Conversion and trust

Why the fields in this theme exist, and what the evidence says about each one. Every element below is API-driven, so a merchant can turn it on with data rather than a deploy.

The number that decides apparel

Apparel returns run 20–40%, the highest of any e-commerce category, against an all-category average of about 19%. Size and fit cause roughly two thirds of fashion returns, with "too small" the single most common complaint. Style and colour account for around 23%, quality about 10%. (NRF via Richpanel, Rocket Returns)

That reframes the whole page. A conversion you win by hiding fit information is a return plus a refund plus return shipping plus a customer who does not come back. The goal is not "add to cart"; it is "add to cart and keep it."

Bracketing — ordering several sizes intending to return most — is now mainstream, up from roughly 40% of shoppers in 2018. Every piece of fit certainty you give someone is a size they do not order speculatively.

What the theme ships, and why

1. Garment measurements per size — product.sizeChart

A letter size means nothing across brands. A chest measurement in centimetres is checkable against a garment the shopper already owns, and it is the only sizing signal that transfers between labels.

The size-chart dialog ends with the one instruction that actually works: *lay a garment you already like flat, measure across the chest 2.5cm below the armhole, double it, match the table.*

Interactive size guides reduce both pre-purchase hesitation and post-purchase returns. (ConvertCart)

2. Aggregated fit feedback — product.fit.feedback

{ small: 34, true: 61, large: 5 }, rendered as one bar.

This is the brand's cut described by the people who bought it, not by the brand. It outperforms any copy a merchant writes about their own fit, because the shopper knows who wrote each.

Do not fabricate it. If you have no post-purchase survey, omit feedback. A made-up distribution produces exactly the returns it was meant to prevent.

3. Model height and size worn — product.fit.model

The cheapest fit signal that exists: it turns a photograph into a scale reference. "Model is 5'9" and wears a S" costs one line and answers the question every shopper is silently asking.

4. Fabric, weight and origin — product.fabric

Drape, warmth and whether something is see-through are weight questions. 190gsm linen and 110gsm linen are different products sold under the same word. Composition, gsm, construction and mill are what a considered buyer is looking for, and almost nobody publishes them.

5. Third-party certifications — product.fabric.certifications

OEKO-TEX Standard 100, GOTS, Responsible Wool Standard, Global Recycled Standard, Leather Working Group. These are independently audited and checkable, which is the entire difference between a certification and a claim — and the reason they survive the greenwashing scepticism that kills self-declared sustainability copy. (Mughal Apparel, Hemptique)

6. Returns and delivery beside the buy button — TrustRow

Placement beats presence. Moving the return policy from the footer to next to add-to-cart lifts add-to-cart rate by around 23% — same words, different position, because the doubt arrives at the moment of commitment. (SplitBase)

The theme also shows a dated delivery estimate rather than a range. "Arrives Thursday 12 September" is a fact to plan around; "2–4 working days" is arithmetic the shopper has to do, and doing it is a moment to leave. GET /delivery-estimate supplies it.

7. Reviews with fit data — review.size review.height review.fit review.photos

A five-star review that says "lovely" is decoration. "Bought size M, 5'11", runs small" is a fitting room.

Two findings worth designing around:

• Real customer photos in the gallery lift product-page conversion by ~35%

versus stock imagery alone.

• The conversion sweet spot for a rating is 4.75–4.99, not 5.0. A perfect

score reads as filtered. The demo catalogue is rated 4.3–4.9 on purpose.

• Specific counts ("302 reviews") convert better than stars alone, so the theme

always renders the number. (ConvertCart)

8. Honest scarcity — product.social

Real counts only: units in stock, bought in the last 30 days, saved. Below a configurable threshold the block renders nothing rather than advertising that a product is unpopular.

There is no "17 people are viewing this right now" in this theme and there will not be. Shoppers recognise the pattern, and the moment they do, every other number on the page — including the honest ones — becomes suspect.

9. Trust signals at the decision point — storefront.trust

Around 70% of online shoppers look for trust signals before completing a purchase, and well-placed ones can lift conversion meaningfully. The theme renders payment marks and a secure-checkout line inside the buy box, not in the footer. (CrazyEgg, Metricuno)

10. Structured enrichment — product.enrichment

Highlights, features and a specification table, in three places rather than one.

The split is the point. A single long table gets read by almost nobody, and the shopper who does want one fact has to hunt for it. Six key/value pairs beside the buy button is a two-second scan that answers "is this the kind of thing I am looking for" before anyone commits to reading a paragraph — and the full table, one group at a time in the detail stack, still catches the person who wants to check the leg opening.

