Blanket Discounts Are the Most Expensive Way to Clear Stock

What does a discount have to sell to pay for itself?

A discount pays for itself only when it lifts volume by more than it cuts margin per unit, and the required lift grows much faster than most pricing decks assume.

The break‑even multiple is simple arithmetic: divide gross margin by gross margin minus discount depth. That gives the factor by which unit sales have to rise for gross profit to stay flat.

• At 30% gross margin: a 10% cut needs 50% more units, a 20% cut needs 200% more, a 25% cut needs 500% more, and a 30% cut can never pay for itself.
• At 40% gross margin: a 10% cut needs 33% more units, a 20% cut needs 100% more, a 30% cut needs 300% more.
• At 50% gross margin: a 20% cut needs 67% more units, a 30% cut needs 150% more, a 40% cut needs 400% more.

Read the middle line again. At the gross margins typical of grocery and general merchandise, a 20% category‑wide discount is a bet on tripling volume. On products whose full‑price demand was intact, that bet loses every time.

Two honest caveats. Basket effects can justify selling a single item below its own break‑even, because the customer walks in for the discounted item and buys six others at full price. That is a promotion decision with its own measurement problem, and we wrote about why most retail promotions lose money separately. The arithmetic above also ignores fixed‑cost absorption, which matters when you are clearing warehouse space you pay for by the pallet.

Neither caveat changes the core problem: a blanket discount applies one number to a set of products whose economics differ item by item.

Why is clearance a different calculation from a discount?

Because the reference point changes. A discount trades margin for volume on stock that would have sold anyway. A markdown recovers value from stock that would otherwise be written off. For that second group, the alternative is not full price. The alternative is zero, plus the cost of disposal.

That single difference inverts the logic. On terminal stock, the correct markdown is often deeper than a pricing manager would ever sign off, because anything above disposal cost is money you were not going to see. On healthy stock, the correct markdown is nothing at all.

Most retailers run one process for both, and a single process split across two opposite cases lands in the worst place available: too shallow to clear the dead stock before it expires, too deep on the stock that was selling fine.

There is a second, slower cost on the healthy side. Discounting an item with intact demand does not create sales, it moves them in time, and it teaches customers which categories are worth waiting for. That behaviour compounds across seasons and shows up later as a structurally lower full‑price sell‑through, which then gets treated as a reason for deeper discounts.

Why does timing beat depth?

Because the number of units you can still sell equals your remaining selling days times the daily rate, and the days run out whether you act or not. Depth only changes the rate. It cannot buy back time.

Take 300 units of a dated product with six selling days left and a baseline of 20 units a day. Assume, purely for illustration, that a 15% cut lifts the rate to 30 a day and a 40% cut lifts it to 45.

• Do nothing: 120 units sold at full price, 180 written off. Revenue index 120.
• Cut 15% on day one: 180 units sold at 0.85, revenue index 153, and 120 units still to deal with.
• Wait until day four, then cut 40%: 60 units at full price plus 135 units at 0.60, revenue index 141, and 105 units written off.
• Cut 15% on day one, then 40% for the last two days: revenue index 156, 90 units written off.

The early shallow move beats the late deep one, and it leaves the option of a second step. The late cut spends more margin per unit to reach fewer remaining days.

Those response rates are assumptions, and that is the point worth taking from the example. The shape of the answer holds regardless of the exact numbers, but the decision of when to start and how deep to go is only as good as your own measured price response. Which brings us to the part that usually blocks this work.

Which stock are you actually pricing?

Before depth or timing, the stock has to be sorted, and three flags do most of the work. Each one is a question your data can answer today.

Is full‑price demand still intact? Compare recent full‑price sell‑through against plan for that item and store format. If the item is tracking to plan, discounting it buys volume you already had.

Is there a hard deadline? An expiry date, the end of a season, a contractual return window, a store closure. A deadline turns the decision into arithmetic with a fixed number of days in it. No deadline means you have the luxury of waiting and watching.

Will the item come back next season? For a repeating line, the reference price is an asset and every deep cut spends a piece of it. For a one‑off or a discontinued line, there is no reference price left to protect.

Those three flags produce four treatments, and only two of them involve a markdown at all.

