The autumn nobody got wrong

A warm autumn, and nothing else

Weather is not a decision. A warm autumn arrives long after the goods have been paid for. They cannot be sent back. This closing lesson takes a season in which that happened and reads the bill in full. The fashionable answer is to buy less up front and re-order fast. This lesson tests three versions of that answer against exactly what happened.

September 2026 was mild. Prestwold's bedlinen department traded at 74% of plan for eight weeks. It then recovered to 92% of plan from week 9, when the weather turned, and it never made the first eight weeks back. Across the phase the department sold 48,220 units against a plan of 57,000, which is 84.6%.

That is the entire external event. There was no supplier failure, no quality problem, no competitor doing anything unusual and no error in anybody's spreadsheet. What follows is what a 15-week commitment lock does to a department when the weather is 10 percentage points off.

Take the printed line, PW-712 Ellerdine, week by week. Intake is the two Kotri deliveries from lesson 1: 8,400 sets booked in during week 1 and 4,500 in week 6.

Check week 6, then read what it means. 6,180 plus 4,500 less 520 is 10,160. A delivery of 4,500 sets landed into a line that had sold 2,220 in five weeks, and every person in the chain was right to let it. It had been unchangeable since 29 June. Dermot could have refused it at the gate and paid for it anyway.

Sell-through, and say the week. Sell-through is the share of what you received that you have sold. At the end of week 8, Ellerdine had received all 12,900 sets and sold 3,780 of them, which is 29.3% at week 8. At the end of week 16 it had sold 10,130, which is 78.5%. On its own that looks like a perfectly respectable season.

The warehouse filled up because the shops slowed down

Now the department. Prestwold's allocation rule sends each shop five weeks of cover at the rate that shop is currently selling, capped by what the fixture and the stockroom can hold. Soft sales mean the shops draw less, so everything else stays at Sarnesby.

Run the same 16 weeks on the plan and the peak is 117 positions in week 6, which is three under the ceiling. The building was sized correctly for the season Prestwold expected. It was over by 35 positions in the season Prestwold got.

The overflow bill: 100 pallet-weeks at GBP 5.40 is GBP 540.00, and 56 pallet movements out and back at GBP 14.00 is GBP 784.00. Total GBP 1,324.00.

The design that was always going to end at the jobber

Ellerdine was bought as three prints, Cresset, Stipple and Wellow, at 4,300 sets each, because in March there was no evidence to buy them unequally. By week 16 the evidence was overwhelming.

Prestwold's disposal ladder for a seasonal print has two rungs. The January sale takes 40% off, so the ticket becomes GBP 27.00. What the January sale does not clear goes to Ockbrook Trading at GBP 6.50 a set. Ockbrook is a clearance jobber: a trader who buys leftover stock in bulk, cheaply, and sells it on through its own outlets. In the event, 1,600 sets cleared at GBP 27.00, which was Stipple's 400 and 1,200 of Wellow, and Wellow's last 1,170 went to Ockbrook.

Hold those two prices against the landed cost of GBP 18.00, because the whole rest of this lesson turns on them:

• A set cleared at GBP 27.00 still returns GBP 9.00 of gross margin. It is a worse sale, not a loss.
• A set sold to Ockbrook at GBP 6.50 destroys GBP 11.50.

Discounting and destroying are different events, and a department that reports both as "markdown" cannot tell them apart.

What the line actually made

That is an achieved margin of 54.2% against an intake margin of 60.0%. Achieved margin is what the line really made, after everything that was marked down. The 5.8 percentage points between the two is the markdown, and it is the number Tobias will be judged on.

Against the plan of 12,000 sets at full price and 900 cleared at GBP 27.00, which is GBP 564,300 of revenue and GBP 332,100 of gross margin, the line is GBP 57,645 short. It decomposes exactly, and the decomposition is worth doing by hand, because it separates the two kinds of loss:

• 700 sets that were planned to sell at GBP 45.00 and cleared at GBP 27.00 instead: 700 × GBP 18.00 = GBP 12,600.
• 1,170 sets that were planned to sell at GBP 45.00 and went to Ockbrook at GBP 6.50: 1,170 × GBP 38.50 = GBP 45,045.

GBP 12,600 plus GBP 45,045 is GBP 57,645. 1,170 sets, which is 9.1% of the buy, caused 78% of the shortfall, because they went down the second rung of the ladder rather than the first.

The rest of the department did not escape either, in a quieter way. Marlbeck finished with 6,690 sets against a planned 3,000, and Denbray with 5,620 pairs against 2,400. Neither is marked down, because plain bedlinen sells all year. But 3,690 extra sets and 3,220 extra pairs is GBP 60,018 of Prestwold's cash carried into a spring phase that had its own buy already committed.

