Promotion optimization often gets misread as running more promotions or cutting deeper, since that is what the pressure to clear stock rewards in the moment. In practice, promotion optimization is the opposite instinct. It forecasts the demand planning and margin impact of a price before it goes live, so the team can choose the discount with the best projected outcome rather than the deepest one.
Promotions fail on margin for a specific reason: they’re chosen without knowing what they’ll do to demand and profit before they run.
What retailers actually mean by "optimization"
Discounting is a single lever, which you can pull harder or softer. Promotion optimization is a decision: given a demand planning forecast, which price or discount lands closest to a defined profit, revenue, or sell-through goal? Adding more promotions increases activity. It does not, on its own, improve the quality of any single pricing decision, which is what actually moves profitability.
The distinction matters because the two get measured differently. A discount calendar is judged on whether events shipped and volume moved. An optimized promotion is judged by whether the outcome matched the goal and whether a different price would have performed better.
Discounting as a single lever
A markdown is one tool, but a blunt one when used on its own. Drop the red price and some customers who were on the fence convert. But a flat discount treats every product and every shopper the same, when their price sensitivity differs.
The result is a lever that reliably increases sales activity while saying nothing about whether the underlying decision — this discount on this product right now — was the right one. Doing more of a thing faster is not the same as doing the right thing. This is the difference between optimization and mere automation.
Optimization starts with a demand forecast
Optimization starts one step earlier than the discount, with a predictive model of how demand will respond to a given price. Once you can estimate how many units a product will sell at the black price versus a range of red prices, and what that does to margin and sell-through, the discount stops being a guess.
Instead, it is a choice between modelled outcomes. That forecast is the foundation of price optimization generally, and it is what separates a promotion scenario that protects margin from one that simply spends it.
Why more discounts erode the profit they are meant to protect
More discounting tends to erode the very margin it is deployed to protect, through two mechanics that compound over time.
The first is customer conditioning. When a retailer runs frequent, predictable markdowns, its most engaged customers learn to wait for the red price. Baseline sales at full price soften, not because the product is less wanted, but because buyers have been trained that patience pays.
The second is cannibalization: a blanket discount is applied to stock that would have sold at full price anyway, so the retailer pays for volume it already had. A simple example shows how quickly this turns negative and how promotional costs exceed the value gained.
If a sweater carries a black price of $80 at a unit cost of $40, each full-price sale earns $40 of margin. At full price, it would sell around 800 units over the promotional window. The team runs a blanket 20% markdown to be safe, taking the red price to $64 and margin to $24 per unit. The discount lifts volume to 1,000 units, a genuine 200-unit sales lift.
But margin falls from $32,000 (800 × $40) to $24,000 (1,000 × $24). The promotion sold 200 more sweaters and gave up $8,000 in margin to do it, because 800 of those buyers would have paid full price. The retailer handed back $16 on every one of them ($12,800) to win 200 incremental volume units worth $4,800.
The unit count went up, and the margin went down. A demand forecast would have flagged that only about 200 of those 1,000 sales were truly incremental, which changes the decision entirely, from a blanket 20% to a shallower, targeted move, or none at all.
The customer-conditioning problem
Conditioning is the slow cost, because it doesn’t show up in a single campaign report. Each predictable markdown teaches a little more patience, and over several seasons the baseline of full-price demand drifts down. The sign: the business always runs a live offer, and conversion drops sharply when promotions pause. Discounting has stopped being a tool and become a dependency.
When promotions cannibalize full-price sales
Cannibalization is the fast cost and harder to see, since sales shift to red price instead of black price. This is how promotions escalate into price wars. A discount triggers a competitive response, which triggers a deeper discount, and margin drains on both sides while unit share barely moves.
In the Hive Barometer 2026, 29% of retailers named margin erosion as their single biggest pricing cost and 38% pointed to missed revenue opportunities, the two things blanket discounting produces most reliably. It lands against a backdrop McKinsey’s State of Fashion 2026 describes as sustained margin pressure, with pricing and inventory among the most strained parts of the model.
What promotion optimization looks like in practice
The alternative is a forecast-first workflow, and it changes the order of operations. Before any price or discount is published, the team simulates its projected impact on profit, revenue, margin, and sell-through, then chooses the discount with the best forecast outcome rather than the deepest one. The decision is made against numbers, not urgency.