• Demand intact, no deadline: leave the price alone and fix the forecast that put the stock there.
• Demand intact, hard deadline, item repeats: a shallow ladder that starts early, with a ceiling that protects the reference price.
• Demand broken, hard deadline, item does not repeat: one decisive cut, deep enough to clear within the window, because there is nothing left to protect.
• Demand broken, deadline passed or nearly passed: stop pricing and move it. A secondary channel, a return to the supplier, or donation will beat a price nobody will pay, and the logistics of that route are worth designing rather than improvising, which is what we did when we modelled distribution scenarios for donated ready meals.

The reason blanket rules survive is that this sorting takes data and the blanket rule takes a meeting. A category‑wide 20% is agreed in ten minutes and nobody has to defend an item‑level judgement. The cost of that convenience does not show up in the pricing report, it shows up in gross margin.

What data do you need before optimising markdowns?

Three layers, and the first one gates the other two.

Age and remaining life at item and location level. Not category averages. A category‑level view will tell you that dairy is fine while 40 stores each sit on stock that expires on Thursday. Most companies have this in the ERP or the warehouse system already.

Measured price response per item and store format. Historical, not assumed. This is where the work usually stalls, because price history, promotional flags and stock movements live in different systems and nobody has joined them at item‑store‑day level. Until they are joined, every markdown rule is a guess with a spreadsheet attached.

An auditable record of who set which price, when, and what followed. Without it you cannot separate the effect of a price cut from the effect of the weather, a competitor campaign or the school holidays, which means you never learn and every season starts from the same argument.

The honest version of the sequencing question: if you have the first layer and not the second, this is a matter of weeks. If you have neither joined nor trustworthy stock data, expect quarters, and fix the data before buying an optimiser. That is also why demand and inventory data and pricing and markdown management belong in the same programme rather than in two separate projects.

The measurement layer itself is not something worth building from scratch. Veritico PRICE pairs the data and process work with the Yieldigo pricing engine, which handles the elasticity modelling and the price‑setting mechanics, so the project becomes a question of data readiness and decision rules rather than of algorithms.

What does good markdown governance look like?

Rules, a cadence, and one place where the decision is recorded.

Rules by stock segment. Decide up front which cases are handled automatically and which go to a human. Short‑dated food with a clear expiry and a measured response curve does not need a meeting. A flagship line at the end of its first season does.

A fixed cadence. Weekly review beats event‑driven panic, because the calendar forces the early, cheap decision to be made while it is still cheap.

A ceiling on cumulative depth. Without one, sequential cuts drift into territory where the item would have been better donated or returned.

Metrics a CFO can read. Share of volume sold at a reduced price, average depth, recovery rate on terminal stock, and written‑off value. The last two are the pair that matters. A high recovery rate with rising write‑offs means you are acting late. A low recovery rate with low write‑offs means you are discounting stock that did not need it.

Note what is missing from that list: “sold out”. Selling everything is not the objective, and it is trivially achievable at a low enough price.

Where this stops being a pricing problem

Repeated deep markdowns in the same category, season after season, are not a pricing error. They are a buying or assortment error that pricing is being asked to clean up.

The test is simple. Count the share of SKUs that needed a markdown in more than one consecutive season. If that share is concentrated in a few categories, the fix is upstream, in assortment and range decisions or in order quantities, not in a smarter discount ladder. Pricing can recover value from a bad buying decision. It cannot make it a good one.

Frequently asked questions

What is the difference between a discount, a promotion and a markdown?

A discount is any reduction from the regular price. A promotion is a temporary reduction intended to drive traffic or volume, after which the price returns. A markdown is a permanent reduction used to sell through remaining stock, and it does not go back up.

How deep should a markdown be?

Deep enough that expected recovery beats the value of holding or disposing of the stock, and no deeper. For terminal stock the comparison is against disposal cost, not against full price, which is why correct markdowns on genuinely dead stock look aggressive.

When should a markdown start?

As soon as the projected sell‑through at the current price falls short of the stock on hand within the remaining selling window. That is a calculation, not a calendar date, and it is usually earlier than it feels.

Does markdown optimisation work in food retail with short shelf life?

Short shelf life makes it more valuable, not less, because the remaining selling window is measured in days and the cost of acting late is a write‑off rather than carried stock. It requires accurate expiry data at store level, which is the main implementation hurdle.

Send us three months of price, sales and write‑off history for one category. We will show you how much of that markdown budget went to products that did not need a discount, and how much value walked out as waste. Talk to an expert about Veritico PRICE.