GMROI is gross margin return on inventory investment: the gross margin earned for each pound of stock tied up. Average stock is the mean of the 16 weekly closing figures, at landed cost. The ratio is for the phase and is not annualised. Denbray comes first here too, on a third of Marlbeck's money. That is the same answer the shelf gave in lesson 2, reached through the bank account instead of the fixture.

Wellow was profitable, and it was the worst thing in the department

Read Wellow on its own.

Wellow made GBP 49,455, an achieved margin of 39.0%. Nobody looking at that line on its own would call it a failure, and a range review that reports margin by line will not flag it.

Now read it the way lesson 2 taught. Wellow is 4 SKUs at 0.34 m, so 1.36 m in every shop. 74% of its margin came through the shops, so that is GBP 36,596.70 of store gross margin against 1.36 × 54 × 16 = 1,175.04 metre-weeks. That is GBP 31.15 a metre-week, on a fixture where Denbray runs at GBP 212.56 and even Marlbeck, the worst of the three lines, runs at GBP 43.83.

Split it by store grade and it stops being an argument. Prestwold grades its shops A, B and C by volume: 12, 24 and 18 shops.

Eighteen shops held 1.36 m of a 12.90 m bedlinen module for 16 weeks and sold nine or ten sets each. That is the loss, and no markdown report contains it.

Three speed models, priced against what happened

Everything so far is history. Now spend the money differently and see whether it helps. All three options below leave the plain and pillowcase lines exactly as they were, and change only the printed line.

Option 2 — half the line responsive. 6,400 sets from Kotri for the launch, and cloth for 6,500 reserved at Hardacre with three call-off windows. Freya calls 2,400 in week 4, 1,000 in week 8 and 900 in week 12, putting every call into the prints that are actually selling, so Wellow is never topped up. Buy 10,700. The line still sells 10,130 at full price and finishes with 570, which the January sale takes at GBP 27.00.

Option 3 — the middle rung on one delivery. Move the week-6 delivery to Honaz and to week 10, which makes it changeable in trading week 4. At 74% Freya cuts it from 4,500 to 1,200. Buy 9,600. This one runs out: the line sells 9,600 and turns away 530 sets of demand in the last weeks.

Option 4 — two prints, the third reserved. Launch Cresset and Stipple only, 4,300 each from Kotri, with cloth for 4,300 reserved at Hardacre in case a third print is wanted, or more of the first two. On the evidence, 500 are called in week 4 and 400 in week 12, both of Cresset. Assume 60% of the customers who bought Wellow buy one of the other two prints instead, which puts full-price demand at 9,358. The buy is 9,500 and 142 are left.

Two figures the table below needs and the story has not yet given. The base cloth Hardacre holds costs GBP 8.40 a set. Prestwold's cost of money is 8.0% a year, which is what it pays to have cash tied up in stock rather than in the bank. Cloth reserved and never printed is charged here for the 20 weeks it waits before the spring programme can use it. And the booking fee is GBP 0.50 on every reserved set, called or not.

Every speed model loses money. That is not a mistake in the table, and it is the most useful thing in this course.

Why, in one line of arithmetic

Take option 2 and reduce it to what happens to a single reserved set. Prestwold reserved 6,500. It called 4,300 and avoided 2,200. Of the 2,200 it avoided, only 1,170 would have gone to Ockbrook. The other 1,030 would have cleared at GBP 27.00.

• Every set actually called costs GBP 4.90 more than Kotri would have charged.
• Every avoided set that would have been destroyed saves GBP 11.50.
• Every avoided set that would have been discounted forgoes GBP 9.00, because a January clearance sale at GBP 27.00 is still profitable on a GBP 18.00 cost.
• Every reserved set carries a GBP 0.50 booking fee whether it is called or not, and the cloth for the sets never made sits at Hardacre earning nothing.

4,300 of 6,500 called is 66.2%, so the premium costs GBP 3.24 a reserved set. Of the 33.8% avoided, just over half would have been destroyed, so the saving is worth GBP 0.64 a reserved set. Take off the booking fee and the cost of money on the cloth, and each reserved set is worth minus GBP 3.19. Times 6,500, that is the GBP 20,703 in the table.

A responsive rung pays for the units it stops you buying that would have been destroyed. It costs you money on the units it stops you buying that would merely have been discounted, and on every unit you end up calling anyway.

