Adding, consolidating and leaving: the base over time

The meeting after the season

Adding a factory to the list you may buy from feels like adding a row to a spreadsheet. Dropping one feels like deleting a row. Both are expensive: in money, in weeks, and in the hours of the people who look after them. Putting a figure on that lets both decisions rest on something other than instinct.

This lesson also weighs putting more work through fewer factories, and how to stop buying from one without lasting damage.

18 November, the outerwear post-season review. DH-7420 sold through, late and at a worse margin than the plan. Two proposals are on the table.

The first is from Verhoek: add a proper padded-outerwear factory to the base, so that the next time Balteem is full there is somewhere to go that has made the category.

The second is from the sourcing director: consolidate. Duinhoven's outerwear is spread over four suppliers, none of whom gets the volume that would earn a real price. Move it to two and take the discount.

Both proposals are sensible. This lesson prices them, and then prices the thing neither of them mentions.

What a new supplier costs before the first piece ships

Buyers talk about adding a supplier as though it were adding a row to a list. Here is Duinhoven's own reconciliation of what it cost, the last time it did it properly.

USD 19,046, and 21 weeks from the first visit to goods on a vessel. That number reframes three things at once.

Your approved list is an asset with a book value. Twenty-two suppliers at roughly USD 19,046 each is about USD 419,000 of spending that has already happened.

Dropping a supplier is spending it again. Not the whole sum, because a re-approval is cheaper than a first approval, but not far off. Duinhoven's own experience of restarting a lapsed supplier is about 60% of the original, near USD 11,428, and 13 weeks.

And the five dormant suppliers from lesson 1 are the most expensive state in the base. Around USD 95,230 was spent to acquire them, and none of it is producing anything. Worse, the asset decays. The audit lapses. The price list goes stale. The merchandiser who knew you leaves. A supplier is only an option while it is warm.

How many suppliers can one team actually hold

There is a real ceiling, and almost nobody measures it. Duinhoven timed it. An active supplier consumes about 6.5 hours a week of a merchandiser: the mailbox, the weekly production report, sample chasing, approvals, one call, and the small emergencies. A Duinhoven merchandiser has about 24 hours a week left for supplier management, after range work, sample reviews and internal meetings.

That is 3.7 suppliers each, and just under 15 for a team of four.

Duinhoven has 17 active suppliers and 22 approved. The base is bigger than the team that runs it.

That is a completely different diagnosis from "we have too many suppliers", and it has a different set of answers: hire, automate the reporting, or shrink. Choosing without the number is how a base ends up at 22 by accident.

Consolidation, priced

The sourcing director's proposal is arithmetic, and good arithmetic.

A volume break is a discount a factory gives once your yearly quantity crosses a stated line.

Balteem's break is worth USD 89,936 a year and Tarkhan's USD 15,500. So the consolidation earns USD 105,436 a year on the same units. It also halves the number of relationships the team has to hold, which the ceiling above says is worth something real.

Now look at what the base becomes. Balteem goes from 73.1% of Duinhoven's outerwear to 80.3%, and Duinhoven goes from 60.1% of Balteem's capacity to 66.1%. Everything lesson 1 said about that quadrant gets more true.

What it gave away, and the honest version

The following March, Balteem was three weeks late on a spring outerwear style of 26,000 pieces. The cause was fabric: its nominated mill missed a dye lot. 4,800 units missed the launch and sold at the first markdown. At USD 74.95 retail and 20% off, that is USD 14.99 a unit and USD 71,952.

The obvious lesson is that consolidation removed the alternative. Girga had been narrowed out of outerwear. Meltem's line had been sold to another buyer, because Duinhoven had told them there was nothing this season. And Tarkhan was full at its new 50,000.

That obvious lesson is wrong, and it is worth being precise about why. The constraint in March was fabric, not sewing capacity. A second factory with free lines and no fabric rescues nothing.

Name the risk you are actually insuring against. Concentrating at a factory concentrates capacity risk, quality risk, compliance risk and solvency risk. It does not concentrate fabric risk. Fabric risk lives at the mill, and it belongs to a different base.

So price the option properly, against the slips a second factory could have covered. Duinhoven's last six seasons carried three slips large enough to reach a markdown.

The rescuable loss is USD 102,500 across three years, about USD 34,167 a year.

What the option costs, and the sentence that decides it

Here is the neat part. Keeping Meltem alive does not cost a premium per piece. It costs a threshold.

If Meltem keeps 9,000 pieces a year, only 7,000 move to Tarkhan. Tarkhan lands at 41,000 and misses its 45,000 band entirely. A volume break is a cliff, not a slope. So retaining Meltem costs exactly Tarkhan's break, USD 15,500 a year, and the consolidation still earns USD 89,936.

Against a rescuable loss of USD 34,167 a year, that looks like a ratio of 2.2, and an easy decision. It is not that easy, and the honest number is worth more than the flattering one.

Meltem is small and slow to start. On its own free line-days it could have absorbed perhaps 7,600 pieces at short notice, not 26,000. Assume it recovers 60% of the rescuable loss. The option then returns about USD 20,500 a year against USD 15,500. A ratio of 1.3. Worth doing, and only just.

Which is why the last sentence of this section matters more than the ratio.

Keep an alternative warm on the work it is best at, or the option stops paying. Give Meltem its 9,000 pieces as in-season repeats and small colour drops, work Balteem would decline and Tarkhan would price at over USD 28, and the option costs USD 15,500. Give Meltem the same 9,000 pieces as a slice of a big single-colour style, the shape lesson 2 showed it is worst at, and you add a premium of USD 1.90 a piece over Balteem, or USD 17,100 a year. That is a total of USD 32,600 against a return of USD 20,500.

