fashion.business fashion.business
One write-off, start to finish: what a compliant record actually looks like

One write-off, start to finish: what a compliant record actually looks like

fashion.business Team
21 July 2026
5 min read

When stock is no longer valuable to a fashion business, its is often written-off. The excercise can lead to the destruction of goods, which is now regulated in the EU. What is allowed, and what is not?

Three thousand jumpers come back to a warehouse outside Venlo across about fourteen weeks. They are a women's lambswool-blend crew neck from AW25, sold across seven EU markets, returned at a rate the brand considers unremarkable. They arrive in ones and twos, are booked in against the returns SLA, fail the post-return quality check, and are consolidated onto pallets in the corner of the third-party warehouse everybody calls the write-off bay.

In February the finance team writes them down. In March somebody asks the 3PL to arrange a collection.

From 19 July 2026 that second instruction, given in that form, is unlawful for a large enterprise placing apparel on the EU market. What follows is the same consignment handled the way Article 25 of the Ecodesign for Sustainable Products Regulation and Commission Delegated Regulation (EU) 2026/296 require. The case is fictional and the money figures are illustrative, flagged where they appear. The structure is not.

The central finding is worth stating before the detail. The derogation decision is not a classification applied to a consignment. It is a sort applied to units, and when the cost-effectiveness test in ground (f) is run against the comparator the regulation actually specifies, it disqualifies most of a damaged consignment rather than justifying it. Of these 3,000 units, 880 can lawfully be destroyed. The rest cannot, and the reason is arithmetic rather than principle.


Who actually decides, and when

Two decisions are taken here by two different people, weeks apart, and in most brands neither knows the other is part of a single compliance event.

The first is the write-down: a finance review, expressed in value, carrying a financial reason code such as obsolete or damaged. It is an ordinary accounting judgement and the regulation does not touch it. Nothing in Article 25 prevents a brand from valuing dead stock at zero.

The second is the physical disposition, taken when an operations manager needs the bay back, expressed in pallets, and communicated to the 3PL as a collection request. That is the decision the regulation now governs, and it is taken by somebody with no visibility of the first.

The first structural change is to reverse that order for Annex VII goods. The disposition decision has to be taken, evidenced and recorded before the units move, because the evidence supporting it comes from an inspection that has to happen while the goods are still in front of somebody. A brand that writes down in February and inspects in March has not broken anything. A brand that writes down in February and collects in March has destroyed the evidence for the ground it will later claim.

One precondition sits underneath all of it. Article 25(1) reaches goods that have been placed on the Union market, and these were sold to EU consumers and returned. Article 2(37) puts the rest beyond argument: an unsold consumer product expressly includes products returned under the right of withdrawal in Article 9 of Directive 2011/83/EU, or during any longer withdrawal period the trader offers. The status attaches on return, whatever the inspection later finds.


Working through the grounds, in order

The inspection is carried out at the 3PL against the brand's own quality assessment procedure. It sorts the 3,000 units into three populations.

2,120 units show pilling, minor seam separation at the shoulder, missing swing tickets, or the general handling wear of a garment that has been tried on, boxed and shipped twice. All of it is correctable by a repair and refinishing operation.

640 units were on a pallet that sat under a roof leak for a weekend in January. They carry mould staining through the body panel. Wet-cleaning a lambswool blend at that level of contamination does not return the garment to a saleable condition, and the inspection records that correction is not technically feasible.

240 units carry a dropped-stitch fault across the yoke, present from manufacture and missed at goods-in a season earlier. The structure of the knit is compromised. Repair is not technically feasible.

Now the grounds, in the order the delegated regulation sets them out, because the order is doing work most operations teams will not expect.

Grounds (a) to (e) do not apply. Nothing here is dangerous within Regulation (EU) 2023/988, nothing is unlawful for a non-safety reason, no rights holder has notified anything, no licence has expired, and the logos are ordinary woven labels that come out without difficulty.

Ground (f) covers products damaged, deteriorated or contaminated to the point of being unacceptable for consumer use, wherever in the chain the damage was caused, including by the consumer, where repair or refurbishment is either not technically feasible or not cost-effective. Recital 9 of the delegated regulation puts the consumer-return case beyond argument, contemplating destruction of damaged products including those returned under the right of withdrawal or a longer trader period, where repair is not technically feasible or not cost-effective. That is two limbs, and they are alternatives. The 640 water-damaged units satisfy the first limb on the inspection evidence. The 2,120 cosmetically damaged units can only reach ground (f) through the second, which is where the arithmetic below decides them.

Ground (g) covers products non-functional because of a design or manufacturing defect where repair is not technically feasible. Note what is absent. Cost-effectiveness attaches to ground (f) only. Ground (g) is technical feasibility alone, so a manufacturing defect that could be repaired at a cost nobody would rationally spend is still not a ground (g) case. The 240 yoke-fault units are, because the knit itself is compromised.