Test the rule by removing the thing that makes it fail. Suppose Prestwold had no January sale channel above cost, and every leftover set went to Ockbrook at GBP 6.50. Then all 2,200 avoided sets save GBP 11.50, which is GBP 25,300. Against GBP 21,070 of premium on the 4,300 called, GBP 3,250 of booking and GBP 568.62 of cost of money, option 2 comes out GBP 411.38 ahead. Barely. And that is the best case for buying speed at Prestwold's margins.

This is why "we should be more agile" is not a strategy and an arithmetic is. A 60% intake margin and a clearance channel that sells above cost together make over-buying cheap, and anything you pay to avoid over-buying has to clear that bar first.

So what was the answer

Not speed. Allocation, which is what this track is about.

Wellow earned GBP 80.65 a metre-week in the 12 grade-A shops and GBP 11.58 in the 18 grade-C shops. Ranging it in grade A only would have freed 1.36 m in 42 shops from week 1. That is 42 × 1.36 × 16 = 913.92 metre-weeks on the best fixture in the department. It would also have roughly halved the Wellow buy without touching the two prints that worked.

The catch is the one lesson 1 opened with. In March, Wellow is indistinguishable from Cresset. That is why they were bought equally, and buying them equally was right. The grade-A-only decision needs evidence that does not exist until about week 4.

That is what a fast rung is actually for at Prestwold, and it is not what the option tables above were testing. Not "buy fewer sets of a print that will be marked down", which the arithmetic says is barely worth paying for. Rather: launch all three prints in the 12 grade-A shops, hold cloth at Hardacre, read four weeks of real sales, and roll the winners out to grades B and C in week 7 on a three-week lock. The intake calendar has to be redrawn for that lock, as lesson 1 showed, or the whole thing collapses back into the season we just read.

The mistake nobody made

Go back through the autumn and find the decision to criticise.

1. Tobias bought three prints at 4,300 each. There was no evidence to buy them unequally, and inventing some would have been worse.
2. Freya phased the intake into four deliveries in weeks 1, 5, 6 and 10. Those are the weeks a container ordered in May and June arrives.
3. Dermot booked all four deliveries into the slots the routing guide allowed and received them clean.
4. The allocator sent five weeks of cover at each shop's own selling rate. That is a good rule, and it is precisely why Sarnesby filled up as trade slowed.
5. Nobody took an early markdown. There was nothing to mark down in September, and taking one would have given away margin on sets that later sold at GBP 45.00.
6. September was warm. That is not a decision.

The department finished GBP 57,645 down on one line, with GBP 60,018 of cash carried into spring on the other two, and 1,170 sets sold for GBP 6.50 against a GBP 18.00 landed cost. And there is no meeting in the calendar at which somebody should have said something different. The season was decided by a lock date on a supplier record and a bay width on a store plan, and nobody in the range meeting had either of them in front of them.

That is the argument for putting them in front of somebody. Not a faster supplier, not an agile programme, not a better forecast. Two numbers on the page where the buy is signed: the last week this quantity can change, and what a metre of this fixture earns.

Act as a supply-chain analyst with no attachment to the word agile. I am being offered a faster supply route for [LINE OR PROGRAMME]. My baseline supplier lands a unit at [AMOUNT] with a commitment lock of [NUMBER] weeks. The fast route lands a unit at [AMOUNT] with a lock of [NUMBER] weeks, and charges [AMOUNT] per reserved unit as a booking or capacity fee whether or not I call it. My phase buy is [NUMBER] units. Last season, of the units I had left at the end of the phase, [NUMBER] were discounted at [AMOUNT] and [NUMBER] were sold below cost at [AMOUNT] - if I have not given you that split, ask for it and stop, because it is the whole answer. My cost of money is [PERCENT] a year. Do the following. First, work out the loss avoided per unit I do NOT buy, separately for a unit that would have been discounted and a unit that would have been destroyed - and note that a discounted unit above cost is a margin FORGONE, not a saving. Second, work out the cost per unit I do call through the fast route, and the fee on every reserved unit whether called or not. Third, express the whole thing as a value per RESERVED unit given a called fraction f and a destroyed share of the avoided units, and solve for the break-even f. Fourth, apply my own history to that formula and give me a single number: what one reserved unit is worth, and what the whole reservation is worth. Fifth, if the answer is negative, do not stop at no - tell me the largest reservation that WOULD pay, and tell me whether the case for speed at my margins is markdown at all or whether it is space, cash or a later allocation decision. Sixth, if any of my leftover stock is held rather than disposed of, price the cash it ties up and the space it occupies separately, and say which of the two is bigger. Show the arithmetic at every step and do not round a break-even into a comfortable answer.