What Is This Shirt Made Of? Shopify Fields

A buyer reaches your T-shirt through an AI shopping answer and asks one question you can almost hear out loud: “What is this shirt made of?” Not the brand story — the fiber. Cotton or a blend? A heavyweight tee or one that goes thin by the third wash? This clinic maps that single buyer question to the exact Shopify fields where its evidence belongs, separates what Shopify documents from what it does not, and leaves you with an audit you can run on five products this weekend.

The short answer: the product Description is the fabric surface Shopify documents as considered by AI platforms, the precise numbers already exist on your legally required care label, and one optional category metafield standardizes the fabric name — each with a bounded, clearly separated role.

What the buyer is really asking

“What is this shirt made of?” is rarely one question. It is five:

1. Fiber content — which fibers, at what percentages (“100% cotton” versus “60% cotton / 40% polyester”).
2. Fabric weight — the grams-per-square-metre (GSM) figure that separates a light summer tee from a heavyweight one.
3. Construction — jersey knit, ringspun, combed, garment-dyed.
4. Feel and stretch — drape, softness, whether any elastane adds give.
5. Behavior over time — shrinkage, fading, pilling.

A listing can support each sub-question, leave it ambiguous, or omit it entirely. Scoring those verdicts, field by field, is the whole clinic.

The exact numbers already exist: your care label

If you sell clothing in the United States, the fiber content is not a merchandising choice. The FTC’s Textile Fiber Rule requires textile products to carry labels disclosing the generic names and percentages by weight of the constituent fibers, plus the manufacturer or marketer name and the country where the product was processed or manufactured.

That means the single most precise fabric sentence in your business is already printed on a label you paid for — and most apparel listings never type it into the product description. Copy it across, percentages first, without improving on it. If your supplier cannot produce a conforming label, that is a sourcing problem no listing edit can fix, and guessing percentages is the one thing this clinic will never recommend.

The one field Shopify documents AI platforms reading

Shopify’s page on optimizing your products for AI platforms lists the product details considered by AI platforms and shopping sites: Title, Description, Images, organization details such as Type, Vendor, Collections, and Tags, Barcode (ISBN, UPC, GTIN), and Variants including Option name. The Description is on that list, and Shopify’s own product description guidance names exactly the content this buyer question needs: “product specifications such as size, material, weight.”

Two boundaries matter, stated plainly:

• Shopify documents that these fields are considered — it does not publish how any platform weighs them. Meeting Catalog requirements and optimizing these fields does not guarantee visibility, a citation, or a recommendation in any AI answer.
• Fabric copy that lives only in a theme section, a review reply, or an image caption is outside that documented field set. If a fact decides the buyer’s question, it belongs in the description body, where the documented field lives.

The same Shopify page documents the Knowledge Base app for reviewing and customizing FAQs that AI platforms use to answer questions — the right home for recurring fabric-behavior answers once the description itself is fixed.

The optional second layer: the fabric category metafield

Assigning a product category from Shopify’s Standard Product Taxonomy — for a tee, something like Apparel & Accessories > Clothing > Clothing Tops > Shirts — unlocks category metafields, including a standardized fabric attribute alongside size, color, and neckline. Shopify documents these metafields for storefront filtering and for selling on channels that require standardized product types.

What Shopify does not document is the fabric metafield as an AI-platform signal. The optimizing-products page names the Description, not category metafields, and the documented Catalog search filters are Color, Size, and Target gender — no fabric filter exists. (We covered that filter set in detail in Agents Filter Color, Size, and Gender.) So set the metafield for standardization, and put the buyer’s evidence in the Description. Two different jobs; do not sell the second one as the first.

The fabric-evidence matrix

Run this against one product at a time. Verdicts: Supported (the buyer can answer the sub-question from the listing), Ambiguous (words present, decision not), Absent (nothing to evaluate).

1. Fiber content

• Evidence to look for: exact percentages by fiber, matching the care label.
• Field: Description — the documented AI-considered surface. Optional mirror: fabric category metafield.
• Typical verdict on real listings: Absent — “premium cotton blend” names no percentages and hides the second fiber.

