Skip to content

Industry · Clothing and apparel

A clothing store runs on the neutral pack.

No dedicated pack is built for this sector yet, so a shop selling shirts, denim, knitwear or shoes lands on the neutral one. Twelve tools exist in the code, nine of them are handed to the model from the first conversation, and the base rule set every store gets is merged in whatever the shop sells. The honest part of this page is where that leaves the first question an apparel shopper asks. Sizing and fit is that question, and what follows counts exactly how far the code reaches toward it — which tool comes closest, what its schema holds and where it stops, and which three switches a pack would open.

Nine of twelve tools, from day one A colour axis written and switched off Zero roles, zero needs, zero survey questions
  • Products
  • Variants & options
  • Prices
  • Live stock
  • Collections
  • Order status
  • Returns
  • Store policies
  • Answers you write
  • A code you choose
Refik’s console, Overview screen: conversations used against the plan, a bar per day, what happened in them, unanswered questions, and orders Refik took part in.
Product screenshot, sample data. The numbers are made up, not a customer’s result. Once installed, every figure here is counted from your own store.

A pack changes what Refik asks; it leaves the ten sources an answer is built from exactly where they were. A clothing store reads the same ten as a placed one — what Refik reads.

What runs

Nine tools, handed over on the first day.

Ten of the twelve tools are present under every pack; a pack only decides whether shade matching and a remembered profile are offered. The ordered-routine tool is the third one that stays shut here, and your catalogue is what shuts it, and a pack is what would open it: it is offered once at least one chain carries a product role, and roles are the thing a pack supplies. Order status, guest order lookup, the return request, the shipping and returns lookup, the stock-back e-mail and the handover to your team all work from the first conversation. A guest lookup needs the order number and the e-mail address together and is told to require both. The shipping answer is read out of your own policy pages and quoted from them, and it comes back as two legs plus a total, because quoting the transit leg alone understates the wait a shopper is told by about a week.

Your own words do the matching

The search tool runs before any recommendation and reads what the shopper actually typed: it resolves barcodes. That leg is code and the same under every pack; the Spanish and Turkish words — and the misspellings — it maps are hand-written entries in the pack’s synonym dictionary, which here is an empty object. Nothing is corrected by algorithm: matching is substring-based, so a word merely cut short still lands and a transposed one does not. Its own instruction is to never state a product name or price from memory, and an item that is out of stock never appears in what it returns. A line carrying a colour, a cut and a size code is answered out of whatever your titles, tags, options and collections already say, in the language your catalogue is written in. Every step of that path is your own listing, read back.

A sold-out size is a question, not a dead end

This trade runs out of a size long before it runs out of a style, and the stock-back e-mail is the tool written for exactly that moment. Its instruction is to offer it whenever a wanted item is out of stock, right after suggesting the closest alternative — some shoppers want the substitute, some want the exact piece in their own size. It records an address only after the shopper has asked to be told and given one, and it is told never to guess an address or sign anyone up unasked.

Counted, not written

The size question waits on a code change, not a pack.

The tool that comes nearest a fit question is the one that matches a shade, and under the neutral pack it is switched off. Its schema is worth reading anyway, because it is where this trade meets a limit that lives in the code. Five fields: a product handle, a depth with five steps, an undertone with three, a finish with four, and a free-text colour family whose own description reads Color family, for products chosen by color. That last one is already written generically, so colour is genuinely a switch away. Size stops there: the schema carries neither a size nor a fit field, it is closed to extra properties, and the pack-side object has five fields of its own — two naming which product types use which axis, two re-describing depth and undertone, one describing how a chosen tone gets tested. All five describe fields that already exist. So a sizing axis here waits on a code change, and that is a straighter answer than a maybe.

5

fields in the shade schema — a handle and four appearance axes — with extra properties refused.

8

words a clothing claim is built from — size, sizing, fit, shrink, fabric, cotton, wool, leather — absent from all sixteen base claim patterns.

3

languages the base set is written in, and all three read a fibre line or an allergy line differently.

A line about fit meets no rule

Those eight words were looked for with a boundary on each side and the search was calibrated in both directions, because an unbounded one finds words inside other words. Eight out of eight are absent. A line saying an item will fit, that a fabric will not shrink, or that something runs true to size reaches no rule at all. The nearest real protection sits outside the rules: the search tool’s standing instruction to quote only what your catalogue holds, so what a shopper is told is what your own listing says.