Simulate the impact before the price goes live
Simulation is the core of it. For a given product, the platform projects what a range of prices will do (units sold, margin, sell-through, remaining stock reach) so the team can see the trade-off before committing. The 20% markdown that felt safe can be compared directly against a 10% move, or a targeted markdown on only the sizes and colorways that are genuinely slow.
The discount stops being an act of faith and becomes the option with the best modelled result and the highest forecast accuracy. This is exactly the kind of decision that measuring trade promotional effectiveness is meant to inform, except that it happens before rather than after the spend.
Set the goal, then let the model find the price
The second shift is target steering. Instead of setting rules and discounts by hand, the team sets the commercial goal, whether a season profit target, a sell-through target, or a revenue floor, and the optimizer finds the prices that reach it.
This is where a platform like 7Learnings fits the argument. It forecasts demand at the product level, then works out the prices and markdown timing that hit the stated goal across the full markdown and clearance lifecycle, adjusting as the season moves. The team steers by outcome instead of managing a calendar of individual offers, gaining actionable insights that drive profitability.
Moving from a discount reflex to demand forecasting
Operationally, this reframes how the pricing team spends its time. Instead of deciding the depth of the next markdown event under deadline, the team reviews forecasts, checks that outcomes are tracking to target, and intervenes where the model flags risk. The promotion planning and forecasting workflow replaces the reflex to discount with a decision grounded in projected demand and artificial intelligence-driven optimization.
This forecast-led approach outperforms the discount-led one where it has been measured directly. Off-price fashion retailer Outletcity used forecast-driven end-of-season sell-off optimization, including handling for broken size runs, to lift profit 26% and revenue 11%, with sales up 5%.
Sports fashion retailer INTERSPORT Krumholz saw an even sharper result in a controlled AB test: a 118% profit increase and a 52% revenue increase, with no sacrifice in sales volume.
Both are European results, so they read as validated benchmarks rather than guarantees, but they show what happens when the discount decision is optimized rather than defaulted to, and the return on investment that forecast-driven decisions deliver.
Optimize the decision before the discount
The point of promotion optimization is knowing the forecasted outcome before you discount. Margin becomes a decision you make rather than a casualty you absorb. A heavier promo calendar answers a volume question. It does nothing for the margin question, because the margin question is decided one level up, at the moment you choose the promotional pricing.
The 7Learnings profit uplift calculator lets you estimate the profit you could recover from smarter promotion decisions in a couple of minutes. Run your own numbers and see how wide the gap is before the next markdown event.
Frequently asked questions about promotion optimization
What is promotion optimization in retail?
Promotion optimization is the practice of forecasting the demand and margin impact of a price or discount before it goes live, then choosing the option that best hits a defined commercial goal such as profit, revenue, or sell-through.
It treats a promotion as a decision to be modelled rather than a lever to pull harder. The aim is to make each promotional decision with a clear view of its projected outcome across units, margin, and remaining stock, rather than simply running more or deeper discounts.
Is promotion optimization the same as discounting?
No. Discounting is a single action: lowering a price. Promotion optimization is the decision process that determines whether, when, and how much to discount, based on a forecast of how demand will respond. You can discount without optimizing — most reactive markdowns do exactly that — but you cannot optimize without a demand forecast.
The difference shows up in the margin report. Discounting moves volume, while optimization protects the margin that volume is supposed to earn.
How do retailers optimize promotions without eroding margin?
They forecast demand first, then simulate several price scenarios to see the projected impact on profit, margin, and sell-through before committing. That lets them avoid the two classic leaks: discounting stock that would have sold at full price, and conditioning customers to wait for the next markdown.
Rather than applying a blanket percentage, they target discounts to the products and variants that genuinely need them, and set a commercial goal for the optimizer to steer toward across the season.
What data do you need for promotion optimization?
At minimum, historical sales and price data, current stock levels, and product cost and margin, ideally at the SKU level so the model can estimate how demand responds to price for each item. Competitor prices, seasonality, and channel information sharpen the forecast further.
The key requirement is enough price and sales history for a model to learn elasticity; where individual products are too slow-selling to learn from alone, machine learning models transfer signal from similar items so even long-tail SKUs can be priced on predicted demand.
