Fashion Brands Plan to Consolidate Sourcing Over Next Two Years, Benchmarking Study Finds

The Analysis

The finding in this year's benchmarking study, published on 29 July, fits on one line: apparel and footwear brands intend to spend the next two years using fewer suppliers, not more countries. I want to read that line twice, because the two halves pull against each other, and which half you act on decides whether your sourcing base comes out stronger or just comes out brittle. The brands surveyed are keeping the geographic spread they already have. What they are cutting is the number of vendors inside it. Volume moves to fewer, better-vetted factories. Flexibility and compliance rigour move to the top of the list. The stated reason will not surprise anybody who works in a sourcing office: tariff regimes and forced-labour enforcement are tightening, and the cost of not being able to prove where your own goods came from has gone up sharply.

I read that as a buyer, which means I read it as a cost line before I read it as a strategy. Every vendor I keep costs me an audit. Every vendor I drop costs me an option. The study says the industry has decided the audits are now worth more than the options, and on the enforcement side I think that is right. I have watched one detention turn into a six-week hole in a season plan, and no amount of vendor count saves you from that. But consolidation is a decision about concentration, and concentration has a price that never appears in a benchmark. It appears on the day a factory carrying a quarter of your volume stops shipping. So before I cut a single vendor I want two numbers side by side: what I save by having fewer, and what I am exposed to once I do.

One thing up front, because the study does not hand me figures I can lean on and I am not going to invent any. It gives me a direction, which is fewer vendors, same countries, two years, and it gives me the reason. Everything below is my own arithmetic on my own assumptions, and I have written the assumptions out so you can drop your own numbers in and see whether the conclusion still holds. On the tariff side specifically, I do not have reliable rates for the Section 301 forced-labour tariff or for what the expiry of the Section 122 levies does to a given line, and I am not going to guess in print. What I know is the shape of it: both change the landed cost of Asian-origin garments, and both change it differently by country and by mill. That is precisely why the number of vendors you hold in each country matters more this year than it did last.

Let me take the cost structure apart, starting with the saving, because it is the easiest number to produce and the easiest to over-claim. Assume a complete compliance file on one vendor costs twelve thousand dollars and has to be refreshed every twelve months: social audit, forced-labour traceability back to the raw material, and origin documentation good enough to survive a detention. Assume you are buying from forty active vendors today. Forty times twelve thousand is four hundred eighty thousand dollars a year, spent purely on keeping your own book defensible. Assume the consolidation programme takes you to twenty-four vendors inside the two-year window, with the average vendor moving from two and a half percent of your spend to roughly four point two percent. Twenty-four times twelve thousand is two hundred eighty-eight thousand. The saving is one hundred ninety-two thousand dollars a year, and it lands straight on the bottom line.

Now the other number, and this is where I stop and argue with myself. Assume your largest vendor, once the programme runs, carries twenty-two percent of annual spend. Assume that vendor is held for six weeks on a forced-labour finding, not because they did anything wrong but because a sub-tier supplier did and the paperwork cannot clear it fast enough. Assume that country runs at twenty thousand units a week for you. Six weeks is one hundred twenty thousand units of exposure. Assume your backup capacity absorbs seventy percent of that within two weeks, which is generous in my experience; the remaining thirty percent, thirty-six thousand units, is simply gone.

At twenty-two dollars of gross profit a unit, that is seven hundred ninety-two thousand dollars. One hundred ninety-two thousand saved against seven hundred ninety-two thousand at risk. One detention eats four years of audit savings. That ratio is the whole argument, and it is why I will not approve a consolidation plan that has only a cost column. What you do inside those six weeks at the destination end is Nadia Pryce's ground, not mine; I only know what the hole costs.

Which brings me to the part of the study finding I think buyers are most likely to skip. Brands are keeping the geography they already built. Read as a buyer's instruction, that says: cut vendors within a country, not countries out of the list. Hold four origins and go from ten vendors each to six, and you have cut the audit bill by forty percent while the worst-case single-factory exposure stays bounded by what one country carries. Hold four origins and drift to three countries with eight vendors each, and a country-level disruption, whether a tariff change, an enforcement sweep or a logistics shutdown, now touches forty percent of the book instead of twenty-five. Nobody decides to do the second thing. It happens because the best-vetted vendors happen to sit in the same place. Concentrating volume and concentrating geography are different decisions, and the difference between a stronger base and a brittle one is whether you knew which one you were making.

