A promotion is not a markdown

The week the two instruments got confused

A weekend sale and a permanent cut to the price on the label both give money away. People treat them as two versions of the same move. They are not. This lesson separates them by what each one leaves behind. The permanent cut changes what every remaining garment is worth, and says so in the accounts. The weekend sale changes nothing about the stock, and quietly teaches the customer to wait.

The reason CF-3140 kept its ticket in week 9 was not that anybody thought the stock was fine. It was that a promotion was already booked for week 10, and a promotion, the room assumed, would do the same job as a markdown.

It gave price away. It sold dresses. It was, by the standards it was set, a success. It removed 160 dresses from a pile of 12,310 and it moved the line's forward cover by 0.8 of a week.

This lesson separates the two instruments. Not by vocabulary: everybody already knows a promotion is temporary and a markdown is permanent. It separates them by the money and by the stock. Those two behave completely differently, and that is where the confusion costs something.

Six differences that are not about wording

Row four is the one that does damage, and it is invisible on a trading sheet.

A markdown revalues the stock the day the ticket changes. Once Callowfield decides CF-3140 sells at 30% off, the 10,590 dresses it still owns are worth GBP 34.42 each in net selling terms, and the accounts say so immediately. A promotion revalues nothing. After four days at 25% off, the dresses that did not sell are still carried at a full-price selling value, because the ticket never moved.

So a department that reaches for promotions instead of markdowns reports a healthier stock position than it has. The units are still there, the customer has been taught what the dress is really worth, and the balance sheet has not been told. Nothing in the weekly numbers will show you this. You have to know you are doing it.

The Spring Dress Event, week 10

Four days, Thursday to Sunday, 25% off every dress in the chain and online, supported by an email and a homepage takeover. Lines already marked down were excluded, which is the ordinary rule and which matters here: it is why CF-3140, still at full price, was in the event at all.

Callowfield's own day-weighting puts Thursday to Sunday at 275 of a normal week's 400 dresses on a line like this one. Here is what the four days did, and the week that followed.

The event sold 520 dresses in four days against a baseline of 275, and the two weeks together sold 160 more dresses than they would have. Both of those are real. Now take the 520 apart, because they are not one thing.

Two of those three rows are the event paying for sales it already had. The 275 are the ordinary substitution any discount makes. The 85 are the specific thing a short promotion does that a markdown does not: it moves a purchase in time. Week 11 sold 385 against a baseline of 470, and the missing 85 were standing in a Callowfield store the previous Saturday.

Cost per dress actually removed

Both instruments spend money to shift stock. Divide what each spent by the stock it shifted.

The event was not a mistake. It made GBP 964.40, and a promotion that makes money is a promotion worth running.

It also cost GBP 27.65 for every dress it took off the pile, against GBP 16.01 for the instrument that was not used. And it took off 160 dresses out of 12,310. So the markdown was not a better version of the promotion. It was a different size of thing entirely. The promotion's real cost is not on this table at all, because what it actually bought was the argument for holding the ticket another four weeks.

The cover test, which settles it in one line

Cover is the number that tells you whether an instrument has changed the situation. Here it is either side of the event, measured the honest way — closing stock against the forecast for the weeks ahead, which for weeks 12 to 26 is 189 dresses a week.

A four-day chain-wide promotion, an email to the whole file and a homepage takeover moved this line's cover by eight tenths of a week. There were fifteen weeks of season left and sixty weeks of stock. That is not a criticism of the event, which was doing its own job for the whole department. It is the answer to whether it could stand in for a markdown on this line, and the answer is not close.

The markdown, by comparison, does not move cover by a fixed amount — it changes the rate for every week that follows. Weeks 9 to 13 at 20% off would have left 9,360 dresses at the end of week 13 instead of 10,590, and every week after that would have been running from a lower base. That is the difference between the two instruments in one sentence: a promotion is a withdrawal, a markdown is a change in the interest rate.

Why the wrong one keeps getting chosen

Three reasons, and none of them is stupidity.

1. A markdown has a visible, budgeted cost and a promotion appears not to. Re-ticketing 12,740 dresses across 96 stores at GBP 0.21 is GBP 2,675.40 of labour. Somebody has to authorise it, in a week when the stores are already short-handed. The event's discount is far larger and nobody signs for it, because it never appears as a cost — it appears as a lower sales figure.
2. A markdown is an admission and a promotion is an initiative. The first says the buy was wrong; the second says marketing is driving trade. In a room, those are not the same sentence, and they are being said by different people.
3. A promotion is reversible. That is a genuine advantage and it is the reason it gets reached for in a week when nobody is sure yet. The trap is that the weeks in which nobody is sure are exactly the weeks in which the markdown is cheapest, because there is still enough season left for the extra rate of sale to do some work.

What a promotion is genuinely good for

This is not an argument that promotions are a bad instrument. It is an argument that they are a different one. A promotion earns its place when

• the category is trading acceptably and you want a week of it, not a permanent price change;
• there is a defined pile with a hard end date — an end-of-line, a colourway, a size block — and the promotion can be aimed at exactly that and stopped;
• newness needs an audience, and the discount is buying traffic for the full-price lines standing next to it;
• the price change genuinely should be temporary, because the reference price is worth protecting.

That last one is a legal point as well as a commercial one. In several markets the "was" price in an advertised reduction is set by law, not by marketing. The law defines it as the price you genuinely charged over some period beforehand. So a line that lives on promotion eventually loses the right to call itself reduced at all. Spend a promotion deliberately. Do not let it leak away.

What lesson 2 leaves you with

Two instruments, and the one that was used removed 160 dresses. Five weeks later the plan's 30% step came in on schedule and the season ran its exit curve to the end: 30% from week 14, 50% from week 19, 70% from week 23.

It ended with 3,060 dresses in a warehouse. Lesson 3 is what those dresses were worth, what it cost to keep them while somebody decided, and what the answer does to the bill for the five weeks in lesson 1. That bill turns out to be smaller than lesson 1 said, for a reason nobody could have known in week 9.

Act as a sceptical retail planner comparing two instruments on one over-stocked line. Facts: [CODE], ticket [AMOUNT], sales tax [RATE], landed cost [AMOUNT], stock on hand [UNITS], weeks of season left [NUMBER], forecast sales for those weeks at full price [UNITS]. Option one is a permanent markdown of [DEPTH] from week [NUMBER] to the end. Option two is a promotion of [DEPTH] for [DAYS] days in week [NUMBER]. My measured rate-of-sale multiple at that markdown depth is [NUMBER]. My own day-weighting says the promotion's days carry [UNITS] of a normal week on a line like this. My re-ticketing cost is [AMOUNT] a unit across [STORES] doors. Do the following, and keep the two instruments strictly separate. First, for each option, tell me how many units it removes that would NOT otherwise have sold in the season, and ask me what an unsold unit eventually fetches rather than assuming it. Second, for the promotion, split the discounted units three ways: substituted, pulled forward from the following weeks, and genuinely incremental — and say which of the three you have had to estimate. Third, give me total money spent for each option, including re-ticketing labour for the markdown and nothing at the till for the promotion. Fourth, give me cost per unit actually removed for each, and net value per unit removed. Fifth, give me forward cover at the end of the affected weeks under each option and under doing nothing, all three measured against the forecast for the weeks ahead. Sixth, state in one sentence whether the promotion can substitute for the markdown on this line, and if it cannot, say by what factor it falls short in units removed. Do not judge either instrument on its sales figure alone, and do not tell me a promotion succeeded because it sold well.