The same fibre claim is read twice

The certain-promise rule exists in all three languages and the three disagree about fibre content. The Turkish one fires on a bare hundred-per-cent claim written the Turkish way round, with the sign in front of the number, as in %100 pamuk. The English one fires only when that number is attached to a word like effective or satisfaction, so a plain claim about cotton content passes; the Spanish one, written out in words, passes too. It is the same claim about the same garment. Which language a shopper writes in decides whether anything reads it.

An allergy line turns on word order

The allergy handover pattern catches allergic to wool, an allergy to wool and an allergic reaction, and it catches the Turkish possessive form that needs no preposition at all. It does not catch a wool allergy or my nickel allergy — the same question with the allergen moved in front of the word. Nickel on a stud and wool against skin are ordinary questions in this trade, and today one phrasing reaches your team while the other is answered as a shopping question.

The patterns can also reach too far, and this trade is where that shows. A shopper writing that she is pregnant and asking which size to order matches the first handover pattern on that one word; the model then stops recommending, offers the ingredient list and points her at a doctor. The match is real, and the question was a shop’s. It is the shape the pet trade already exposed — a pattern written for people meeting a trade it was not written for — with the direction reversed: there a handover named the wrong professional, here a question is handed over that needed answering. The one part that lands is the ingredient list, which in this trade is the fibre and care label. What the base set carries.

What this page is a picture of

Refik card for Industry · Clothing and apparel: the sentence this page turns on, and the store data it reads.
Drawn from this page’s own data — the sentence and the sources shown are the ones shipped, not an illustration of them. Regenerated by scripts/sektor-maket-yaz.py.

From the schema

Four things a pack for this trade would add.

A pack is a file of named fields, and the schema says which of them a pack has to carry. These four are the ones this trade would fill, in the order of what they unlock — and the last is the weakest of them, which is worth saying out loud.

  • A size profile, and the sentence that explains it

    The remembered-profile tool is one of the two a pack switches on, and switching it on makes a second field mandatory: the sentence the model is shown describing what this store’s profile is even for. The placed pack carries five profile fields and that sentence; the neutral file carries an empty list and no sentence at all, because supplying a default would mean saying something on a pack’s behalf that the pack itself left blank. Which fields a clothing pack should hold is a decision for whoever writes it; the schema only insists the fields and the sentence arrive together.

  • A synonym dictionary for cut, cloth and colour

    The synonym field maps the shopper’s word onto your catalogue’s word, and it is where a typo, a second language and a trade’s own vocabulary meet. The placed pack carries fifty entries. In this trade the useful entries are the ones a size chart leaves out: the words for a cut, for a cloth, and for a colour a shopper names differently from the way the label does.

  • Claim rules for fit, shrinkage and fibre

    A pack carries its own claim rules and they sit in front of the base ones, so a pack rule is still the first one reported when both would fire. The eight words counted above are what a pack for this trade would cover. A pack rule can also be bound to a single market, and that matters here: fibre labelling is a regulator’s subject and regulators differ by country, so the right rule is one written market by market.

  • Replenishment, and why it is the weakest of the four

    A role maps to a lifetime in days, and a reminder candidate is only produced between twenty-one and two hundred and forty days after the order. Under the neutral pack the candidate list comes back empty for two independent reasons: the switch is shut, and the role-to-days table is one of the nine empty collections in the file. Of these four this is the one the trade has least use for — socks and basics repeat on a clock, a coat does not — and saying so costs less than shipping a reminder nobody wanted.

Unchanged

What stays yours, under every pack.

These three are the same on every vertical page on this site, and that is the point: they describe the part of Refik that your store decides and a pack only reads.

You choose it, and the server measures again

Refik counts how many of your products carry any of a pack’s words and shows you the count, the ranked score, up to eight of the words it matched and up to three example products for each. A pack is only suggested when the share clears a threshold, at least three different words matched and at least five products matched. You confirm it, and the server runs the measurement again on its own before it writes anything. How a pack gets chosen.

Your catalogue supplies the products

A pack supplies names; your catalogue supplies the products. Every option Refik offers has passed two gates in your own store — sellable, and in stock — so the products a shopper sees are the ones you carry today. A pack tag that matches zero products in your catalogue is counted and shown in the console, with the number it matched.

The base set stays merged in

Sixteen claim rules and nine handover patterns are merged in when any pack loads, in English, Turkish and Spanish, and they sit behind the pack’s own rules so the pack’s rule is still the first one reported. A base rule that would report the same words twice stays quiet. An empty set makes the app refuse the pack outright. The base set.

A pack arrives for this trade the day a catalogue in it is measured, and until then the neutral file is the honest one — the neutral pack, counted. The industries page lists the packs that exist and the three gates a store has to clear for one of them; this sector sits among the stores those gates leave unplaced — how a pack gets chosen.