Ground (h), donation, is the one the operations team will reach for first and the one they will get wrong. It is not a parallel option sitting alongside the others. Article 2 of the delegated regulation makes it available only where none of grounds (a) to (g) applies. For the 640 and the 240, grounds (f) and (g) apply, so ground (h) was never available for those units, whatever donation attempt was made and however it went. This sequencing is the single point most likely to be inverted inside a business, because operationally donation feels like the responsible route and therefore the safe one to claim.

For the 2,120, ground (h) does remain available, and section 4.2 of the destruction ban report sets out why it is expensive in calendar time rather than in money: three qualified social economy entities in the Union, or a public listing running a full eight weeks.


The cost-effectiveness sum, done in full

Everything turns on the second limb of ground (f), and on what the comparator is.

Article 1 of the delegated regulation defines cost-effective as the cost of repairing or refurbishing a product not outweighing the total cost of destruction of that product plus the materials, manufacturing, packaging, transport, stocking and any other administrative or logistical expenses of replacing that same product.

Read the second half again. The comparator is replacement. It is not the residual value of the garment and it is not what the garment would fetch in a clearance channel. A brand asking whether repair costs more than the jumper is now worth is answering a question the regulation does not pose.

Here is the sum for the 2,120 units. Every figure is illustrative, chosen to show the shape of the calculation rather than any real garment.

Repair side, per unit: de-pilling and finishing £1.60, seam correction £0.90, press, re-tag and re-bag £0.90. Total £3.40.

Replacement side, per unit: materials £4.60, manufacturing £2.90, packaging £0.35, inbound transport and duty £1.15, stocking and handling £1.05, administration £0.60. Total £10.65.

To that replacement figure the definition adds the total cost of destruction of the product. This piece does not carry a figure for it, because no verified cost exists for the disposal routes in question and inventing one would be worse than omitting it. Omitting it is safe, and instructive, for a reason worth pausing on: the destruction cost is a positive number, so it can only push the comparator higher. The comparator is therefore at least £10.65 and the repair cost is £3.40. Repair does not outweigh it. Repair is cost-effective within the meaning of Article 1, and ground (f) is not available for these units.

Now run the same units through the comparator brands actually use. The residual clearance value of the style is, illustratively, £2.00. Against that number a £3.40 repair looks obviously irrational, disposal looks like the commercially responsible choice, and the write-off file that results claims ground (f) with an inspection record attached and a sum performed against the wrong denominator. It will read as complete. It will fail on the ground it claims, and it will fail for a population of 2,120 units rather than for one.

The 2,120 must therefore be repaired and returned to a commercial route, or offered for donation and only then, if not accepted after the full eight weeks, destroyed under ground (h). Eight weeks is not a formality. It has to begin while the stock is still commercially live, which is the argument Decision 6 of the destruction ban report makes in full.

That leaves 880 units lawfully destructible: 640 under ground (f) on technical infeasibility, 240 under ground (g).


The evidence pack, and why the process-level route is cheaper

Article 3 of the delegated regulation gives nine evidence limbs for ten grounds. Grounds (f) and (g) share one, and it offers a choice.

The first option is documented quality assessment procedures, including sorting that prioritises restocking and repair. The second is an inspection record documenting the type and severity of the damage and the unfeasibility of correction.

A returns operation running thousands of units through a bench should hold the first. The procedure document is written once, versioned, and covers every consignment passing through the process for as long as it stands unamended. The per-unit inspection record has to be produced, stored and retrieved 880 times, and 900 more at the next consignment. Recital 13 of the delegated regulation permits documentation to be prepared collectively where the same circumstances affect multiple products, which is what makes the process-level route workable at volume rather than merely cheaper.

In practice this operation holds both, because the procedure is the standing evidence and the inspection output is the record of applying it. The pack for this event is the versioned quality assessment procedure showing the sort hierarchy, the inspection summary with the three populations and their counts, the cost-effectiveness working for the 2,120 with its inputs identified, and the disposition instruction naming ground (f) for 640 units and ground (g) for 240.

Article 3 requires it to be held electronically for five years after the product has been destroyed, and produced to a competent authority within 30 days of a request. Note the trigger. The clock starts at destruction, not at the write-down, not at the inspection and not at the disclosure.

A second clock runs alongside it on a different class of document. Documentation demonstrating the delivery and reception of discarded products must be kept for five years from the date the disclosure was made, a point stated in the same terms by Cooley, Bird & Bird and Beveridge & Diamond in their 2026 briefings on Commission Implementing Regulation (EU) 2026/2. For this consignment the disclosure falls due within twelve months of the financial year end, so the second clock starts later than the first and runs later. Filing both against the same reference is the only way either is retrievable.