Same supplier, same volume, same intention. The decision flips from sensible to indefensible on the shape of the work.

Leaving, and the two ways to do it

Duinhoven did not exit Girga. It narrowed it. Girga stays in the base for shirting, which it is genuinely good at, and comes out of outerwear.

That distinction is most of good pruning. A narrowing keeps a USD 19,046 asset alive at almost no cost, and takes away the thing that went wrong. A full exit throws the asset away, and has to be done properly, because a supplier relationship does not end when you stop ordering.

Start with what is physically yours at their site, on the day you decide. At Girga, on 18 November: 1,940 metres of shell fabric at USD 2.20 a metre, USD 4,268; 6,200 unused zips at USD 0.41, USD 2,542; and 21,000 printed care labels and 19,000 polybags carrying the Duinhoven name, USD 1,113. That is USD 7,923 in total, of which USD 6,810 has recoverable value and USD 1,113 has none.

It is the worthless USD 1,113 that carries the risk. Labels and packaging with your name on them, left at a site you no longer approve, will end up on a garment you did not make. Have them destroyed, and ask for the certificate.

Then run a sequence, not a phone call.

1. Decide narrowing or exit, and write the reason down. If the reason is one order, it is probably not a reason.
2. Place the replacement work before you give notice, so the alternative has shipped one full order before the incumbent's last one.
3. Give notice in writing, on whatever your own supplier terms require, and say plainly whether it is a narrowing or an exit. Ambiguity here costs you a supplier who would have taken you back.
4. Settle open liabilities: committed fabric and trims bought against your orders, work in progress, and tooling you paid for.
5. Recover or destroy your property, and get the certificate for anything carrying your name.
6. Leave the record accurate. A supplier marked inactive with the true reason is worth something to whoever reads it in three years. A supplier deleted is a lesson deleted.

When the supplier has outgrown you

The exits people find hardest are the ones where nothing went wrong.

Duinhoven had a supplier that won a very large account and grew. Duinhoven's share of its capacity fell from 34% to 6% over three seasons. Nobody's attitude changed, nobody was rude, every meeting was cordial. And on-time delivery from that site fell from 94% to 61%.

That is lesson 1's arithmetic running in reverse. Priority is a queue, and the queue is ordered by minutes. When you become a rounding error to a supplier, you get rounding-error service. No amount of relationship work reverses it. You can only add minutes, or leave.

Albert Hirschman's Exit, Voice, and Loyalty, from 1970, is about exactly this choice. When something declines, you can complain or you can leave, and loyalty is what delays the leaving. His unwelcome point is that loyalty can keep you complaining long past the moment leaving was the cheaper option. Course 27.5 covers how the on-time number that shows you this is actually built.

Act as a sourcing analyst who is neither for nor against consolidation, and who reports numbers. I am considering consolidating one category. Today: [FOR EACH SUPPLIER — NAME, PIECES A YEAR, AVERAGE FOB, MY SHARE OF THEIR SEWING CAPACITY, THEIR SHARE OF MY CATEGORY UNITS, THE ORDER SHAPES THEY ARE GOOD AT, FREE LINE-DAYS IN A TYPICAL WINDOW]. The proposal: move [VOLUMES] to [SUPPLIERS], leaving [SUPPLIERS] with nothing. Volume breaks offered: [SUPPLIER, DISCOUNT PER PIECE, THRESHOLD QUANTITY]. My last six seasons of delays large enough to reach a markdown: [SEASON, CAUSE, UNITS AFFECTED, MARKDOWN COST]. Do the following. First, compute the annual saving from the consolidation, break by break, and show the concentration afterwards in BOTH directions for every remaining supplier. Second, classify each historic delay by cause, and say honestly whether a second garment factory with free capacity could have covered it. Treat a fabric or trim failure as NOT coverable, because a second factory with no fabric rescues nothing. Third, total the coverable loss and express it per year. Fourth, price the option of retaining one alternative. Check whether holding volume back drops another supplier BELOW a volume threshold, since a break is a cliff and not a slope, and give me the true annual cost of retention, including any break lost entirely. Fifth, discount the option's value for the alternative's real size, meaning how many pieces it could actually absorb at short notice from its free line-days, and give me both the crude ratio and the honest one. Sixth, tell me what shape of work keeps the alternative warm most cheaply, and what it would cost to keep it warm on the WRONG shape. Seventh, state the case against retention as strongly as you can. Show every calculation. Do not give me a recommendation without the arithmetic under it.

What to take away

1. A supplier costs USD 19,046 and 21 weeks to acquire, near USD 11,428 and 13 weeks to reactivate, and nothing at all to leave sitting on a list decaying. Know your own version of those numbers.
2. Your base has a ceiling set by the hours your team has, not by the market. Measure it before you argue about size.
3. Consolidation earns real money. Price it, then price the alternative it removes, against the slips a second factory could actually have covered, not against every slip.
4. A volume break is a cliff. Find your suppliers' thresholds before you split a volume, or you will give one away for the sake of a few thousand pieces.
5. Keep an alternative warm on its own shape of work. Warm on the wrong shape costs about twice as much, and stops being worth it.
6. Most exits should be narrowings. When it is a real exit, run the sequence: replacement first, notice in writing, liabilities settled, property recovered or destroyed with a certificate, and the record left honest.