What goes on the scorecard, concretely, is four things with weights: documentation quality and retrieval speed, landed cost on the four-line basis I come to below, flexibility expressed as MOQ and lead time, and capacity headroom in the country. Weight the first at no less than twenty-five percent, because it is the one that decides whether a shipment clears. Anything that cannot be scored goes into a notes field and gets read by a human before the cut list is finalised, which sounds soft and is the single cheapest control available to you.

Flexibility is the second priority named in the study, and it is the one I would actually pay for, because unlike the audit saving it shows up inside the selling season rather than in overhead. Assume flexibility means two measurable things: minimum order quantity comes down and replenishment lead time comes down. Assume the current MOQ is one thousand two hundred units and replenishment lead time is forty-five days. Assume the consolidated vendors, because they are getting more of your volume, will sign six hundred units at thirty days. Assume a sixteen-week season and clearance at forty percent off the ticket.

Here is what that is worth. Assume season volume of four hundred thousand units at forty-nine dollars retail and twenty-seven dollars unit cost, so twenty-two dollars of gross profit at full price. Assume clearance at forty percent off, which is twenty-nine forty, leaving two dollars forty: every unit that fails to sell at full price costs me nineteen dollars sixty of margin. Assume your pre-season forecast runs at seventy-two percent accuracy and your in-season read at eighty-eight percent, and assume the shorter lead time lets you move fifteen percent of the season out of the pre-season guess and into the in-season read.

Fifteen percent of four hundred thousand is sixty thousand units. Guessed, at an eighteen percent markdown rate, ten thousand eight hundred units go to clearance, and at nineteen sixty each that is two hundred eleven thousand six hundred eighty dollars. Read into the season instead, at six percent, three thousand six hundred units go to clearance, which is seventy thousand five hundred sixty dollars. The difference is one hundred forty-one thousand one hundred twenty dollars, and that is what fifteen days of lead time is worth on one season of one programme.

Put the two together and the consolidation programme is worth three hundred thirty-three thousand dollars a year to me, one hundred ninety-two thousand of avoided audit spend plus one hundred forty-one thousand of avoided markdowns, set against a seven hundred ninety thousand dollar downside if I execute it badly. I take that trade, but I take it with my eyes open, and I take it only because the downside is one I can buy down. Backup capacity is the thing that turns seven hundred ninety thousand into something survivable. Without it, the same programme is a bet that nothing goes wrong for two years, and I have never seen a sourcing book that earned that bet.

There is an opportunity side too, smaller than people assume but real. Concentrating volume buys you things beyond a lower audit bill: priority when a mill is running tight, better payment terms, and a supplier willing to hold fabric or key trims against a forecast rather than against a firm order. Assume the last one is worth something genuine. Assume holding greige or critical trims against a forecast cuts replenishment lead time by a further seven days on top of the fifteen. On the arithmetic above, fifteen days was worth one hundred forty-one thousand. Seven more days is not linear and I will not pretend it is, but even at half the rate it is another seventy thousand, and it costs nothing except a conversation. Ask for it in the same negotiation where you commit the volume, because that is the only moment you hold any leverage to ask.

Now the tariff part, where I want to be careful, because this is where most sourcing decks quietly break. The study ties the push to consolidate to tightening tariff regimes and forced-labour enforcement, and it names two specific things reshaping landed cost on Asian-origin garments: the Section 301 forced-labour tariffs, and the expiry of the Section 122 levies. I am not going to quote you a rate for either. I do not have numbers for this season that I would defend in front of finance, and a sourcing decision built on a rate somebody half-remembers is worse than no decision at all. What I am sure about is what to do with the uncertainty.

Landed cost gets broken into four lines: FOB, tariff and related levies, freight, and the amortised annual compliance cost per vendor. Any vendor quote arriving with fewer than four lines does not enter the comparison. That rule sounds bureaucratic and it is not. When tariff treatment is moving, a cheap FOB is not a cheap landed cost, and the only way to see the difference is to force the tariff onto its own line where somebody has to own it. If you cannot tell which country's treatment is doing what to the number, you are comparing FOB values and calling it sourcing.