The statement that goes to the waste operator

Article 4 requires the operator to give the receiving waste treatment operator a statement identifying the applicable derogation. It does not prescribe the content and it does not prescribe the format. Nobody should circulate a template as though the regulation contained one.

What a defensible statement carries, on the same reasoning that produced the evidence pack: the identity of the economic operator, the consignment reference used in the brand's own records, the derogation ground relied on for each population in the load, unit count and weight against each ground, the date, and a named contact. Two grounds in one load means two lines, not one statement claiming the higher of them.

Where destruction is lawful, the waste hierarchy in Article 4 of Directive 2008/98/EC still applies, prioritising recycling over other recovery and over disposal, as Recital 3 of the delegated regulation makes explicit. The derogation permits destruction. It does not permit any method of destruction.


When the ground changes halfway through

Suppose the 2,120 go out on the eight-week public listing under ground (h), and in week six a sprinkler discharge damages 300 of them beyond repair.

Nothing in the delegated regulation bars reclassification, and nothing requires an operator to stay with a ground once goods have been earmarked for it. What governs is the actual circumstance at the time of destruction. Two rules follow. The documentation for the final ground must be in place and retained. The Article 4 statement must reflect the final ground, not the one originally planned.

The sharper point is about ground (h) itself. Because it is residual, damage discovered during the donation window does not merely change the ground for those 300 units. It reveals that ground (h) was never available for them from the moment the damage occurred, because ground (f) applied. Retain the donation offer evidence alongside the new inspection record, not as the basis of the claim but so that the chain of decisions is visible to anybody reading the file two years later.


Keeping 880 units joined to one reference

The regulation imposes no segregation requirement. Nothing obliges a brand to physically separate derogated stock, label a pallet or restrict its movement. What follows is operational good practice rather than legal duty, and it is worth doing because the alternative is a disclosure that cannot be reconciled.

The consignment carries one write-off reference, and the two populations carry sub-references because they carry different grounds. Movement, restacking and picking over are logged against the reference rather than the pallet, since the pallet will not survive the month. The 3PL reports the treatment outcome back per consignment against that reference, which is the second of the three clauses Decision 5 of the destruction ban report recommends putting into the next contract renewal.

Weight comes from the product master rather than a scale. Section 4.4 of the report sets out why: the figure already exists in PLM from development and in ERP for customs and shipping. At an illustrative 0.42 kg per unit, the 880 destroyed units are 369.6 kg, of which 268.8 kg sits against ground (f) and 100.8 kg against ground (g). That is a multiplication, not a data capture project, and it lands comfortably inside the 10% tolerance that Annex III of Commission Implementing Regulation (EU) 2026/2 treats as compliant.


What this event looks like in the published table

One consignment produces a small number of rows in the Annex I disclosure: the customs code at two-digit level unless the category appears in Annex II and requires four, units discarded, weight in kilograms, whether packaging is included, the reason, and the percentage split of the treatment outcome across reuse, recycling, other recovery, disposal and unknown.

The 2,120 repaired units do not appear at all, because they were not discarded. Nor would they if they had been successfully donated, or prepared for reuse and placed. The disclosure counts what was thrown away, not what was written off, and it counts nothing that found a home. That is worth saying plainly to anybody who assumes it is a stock loss report. It is a discard report, and a brand that routes volume successfully out of the waste stream sees its published number fall for the right reason.

The timing has been widely misstated and it matters for planning. The obligation itself is already running: published on the company website within twelve months of the financial year end, in the operator's own format. The prescribed Annex I format applies to financial years starting on or after 2 March 2027, so for a March year end the first affected year runs to March 2028 and the disclosure falls due by March 2029. The runway is longer than most commentary suggests, and the intervening years are the cheap window for building the record rather than a period in which nothing is owed.


Back to the bay

The pallet in the corner of the Venlo warehouse was, in the old sequence, a single object with a single fate: 3,000 units, one collection, one transfer note describing a load of mixed textiles.

Under the new sequence it is three objects. Three populations, two grounds, one reference, one evidence pack, and five years of retention starting the day the skip leaves.

The brands that will struggle are not the ones that destroy too much. They are the ones that arrive at the bay with the decision already made and look for the ground that fits it. The sort has to come first, and it has to come while somebody is still standing in front of the stock.


References


Share this report

LinkedIn

Read Next

Navigating the Digital Product Passport Mandate: What Fashion Brands Need to Know
Sustainability

Navigating the Digital Product Passport Mandate: What Fashion Brands Need to Know

30 January 2025 8 min read
Who is the economic operator? The scope questions the destruction ban leaves open
Compliance

Who is the economic operator? The scope questions the destruction ban leaves open

21 July 2026 5 min read
The ten derogations: a working reference for fashion operations and systems teams
Technology

The ten derogations: a working reference for fashion operations and systems teams

21 July 2026 5 min read