Article numbers. The one field every other system will join on. Worth deciding once, deliberately.
What the number is for
An article number is not a description. It is the handle that a tech pack, a sample round, a purchase order, a warehouse and an accounting system all use to mean the same garment. Everything else about the article can change: its name, its colourway, its supplier, its price. The number has to survive all of it.
It is also worth being honest about who the number is for. It is not for your customer. Nobody has ever declined to buy a coat because the code was hard to remember. The article number is internal plumbing, which is exactly why it deserves ten minutes of deliberate thought and then no thought at all, ever again.
Every brand starts with a coded number
When a brand is small, the obvious move is a coded number: a few letters of the style name, an attribute, a colour, so anyone can decode it on sight. It genuinely works, and it works for years.
Then the logic runs out of room.
Nothing here was a mistake. Each patch was the sensible thing to do that week. But the number is now a set of rules with a set of exceptions, and the only people who can read it are the ones who were there when each exception was made. A new joiner sees noise. And the time spent maintaining it goes into something no customer will ever see.
The rolling number
The alternative is a rolling number: stop asking it to describe anything and let it simply count. Think of it as a key: the thing you hand a system to get the real data back. Three levels, each rolling on from the level above.
An article is not always a colour, incidentally. It is whatever varies below the style while the style stays the same garment: a colourway, an organic and a non-organic version of the same shirt, a men’s and a women’s size run. Keeping the number neutral is what lets the article level mean different things for different products without anyone having to redesign the scheme.
The pay-off is quiet, which is why it is easy to undersell. You can mint a rolling number the day a style is sketched and keep it to end of life. You can recolour, recategorise, change supplier, move a style to next season, add a gender, and nothing anywhere needs renumbering. And you never again spend a meeting deciding what a code should be.
The real cost, and when it lands
This is the argument that matters most, and it is the one most easily missed, because it does not show up until months after the numbering was decided.
Development is development. A colour called Dusty Rose at the first fitting is Clay by the time the range is signed off. A material changes. A style moves category. None of that is a mistake; that is the process doing exactly what it is for. But with a coded number, every one of those is also an edit to the number, so a completely normal act of product development turns into a systems decision, and both of the available answers are bad.
Retire it and start again
The style’s history splits in two. Samples, comments, measurement rounds and costings stay on the retired article; the new one starts empty. And if an order is already placed, the article you ordered is no longer the article you are producing.
Let the number update
The number on the purchase order, the label draft, the supplier’s spreadsheet and three email threads is now a number that does not exist. Technically nothing broke. Somebody still has to reconcile it, under time pressure, more than once, until everyone downstream has caught up.
The second-order cost is the one that should decide it. Once a team learns that renaming a colour is expensive, they stop renaming colours. The admin has quietly begun making product decisions, and a system that discourages a design team from improving a name has taken something from the brand that no efficiency gain will pay back.
A rolling number has nothing to react to. Rename the colour, change the material, move the category, place the order. It does not move, because it was never a claim about any of them.
The colour number, and why it is the hardest one to give up
Almost every brand we onboard has the colour built into the article number, and almost every one tells the same story.
At the start it is obvious. Black is 9000. White is 1000. Everyone learns it in a week and it feels like good housekeeping.
Then the assortment grows. An off-black arrives, so 9001. A black with white trim: is that a black? Someone decides it is, so 9002. Two years on, a plain true black is added because a buyer asked for one, and there is no number left that says this is the actual black. It takes 9003, and from that day the number space no longer means what everyone was taught it means.
Nothing failed. The logic simply ran out of room, the way it always does.
The version that survives: a group, and a queue
If a colour reference in the number really is worth the upkeep, keep only the part that is stable and let the rest roll.
The colour group is stable. A black stays in the black family this year and next, whatever the yarn does. So 9 means “this one is a black”. The digits after it mean exactly one thing: the order it was created in.
9000 is not “the real black”. It is “the first black”. The number after the group prefix is a queue position, not a verdict.
This is the part that takes longest to land, and it is worth saying plainly, because the objection is always the same: surely the proper black should be 9000? No, because the number was never a claim about the colour. It is a place in a queue. Once a team accepts that, the argument about which black deserves 9000 evaporates, since no black deserves anything. It takes the next free number, and the colour name, the Pantone reference and the colour group travel with it into every system downstream.
Worth deciding alongside this: whether your colour library is per brand or shared across the group, and whether a colour is identified by its Pantone or by the name someone typed. Those two questions cause more duplicate colours than the numbering ever does.
“But then I can’t tell what it is”
Fair, and worth answering properly rather than waving away.
You will still learn your best sellers. The number on the style that pays the rent will be in your head within a season, exactly as it always was.
