Most Retail Promotions Lose Money. Here's How to Find the Ones That Don't

Why do most retail promotions lose money?

Most promotions lose money because they are judged on apparent lift, meaning sales during the promotion measured against a rough baseline, which quietly counts volume the retailer would have sold anyway.

McKinsey has put the share of value‑destroying trade promotion spend in developed FMCG markets at 70 to 90 percent once incrementality is taken into account. NielsenIQ's promo effectiveness work over 2018–2023 lands in a similar place: between 60 and 80 percent of EU retail promotions are not ROI positive. Our own read of the market matches it: up to half of retail sales come from promotions, and only around a third of those promotions end in profit.

The uncomfortable part is that everyone in the process is behaving rationally. The retailer requires promotional support in exchange for shelf position and leaflet space. Trade marketing bonuses are tied to volume and sell‑in, not to margin contribution. And measuring incrementality properly requires a test design, clean baseline data and a model that most mid‑size planning teams do not have. So the calendar grows, the discount deepens, and nobody can prove which half is working.

That is a solvable problem, but not with a better spreadsheet. It needs a different definition of lift.

What is the difference between apparent lift and incremental lift?

Incremental lift is the share of promotional volume that would not have been sold without the promotion. Apparent lift is everything you see in the promotional week, including volume you had already earned.

Take a promotion that sells 3,000 units over two weeks against a 1,000‑unit baseline. Apparent lift is 200 percent, and it looks like a win. Then subtract the shoppers who would have bought the following week anyway, the sales pulled off sister SKUs in the same brand, and the households that stocked up and disappeared for two months. In many categories that leaves something closer to 30 percent of promotional volume as genuinely incremental, and 30 percent of volume sold at a 25 percent lower margin is a very different business case.

The arithmetic is simple once the numerator is honest:

Promo ROI = (Incremental Margin − Promo Cost) / Promo Cost

Incremental margin is incremental volume multiplied by the margin actually earned during the promotion, not the everyday margin. Promo cost is the discount, the trade investment and the activation cost: display, listing fees, leaflet space. Run the same promotion through apparent lift and you can report ROI above 200 percent. Run it through incremental lift and the same event can be negative. Nothing about the promotion changed; only the measurement did.

Academic work has been consistent on the size of the leakage. Nijs and co‑authors, writing in Marketing Science, found that combined pull‑forward and stockpiling effects absorb 40 to 65 percent of apparent lift across many categories. The exact split varies by category, pack size and shopper base, so treat any single ratio as a starting hypothesis to test, not a constant.

Where exactly does the money leak?

Four mechanisms account for nearly all of the gap between apparent and incremental lift, and each leaves a distinct fingerprint in weekly data.

• Pull forward. The shopper intended to buy and simply moved the purchase into the promotional week. The signature is a clean spike followed by a matching dip in the next one to three weeks. Net incremental volume is near zero, and the margin sacrificed during the spike is never recovered.
• Cross‑SKU cannibalisation. The promotion on one variant takes sales from the non‑promoted variant next to it. The signature is a spike on the promoted item alongside an unexplained dip on its siblings. Brand share holds, brand margin does not, because the promoted item carries the lower margin.
• Stockpiling. Households buy several units and stay out of the category for 8 to 12 weeks. The signature is a large spike followed by a long, shallow trough, easy to miss because it sits below the noise floor of a monthly report.
• Base erosion. Frequent deep discounts teach shoppers to wait, so non‑promoted sales weaken and the next promotion has to work harder for the same volume.

NIQ's February 2026 analysis of Western European FMCG shows what base erosion looks like at market level: promotional pressure has risen slightly while unit growth has slowed sharply, and their conclusion is blunt — promotions are now supporting demand rather than stimulating it. Categories and markets with declining non‑promoted sales tend to decline overall, even when promotional activity increases.

The most instructive detail in that data concerns private label. While branded manufacturers increased their reliance on promotions, private label held its promotional mix steady and still outgrew brands in promotional units. The advantage came from how promotions were designed, not from how deep they were cut. That is the whole argument for measurement in one data point.

Base erosion is also where promotion planning stops being a marketing question and becomes a pricing and markdown discipline question. Everyday price architecture decides how much work promotions have to do in the first place.

How can you measure promotion incrementality without an enterprise stack?

A matched test/control design on comparable stores answers most of the question, and it does not require a modelling platform to get started.

The setup is quasi‑experimental. Pick test stores where the promotion runs and control stores matched on baseline volume, format, catchment and customer mix, where it does not. Measure four weeks before the promotion to establish the baseline, the two to four promotional weeks, and — this is the step teams skip — four to eight weeks afterwards, which is the only way pull forward and stockpiling become visible. The difference between test and control across that whole window is your causal lift.

Common practice is 30 to 50 stores on each side for statistically comfortable results. A mid‑size retailer or supplier can run a simplified version with 5 to 10 stores per group; the noise is higher and the confidence interval wider, but it is directionally sound enough to settle an 80/20 decision about whether a mechanic deserves to stay in the calendar. Both figures are planning rules of thumb rather than a standard, and the right sample depends on how variable your store base is.

Manual test/control stops scaling at the point where the promotional calendar has hundreds of overlapping mechanics across banners, categories and channels. Cross‑effects between simultaneous promotions cannot be isolated by store matching alone, and by then the analyst is spending more time reconstructing baselines than deciding anything. That is where item‑level models earn their place: Veritico PROMO forecasts promotional uplift, cannibalisation, halo and post‑promo effects per item, so the evaluation runs as part of the planning cycle instead of as a project.

What does an honest promotion KPI set look like?

