The SKU-Level Discount Ladder: When to Hold Price, When to Discount, and When to Liquidate

I'll say the unpopular thing first, because most founders I talk to are doing the opposite of it. Almost every sitewide sale I've ever seen was a panic, not a plan. Stock was piling up, the month looked soft, someone reached for a "20% off everything" banner, and the brand quietly handed away contribution margin on the products that were selling perfectly well at full price.

That's the trap. A blanket discount treats every SKU as if it has the same problem, when in reality only a handful of them do.

So here's my take on what a discount should actually be. It's not a marketing event. It's a response to a specific signal from your inventory. And the order you do things in matters far more than the percentage off.

Let me lay out the ladder I'd actually use.

The signal you're responding to is inventory age, not the calendar

The mistake underneath the panic sale is timing the discount to the month instead of to the stock.

Every SKU you carry is a little bet you placed on future demand. The size small of a jean, the dark green colourway, the third scent you weren't sure about. Each one has its own demand curve, and they do not move at the same speed. Some sell through before you can reorder. Some sit.

The cleanest way to read that is days of inventory remaining: how many units you've got, divided by how fast they're selling. Pair that against how long the product takes to reproduce, and you get a rough grade for each SKU. Plenty of cover and steady sales is healthy. A growing pile against slowing sales is a warning. Years of cover on something nobody's buying is a problem you're paying rent on.

That grade, not the date, is what tells you which rung of the ladder a product sits on. Most SKUs in a healthy catalogue should never get discounted at all.

Rung one: sell it at full price with no ad spend

The most profitable sale you'll ever make is the one to a customer you already own, at full price, with zero acquisition cost attached.

This is your email and SMS list. Before any paid discounting, before any banner, the first lever for moving stock is simply telling the people who already like you that the thing exists. No margin given away, no media cost. Pure contribution.

I'd go further. If you can sell a chunk of a product run this way at full price, you do it first, every time, on anything that's even slightly aged. A homewares brand I'm thinking of had a candle line sitting at about four months of cover. Before touching the price, they ran a simple "back in your favourite scent" flow to past buyers of the other scents. A meaningful slice of that excess cleared at full price, off owned channels, at no media cost. The discount conversation got a lot smaller after that.

If you find yourself constantly selling out this way with nothing left over, by the way, that's not a win either. It usually means you under-ordered.

Rung two: warm paid traffic, still at full price

When owned channels have done their work and there's still stock to move, the next rung is warm paid audiences. Past customers, site visitors, people who've already met you. Still at full price.

The logic is about where the margin leaks. The second you introduce ad spend, you're taking a bite out of contribution. Warm audiences are the cheapest paid traffic you have, so they take the smallest bite. You can keep the price intact and still shift volume, because these people don't need a discount to come back, they need a reminder.

Hold the line on price as long as the audience is warm. The discount is a tool you're keeping in reserve, not your opening move.

Rung three: full price at your target acquisition cost

Now you're into genuinely new customers, at the acquisition cost you'd normally accept, still selling at full price.

This is just your normal prospecting. I include it on the ladder because of what it signals: if a product can still pull new buyers at full price and a sensible CAC, it has no business being discounted at all. The aged-stock playbook hasn't kicked in yet. You're selling a healthy SKU to a growing audience, and you should be first-order profitable on every unit.

The discount only enters the picture when the signal changes. When demand at full price stops keeping pace with the stock you're holding.

Rung four: a targeted discount, to a subset, off owned channels

Here's where the discount finally appears, and notice how narrow it is.

When a SKU has aged past comfortable and full price isn't clearing it, the first markdown I'd reach for is small, aimed at a slice of your list, and run with no paid media behind it. The classic version: take people who've bought product A but never B, and offer them B at a modest discount. You're expanding what your existing customers buy, clearing the aged unit, and you're doing it at zero CAC so the margin hit is just the discount itself and nothing more.

This is the rung most brands skip straight past on their way to a sitewide banner. It's also the most accretive discount available to you. Small percentage off, no media cost, aimed only at the people most likely to bite.

The reason to keep paid out of it at this stage is simple maths. Discount the margin and stack a CAC on top, and contribution gets squeezed from both ends at once. Off owned channels, you only pay for one of those.

Rung five: discount with paid, then liquidate

If the stock is still sitting after all that, the inventory is telling you something you need to hear. The demand isn't there at this price.

From here the rungs get less pretty, and each one trades margin for cash. First, a discount pushed through warm paid at a tighter efficiency than usual, so you're still making money per unit while you clear it. Then, when the age becomes a genuine drag on cash, a discount at your normal target CAC, accepting you're probably not profitable on those units anymore. And finally, if nothing's moved it, liquidation: clearing at whatever it takes, taking a loss per unit, to turn dead stock back into cash you can put behind better inventory.

That last step feels like failure. It isn't. A unit gathering dust on a shelf is cash you can't reinvest, and it's quietly deteriorating. Recovering even a slice of it to fund a better bet usually beats holding it out of pride. If you run an apparel catalogue and you actually pull this report, I'd put money on you finding a long tail of SKUs sitting in this zone right now that the business would be better off clearing.

There's a flip side worth naming. The thing that genuinely breaks my heart is a brand running a discount, with paid behind it, on a product that's about to sell out anyway. You're paying to give away margin on stock the market already wants. If anything's running low, the move is the opposite of a discount: tighten the spend and let it sell through at full price.

The variant-level markdown, the way the big apparel brands do it

The other shift I'd push you toward is to stop thinking about discounts at the product level and start thinking at the variant level.

Go and look at how a brand like Nike prices a hero shoe. The core colourways, the black, the white, sit at full price and never go on sale. Then the odd sizes and the stranger colours, the ones with low demand and a bit of stock left, quietly carry a markdown. Same shoe, different price by variant, set by how much demand each one actually has.

That's the model. The slow size and the weird colour are the SKUs to mark down, not the whole line. Discounting the entire product to clear a few stray variants is how you nuke margin on the variants that were selling fine. A homewares range might hold its bestselling neutral at full price and trim only the bold print that overstocked. Apparel might hold the mid sizes and mark down the extremes of the size curve.

It's more fiddly than a sitewide banner. It's also the difference between protecting contribution and donating it.

Where to from here

If there's one habit to take from all this, it's that a discount should always answer a question the inventory asked first. Not the calendar, not a soft month, not a competitor's sale. Which specific SKUs, at which specific ages, are telling you demand has slipped behind supply.

Most brands have never actually laid their catalogue out this way, and the markdown spend hiding in a blanket sale is usually larger than anyone realises. If you'd value a clear read on where yours is leaking, the unit-economics and offer side of a Signal/Noise Audit maps exactly this: which products are quietly carrying the discount load, and which ones you should have been holding firm on all along.