Resale needs a grading language before it needs a marketplace

Retailers that decide to sell used versions of their own products usually begin with the storefront. They choose a resale platform, design a section of the website, photograph a few hundred items and write a description for each one. That was the plan at Sablemere in 2022, and it was the wrong place to start. The constraint in resale is not the shop window. It is the language used to describe condition: whether a grader in Montréal and a grader in Vancouver would describe the same jacket in the same way, whether a customer in Rimouski and a customer in Kelowna would understand the description, and whether it could be written once rather than thousands of times.

We had been taking used gear back for store credit since a small pilot in 2021. By the autumn of 2022 about 9,000 items sat in storage in Vancouver. Most were good. Some needed a repair our workshops could do in an hour. A few were beyond saving and were stripped for parts. None could be sold, because we had no way to describe them that a customer would trust.

Every used jacket is a new product

A new product is described once. Its page is written, photographed, translated and reviewed before launch, and thousands of identical units share that work. A used item has no twin. Its colour has faded to its own degree, its seam tape lifts in its own place, and its previous owner patched an elbow or did not. Every item is in effect a new product with a catalogue of one.

The natural response is free text: a grader looks at the jacket and writes what she sees. Our marketing team liked this, and not without reason. Early pilots showed that customers enjoyed reading that a jacket had crossed the Chilkoot Trail. They proposed that graders write a short story for each item, in English, and that we launch outside Québec while volumes were low.

The difficulty was partly arithmetic and partly law. Québec customers made up about 21 per cent of our online sales and sent a large share of our trade-ins to the Montréal repair centre. Québec's Charter of the French language requires commercial information published there, including what we say about the products we sell, to be available in French. An agency quoted about C$7 a listing and two days' delay to translate free text, on items whose average resale price would be about C$118. Unreviewed machine translation was suggested and rejected by our Montréal writers on quality grounds before our counsel needed to offer a view. Launching in English and adding Québec later would have meant buying trade-ins from Québec customers and selling them to everyone else.

A vocabulary written twice

The alternative was to put the substance of a description into structure. With graders from both centres, our content team wrote a vocabulary of five condition grades and 64 codes, each describing a defect, a repair or a sign of wear, from a lifted seam tape to a replaced zip pull. Each code was written once in English and once in French, by writers working in their first language, and reviewed by the graders who would use it. A listing became a set of selected codes, photographs at fixed angles, measured dimensions and an optional note. The page assembles the description from the codes in the customer's language. The grader's note, where there is one, appears in the language it was written in and is labelled as the grader's own.

It cost us something. Launch slipped eleven weeks, from April to July 2023. The jacket stories largely disappeared, and the campaign built around them was dropped. Graders were slower for the first month while they learnt the codes. One of our marketing directors told me, fairly, that we had turned something charming into something bureaucratic.

The benefits arrived in places we had not fully foreseen. Once graders knew the vocabulary, listing time fell from 14 minutes to about six. Québec customers bought 23 per cent of resale items in 2025–26, slightly more than their share of new online sales. Returns of resale items ran at 7.8 per cent, against 16.5 per cent for new clothing bought online, because the codes told customers precisely what they were getting. And because the same codes now classify returns and repairs, our designers can see when a product fails at the same seam in both centres. Two products have been changed as a result.

What the codes cannot say

A vocabulary has limits, and pretending otherwise would be the next mistake. Codes describe what a grader can see and measure. They say little about how a fabric now feels, or whether a waterproof coating still performs, which would need a test we cannot run on every item. We state that limitation on the resale pages and offer a straightforward return, which is part of why the return rate is not lower still.

The vocabulary also ages. Our first list was written in Vancouver and checked in Montréal, and within two months graders there had shown us that eleven codes described defects in ways French-speaking customers read differently. We rewrote them. New products bring new materials and new failures, so the list is now reviewed twice a year by graders, writers and designers together, and a code is added only when both centres agree on how to describe it.

Finally, structure does not replace judgement about what to sell. Some items satisfy the codes and still should not be resold, because a repair would be visible enough to disappoint or because the product had a known fault. Graders can withhold any item, and they do so about 8 per cent of the time.

Retailers considering resale tend to ask which platform to use. I would ask a different set of questions first. Who will grade, and will graders in different places describe condition in the same way? In which languages must a description exist, and can it be written once? Will returns and repairs use the same vocabulary, so that what graders learn reaches the people who design the next product? A retailer that answers those questions will find the storefront the easy part.