The second thing to do with the uncertainty is to stop letting it be averaged away. If the book has four origins and the tariff treatment differs across them, an average landed cost is a number that will cost you money in exactly the season you most need it to be right. I want landed cost quoted per origin and per fabric, refreshed whenever the treatment changes, with the date of the last refresh printed on the sheet. I do not trust a landed cost figure that cannot tell me how old it is, and in a year like this one a landed cost number without a date on it is not a number I will sign against.

Two years in a study is a finding. On my desk it has to become a calendar or nothing happens. Assume the study landed on 29 July. That leaves the rest of this year to build the vendor scorecard, the first half of next year to move the volume, and the season after that to find out whether any of it was right. The trap inside a two-year programme is that the savings arrive at the end and the risks arrive in the middle. The audit money does not come back until vendors are actually off the list. The concentration risk starts the day the first vendor's share goes up. Sequencing matters more than the target, and the sequence I would hold to is simple: qualify the backup before you grow the incumbent, and never let a vendor's share rise ahead of the backup's readiness.

Timing inside the window matters more than the window itself. Do not cut vendors in the eight weeks before a season launch. The audit saving accrues monthly while the disruption risk spikes during cutover, and if a vendor's last order overlaps a launch, the knowledge loss and the qualification gap both land at the worst possible moment. I schedule cutovers in the four to six weeks after a season closes, when the line is quiet and a two-week qualification delay costs me a sample rather than a shelf.

Price is the worst possible basis for this decision. A vendor that wins on FOB and loses on traceability documentation is not cheaper, it is a detention waiting for a date. The study's own answer to that risk is compliance rigour, which means the scorecard needs a documentation score on it with real weight rather than a tick box. If it does not have one, consolidating moves you into precisely the exposure you were trying to buy down.

The country column needs watching as closely as the vendor column. Set an intended share per origin and review it monthly while the programme runs. If one country drifts more than five points above its intended share, stop and rebalance before the season starts rather than after, because rebalancing mid-season costs money you did not budget.

Then there is the MOQ promise that never made it into the contract. Six hundred units discussed over dinner is worth nothing in the week you need it. It has to be a number in the agreement, with the lead time beside it and the unit price at that quantity stated in the same clause, because vendors will quote a low minimum and a high unit price and the two together are frequently worse than the deal you already had. I check the landed cost at the new minimum, not the minimum itself.

And the audit file itself. A compliance file that lives at the vendor, in their format, on their system, is not evidence you can produce in seventy-two hours when a shipment is held. I hold a copy, in my format, refreshed inside twelve months, and I test retrieval once a year by asking for one at random and timing the response. A file you cannot produce quickly is a file you paid for and do not have.

The last trap is the one nobody scores. Cutting a vendor you have used for eight years costs you institutional knowledge: who runs long at the dyehouse, which trim supplier is the real bottleneck, what the mill does when a colour comes off. None of that appears on a scorecard and all of it turns into lead time the moment it disappears. Before the final order goes out, get the process notes in writing from somebody who has actually stood on that floor.

If I were signing this, here is what I would sign. I would consolidate, because the study is right that the compliance burden is now the binding constraint and I would rather carry twenty-four vendors' worth of audits than forty vendors' worth of exposure to a file I cannot produce. I would consolidate inside the geography rather than across it, and I would not let any single vendor's share cross twenty-five percent of annual spend without a qualified backup already activated somewhere else. I would price the programme at three hundred thirty-three thousand a year of benefit against a seven hundred ninety thousand downside, and I would accept that trade only with the backup capacity in place. I would write flexibility into the contract as three numbers, MOQ, lead time and the price at that MOQ, rather than as adjectives. And I would refuse to compare vendors on anything except four-line landed cost with a date on it.

Two years from now the buyers who got this right will not be the ones with the fewest vendors. They will be the ones who could say, on any given day, what fraction of the book sat behind one factory gate and what it would cost if that gate stopped. That number is knowable this afternoon. Most books I have looked at cannot produce it. Getting to the point where you can is the actual job, and everything else in the study is commentary.

↑ Back to the key points

— By Grace Yang