example

By revenue, B clearly wins. By incremental contribution margin, the promotion with the higher discount brings in more than double. The size of the discount alone therefore does not decide. What matters is how much additional business a promotion triggers. A revenue report misses exactly this reversal.

Include campaign costs

Some promotions have direct costs: printing and postage for a mailing, additional agency costs caused exclusively by the promotion, a dedicated media budget. These costs should be deducted from the incremental contribution margin, not from revenue. What remains is the promotion result.

An example: a mailing promotion, such as a postcard to 10,000 customers, costs €0.60 per piece, or €6,000. The promotion result is then €18,100 for promotion A and €5,400 for promotion B.

It is also worth comparing the same promotion with and without the promotional material. Do customers who receive the mailing buy more than customers who only see the online promotion? Only then does it become clear what the mailing itself delivers.

Checklist for the next promotion

  • Calculate the threshold beforehand: how many additional orders does the promotion need to break even?
  • Calculate net, keep cost of goods unchanged, and include fulfillment costs per order.
  • Assign only costs that arise additionally because of the promotion – including extra warehouse or service staff.
  • Assign every order unambiguously to a promotion, for example via coupon code or campaign ID.
  • Wherever possible, define a random holdout group before the promotion.
  • Align the time window with the purchase rhythm: at least one repurchase cycle after the promotion ends.
  • Report results as a range, not as a point value.
  • Compare promotions by incremental contribution margin and promotion result, not by revenue.

Conclusion: calculate properly instead of judging across the board

Discount promotions are not a mistake. Even a high discount can pay off if the promotion delivers contribution margin that would not otherwise have existed. That the individual order brings in less is no argument against it. But promotions need an honest calculation: contribution margin instead of flat allocation, overall result instead of amount per order, incremental instead of purchases that would have happened anyway. Most of the data needed is already available: orders and coupons in the shop, purchase prices in the ERP, campaign costs in accounting.

Anyone who calculates this way will find promotions that are better than their reputation, even with a high discount. And others that merely shift revenue. Only this view shows which promotions are worth repeating.

Frequently asked questions

What is a cannibalization rate?

The share of promotion revenue that would have been generated even without the promotion. At a rate of 60 %, six out of every ten euros of promotion revenue would have come in anyway, just at full price or later. The rate can only be estimated, most reliably with a random holdout group.

What is the difference between contribution margin, incremental contribution margin, and promotion result?

Contribution margin is what remains after the discount and the costs per order. Incremental contribution margin is what the promotion brings in on top of the case without a promotion. The promotion result deducts from that the costs that arise only because of the promotion, such as for a mailing.

Why is comparing with the previous year not enough?

Season, assortment, prices, and customer base change. A control group shows better what would have happened without the promotion. A random holdout group, defined before the promotion, is the most reliable.

What data do I need for the analysis?

Orders with customer ID and promotion ID, list and selling prices, purchase or production costs, plus fulfillment and campaign costs. For the control group, also the customers’ purchase history.

Should I use contribution margin or full costing?

For the decision on an individual promotion, use contribution margin. Existing fixed costs are incurred even without a promotion. The main problem is a flat allocation by revenue: it penalizes high-revenue promotions. Full costing, by contrast, shows whether the business as a whole is profitable.

Which marketing costs belong to a promotion?

Only costs that would not have arisen without the promotion: for example printing and postage, additional media budget, free gifts, or promotion commissions. Agency retainers, software, and the in-house team are incurred anyway. If they are allocated proportionally, high-revenue promotions look worse than they are.

Sources

Inhalt

    InsightsIns Gespräch kommen