2. Fabric weight

• Evidence: a number with units (GSM or oz/yd²) plus a plain-language anchor, since a buyer comparing stores needs both. Lighter tees are commonly marketed near 150 GSM; “heavyweight” branding commonly starts around 200 GSM — give the number, not just the adjective.
• Field: Description.
• Typical verdict: Ambiguous — “heavyweight” with no figure, or a figure with no comparison point.

3. Construction

• Evidence: knit type and spin — ringspun, combed, jersey, garment-dyed.
• Field: Description; optionally the Title for one flagship claim a supplier document supports.
• Typical verdict: Absent to Ambiguous.

4. Feel and stretch

• Evidence: elastane percentage, drape and hand-feel phrasing a buyer can act on.
• Field: Description.
• Typical verdict: Ambiguous — “soft, stretchy, comfortable” is unfalsifiable and gives an agent nothing to reason with for a buyer who avoids synthetics.

5. Behavior over time

• Evidence: shrinkage treatment, wash-fade notes, care behavior.
• Field: Description for the product-specific facts; Knowledge Base FAQs for recurring questions.
• Typical verdict: Absent.

The matrix is the repeatable method: one buyer question, five sub-questions, one verdict and one field location each. Run it across your five best-selling tees and your gap pattern will be obvious within the hour.

A controlled before-and-after (illustrative example)

Constructed example — not a real store, and no performance claim is made or implied.

Before — a typical 26-word paragraph:

The Classic Tee. Cut for everyday wear in a premium cotton blend, with a relaxed fit you will reach for again and again.

After — label-conform, still one paragraph:

The Classic Tee is 100% combed ringspun cotton, 180 GSM jersey knit — a mid-weight tee with a soft hand and no stretch, pre-shrunk so the fit holds. Relaxed cut through the body. Wash cold, tumble dry low.

Nothing was invented: every figure is the kind of fact a care label and supplier spec sheet already state. The after-version answers four of the five sub-questions in roughly the word count most brand paragraphs already spend.

What this can and cannot do

A description that states fiber percentages, weight, and construction gives any reader — human, or an AI platform shopping on a buyer’s behalf — the raw material to answer “what is this shirt made of?” truthfully. It removes ambiguity you created. What it does not do: control whether any AI answer cites, ranks, or recommends your product. Shopify documents consideration, not outcomes; the Agentic channel’s Listing quality meter, which we examined in Shopify Now Grades Your Listings for AI, is itself a data-completeness check, not a ranking promise.

The rest of the buyer’s decision chain

Fabric is one link. The same apparel buyer usually stacks it with fit — our Shopify fit evidence clinic covers the variant and size-chart side — and with the return question behind it, covered in Can I Return This?. The full decision chain is mapped on the Shopify apparel product discovery hub, and the practice of structuring listing evidence for AI channels more broadly is what we call Generative Engine Optimization — see the GEO for marketplace sellers guide for the general method this clinic applies.

Run the audit this weekend

1. Open your best-selling tee in the Shopify admin and read only the description, asking the five sub-questions above.
2. Record a verdict — Supported, Ambiguous, Absent — for each, and the field you would fix.
3. Pull the care label or supplier spec sheet and rewrite the description paragraph with fiber percentages, weight, construction, and stretch, changing nothing you cannot verify.
4. Assign the clothing-tops category and set the fabric category metafield to its standardized value.
5. Repeat for your next four products, then re-read the first one. The verdicts move fast once the source of truth is the label.

How FirstShelf can help

FirstShelf is built around this exact method: take one buyer question from the apparel decision chain, audit the listing evidence a buyer and an AI platform can actually verify, and show the seller-controlled fields that close the gap. We call it the Buyer Match Audit. We do not connect to your Shopify admin, we do not publish changes for you, and we do not promise AI placement — we make the evidence legible so the truthful version of your product is the one platforms can read. Start with fabric on one tee and the rest of the chain follows.

{cta}