Six metrics keep a promotional portfolio under control, and each needs a threshold agreed before the calendar is built, not after the results come in.

• Incremental promo ROI – above 30 %, measured per promotion
• Pull forward share – below 30 %, per promotion
• Cross‑SKU cannibalisation – below 25 %, per promotion
• Average discount depth – below 25 %, reviewed annually
• Share of volume sold on promotion – below 35 %, monthly
• Promo margin contribution vs. baseline – positive, per promotion

These thresholds are common planning rules of thumb, not an industry standard; a discounter and a premium brand will set them differently. What matters more than the exact number is the cadence. Evaluate per promotion, not per month: monthly aggregation reliably hides loss‑making mechanics behind profitable ones, which is precisely why portfolios full of negative promotions look acceptable in the P&L.

Promotional forecast accuracy belongs in the same report as margin, because a promotion that was profitable in theory and unavailable in practice earned nothing. At Mondelez, Logio consolidated promotional and non‑promotional demand into a single forecasting system and standardised the approval workflow; forecast accuracy moved from 50 to 70 percent, which is what made promotional decisions arguable with numbers rather than with negotiating power.

Can you cut the loss‑making half without losing sales?

Yes. A promotional portfolio can usually be reduced by roughly a quarter at unchanged sales, provided item selection is driven by scoring rather than by who negotiates hardest.

METRO Slovakia is the clearest example in our portfolio. The wholesaler runs a 40,000‑plus item assortment across six stores serving more than 180,000 business customers, and the goal was a leaflet where every featured item earned its place. Veritico PROMO scores promotional candidates on suitability and profitability, then assembles the optimal handout for each period against business rules and margin impact. The brochure shrank by 25 percent with sales unchanged, promotional margin rose by one percentage point, profit on featured items improved by 3 percent, and buyers accepted up to 80 percent of the system's automatic suggestions, which matters, because a recommendation engine nobody trusts changes nothing.

Two more data points from the same module. At Kofola, unifying promotional data across departments lifted promotional forecasting to 76 percent and cut depreciation of expired stock by 14 percent. At Apatinska Pivara, part of Molson Coors, centralised promotional data and a cross‑functional workflow raised promotional margin by 1.5 percent.

The mechanics behind results like these are unglamorous. Replace deep discounts of 30 to 40 percent with controlled discounts of 10 to 15 percent supported by activation such as display, demo or secondary placement, which shifts spend from price to visibility. Move from always‑on promotion to a menu of three or four planned events per year per category. Pilot the change in one category with one retail partner, then negotiate the rollout with the results in hand. Retailers are more willing to discuss this than most suppliers expect, because their margin on a value‑destroying promotion is thin too.

Why are promotion planning and supply chain the same decision?

A promotional plan that is not connected to inventory and pricing pays for itself twice: once in lost sales when the promoted item runs out, and again in write‑offs when it does not sell through.

The dependency runs in both directions. Promotional forecasts at item and store level drive replenishment and allocation, so a forecast that is 20 percentage points too low turns a successful campaign into an out‑of‑stock and a customer who buys the competitor. A forecast that is too high leaves promotional stock to be marked down, which is how a positive promotion becomes a negative one after the fact. The post‑promotional dip belongs in the inventory plan as a known event, not as next month's explanation for excess stock.

This is where forecasting alone runs out of road. As we argued in Better Demand Forecasting Won't Fix Your Stockouts, the availability gap closes in execution: in replenishment, allocation and the operational discipline around them, not in the accuracy of the number itself. Promotions are the sharpest test of that, because they compress a month of demand variability into four days.

Logio works this problem from both ends. Consulting establishes how the promotional portfolio actually earns: which mechanics, suppliers and formats pay back, and which are funded by everyday margin. Veritico PROMO then runs the process at scale, with promotional decisions made alongside pricing and inventory rather than in a separate spreadsheet. We find the problem in the data, design the change, and build the tool that operates it.

Frequently asked questions

What is incremental lift in retail promotions? Incremental lift is the share of promotional volume that would not have been sold without the promotion. It excludes purchases pulled forward from later weeks, sales taken from other SKUs of the same brand, and stockpiling that suppresses future demand. Apparent lift is total sales during the promotional period against baseline, which is almost always the larger and less useful number.

How much of a promotional uplift is usually not incremental? Research in Marketing Science found that pull‑forward and stockpiling effects together absorb 40 to 65 percent of apparent lift across many categories, before cross‑SKU cannibalisation is counted. The split varies substantially by category, pack size and shopper base, so the honest answer for any specific portfolio comes from a test, not a benchmark.

Can mid‑size retailers measure promo ROI without enterprise tools? Yes, in a simplified form. A test/control comparison across 5 to 10 matched stores per group, monitored for 8 to 12 weeks including the post‑promotional window, gives directionally reliable answers for most decisions. The limitations are more noise and weaker statistical validity, which makes it suitable for deciding whether a mechanic stays or goes, and unsuitable for fine‑tuning discount depth by a percentage point.

Are all promotions bad? No. Promotions create real value when they have a specific job: introducing a new product, converting light buyers into repeat buyers, defending a position against a competitive push, or clearing end‑of‑season stock. The problem is always‑on deep discounting with no defined objective, funded out of everyday margin and measured on apparent lift.

Make your promotional calendar earn its place

If promotions drive a large share of your revenue but nobody can say which ones pay back, that is a measurement problem before it is a strategy problem. Logio will review your last four quarters of promotional activity, calculate incremental ROI by mechanic, and show how much of the calendar can be cut without touching revenue.

Talk to an expert, or see how Veritico PROMO plans, forecasts and evaluates promotions in one place.