Everything you wanted to read is still there, and there is more of it. Name, colour, Pantone, colour group, season, supplier, category, gender. A code that carried three of those carried them badly. Fields carry all of them precisely, and any modern system can group, filter and report on them without the number’s help. If a report cannot do what you need, that is a conversation about the report.
If a short human reference genuinely helps, build one. For the warehouse shelf or the collection meeting, a readable label is a reasonable thing to want. Just don’t make it the article number. Make it a field, so it can be changed when it stops being useful.
The one answer worth pushing back on is “our ERP requires that format”. Sometimes true. More often it is a preference someone configured years ago that nobody has re-examined. Ask the question before you design the next decade of numbering around it.
History stays. Only what comes next changes.
This is the question that stops most of these conversations before they start: does this mean renumbering six thousand articles? No. It never does, and it should never be attempted.
Changing your approach is a line in the sand, not a migration project. Everything that already exists keeps the number it has, for life. Every purchase order, care label, invoice, warehouse location, spreadsheet and email thread that quotes an old number stays correct, because the number it quotes still points at the same garment. That is the one job a number has, and a number that already does it well should never be disturbed.
What changes is everything created from that point on: new styles, and new colourways added to existing styles. Draw the line at a season boundary and it is almost invisible in practice.
Before the line
Keeps its number, exactly as it is. Never rewritten, never reissued, never reconciled.
After the line
Rolls. New styles and new colourways take the next number in the sequence.
You will have two conventions living side by side for a few seasons, and that is correct rather than untidy. The old numbers are not wrong; they are simply finished. As the old styles retire, the mixed estate resolves itself without anyone doing anything about it.
In Resourced this is how it behaves by default. Numbers carried in from your previous system are held exactly as they arrived and are never regenerated. Saving a new template changes what gets created next; it does not reach backwards into what already exists. Renumbering an existing catalogue is possible, but it is a separate action somebody has to deliberately choose to run, and our advice is almost always not to.
How this is set up in Resourced
Numbering is configured per brand on the Article Numbers page: one template for each of the three levels. Each is built from the variables available at that level, so the structure is yours rather than ours.
{% if hierarchy_code[0] %}{{ hierarchy_code[0] }}-{% endif %}{{ sequential_number }}
F-1042
{{ product_number }}-{{ sequential_number | pad_left: "000" }}
F-1042-001
{{ article_number }}-{{ size_code }}
F-1042-001-M
Those three are the defaults, and they are what we recommend for most brands. The product number rolls brand-wide and can start wherever you like. Set it to 1000 and your first style is 1000, which spares you the look of a catalogue with four items in it. The article number rolls within its product, so the first colourway on every style is 001. The SKU takes the size code from your size chart.
The variables go well beyond a counter, though, and it is worth knowing what is on the shelf before you decide you don’t need it. At product level: the hierarchy codes, creation date, gender, tags and your own non-seasonal custom fields. At article level: the product number, colour name, colour code and colour group code (the mechanism behind the 9-for-blacks example above), alongside the article’s own gender, tags and custom fields. At SKU level: size code and size value.
There is a deliberate escape hatch, too, for brands that need colour visible in the number. Colour codes can be minted automatically from a numeric range you set: one strategy holds a single stable code per colour for life, another allows the same colour to carry more than one code where that is genuinely meaningful: three blacks that differ by material, say. Either can be paired with an automatic collision number for the cases where two articles would otherwise land on the same string. All of it works, it is supported, and we will happily set it up. It is simply more to maintain than a rolling number, and we would rather you chose it knowingly.
The sharp edge worth knowing about
Whatever you build the number from, you have also decided what can renumber it. Put a colour code in your article template and recolouring an article changes its number; put a gender initial in it and the day a style becomes unisex, the number moves. That is not a flaw in the templating; it is the cost described above, made mechanical and visible.
Individual numbers can be set manually and are then left alone, which is the right answer for legacy articles carried in from an old system, and the wrong answer as a way of life. If you find yourself locking numbers routinely, the template is telling you something.
Before you have this conversation
Six questions. If you can answer them, the numbering decision usually takes twenty minutes rather than three meetings.
- Who actually reads the number, and to do what? Picking from a shelf, quoting a supplier, referring to a style in a meeting. Each has a different fix, and most are not the number.
- What would have to change for the number to change? The target answer is “nothing”.
- What are you encoding today that is really a filter? Season, gender, category and division are almost always fields wearing a costume.
- Is the legacy constraint real? Does a downstream system truly require a format, or does someone prefer it? Get that in writing before you design around it.
- Does it survive ten times the catalogue? Do the arithmetic on digits now, not at 8,000 articles.
- Who owns it? One owner, one place. Numbering invented independently by three teams is the actual failure mode.
And one thing to settle in the same sitting, because the two decisions lean on each other: how deep your product hierarchy goes and what each level is called. More on hierarchy decisions →