It sits between the highlights and the picker, and none of it is closed. Three placements were tried and two were wrong in the same way:

• A *full-width section below the fold* asks a shopper to scroll the buy button

off the screen to find the fabric weight. Most do not go looking, and the ones who would have were not the ones you were losing.

• A *stack of accordions* asks for a click per section. Same failure, smaller

scale: the shopper willing to open four panels was already going to buy.

• *Under the buy button* still put the reason a shopper trusts the piece a

screen and a half below where they start reading. Detail that arrives after the decision is detail that did not help make it.

So it is one open block with tabs, directly under the highlights and above the colour picker. The first tab renders on arrival; the others cost one click each rather than one click per section.

Yes, this moves "Add to bag" down, and that is the right trade. The sticky buy bar covers it, and the observer driving that bar fires on *not visible* rather than *was visible and left* — so a button below the fold on arrival means the bar is present from the first paint. There is never a moment with no way to buy. Trading button position for detail position is only safe *because* that bar exists; on a page without one, do not make this move.

Tab order follows the order the questions arrive: what is it made of → what are the numbers → what is different about it → how is it built → who made it. The cost of moving into the column is width, which is why the specification table pages by group instead of laying out two columns — the groups are the units a shopper thinks in, so a slide is a complete answer rather than an arbitrary slice.

Feature cards sit side by side with the next one deliberately cut off at the edge. A half-visible card is the only reliable way to say "there are more of these" without a caption saying so, and it is why the row is 86% wide rather than a tidy 100%.

Two rules that keep it useful rather than decorative:

• Highlights are capped at six. More and it stops being a scan and becomes

the specifications table with delusions, which is worse than either.

• Attributes come from a shared vocabulary. Free text everywhere produces

Fabric, fabric, Material and Composition as four attributes across four products, and nothing can ever be filtered or compared. The vocabulary is suggested rather than enforced, so nobody is stopped from describing what they actually sell.

manufacturer is a separate matter: the manufacturer and packer address, the country of origin and the net quantity are legally required on an e-commerce listing in several markets, India included under the Legal Metrology rules. It is compliance, not conversion, and it belongs in a labelled block rather than buried in a description.

11. Key facts over the first photograph — ImageKeyFacts

Two or three chips in the corner of the main image: Pure cotton · 140 gsm · Relaxed.

The cheapest trust signal on the page. A visitor who has looked at nothing but the picture has still read the three facts that decide whether this is the right kind of garment — no scroll, no tab, no click. On a product page that is a large share of the traffic, and it is the share least likely to be reached by anything below the fold.

Three constraints stop it becoming a sticker on the product:

• First image only. Shots two onward are the detail crops — collar, weave,

hem — and covering those covers the answer somebody opened them for.

• Opaque chips, not text on the photo. White text needs a scrim, a scrim

darkens the garment, and the garment is what is being sold.

• Values longer than 24 characters are skipped, not truncated. A composition

like "Recycled polyester shell, Recycled polyester fill" wraps the strip onto a second row and turns a glance into a caption block. The next fact takes its place instead.

It is aria-hidden and pointer-events-none: every fact is also in the highlights list, so a screen reader hears it once, and nothing decorative intercepts a click meant for the image.

12. The services block — product.enrichment.assurances

Returns, exchange, repair and payment, in a short list under the buy button.

This is the block a marketplace listing puts directly under the price, and it is doing different work from everything above it. The specification table answers *is this the right thing*. This answers *what happens if it is not* — and in apparel, where the shopper cannot try it on and knows it, that is usually the last question standing between a considered buyer and the button.

Two details that matter more than the copy:

• The explanation is behind a disclosure, not printed. A returns policy set

out in full is four lines of legal prose beside a call to action. The label is the reassurance; the wording is there for the one shopper in twenty who checks it before committing.

• Product rows replace the store's, they do not merge. A coat with a

ten-year structural guarantee must not also advertise the store's two-year one. Omitting the field inherits; an empty array means *this product has none*.

13. Who made it — product.enrichment.maker

This started as a marketplace seller card — name, rating, years with us — and it was the wrong shape twice over. On a marketplace the seller is the variable and the rating is the reassurance; on an own-brand store the seller is never in doubt and the card has nothing to reassure anyone about. Worse, a rating for a supplier nobody can review is a number somebody typed, which is the fastest way to make the review count and the stock level look typed too.

What survived is the part that was information rather than decoration: who wove the cloth and where, as two rows at the top of the compliance block. That is where a shopper already looks for manufacturing facts, and the address on those rows is the brand's — so the mill is new information rather than a second copy.

The same instinct applies to countryOfOrigin. It is legally mandated in several markets, and answering it with "see product specifications" is a shrug on a required field when the answer is one field away in the fabric data.

14. The variant sheet — mobile

On a phone the buy box is one column, so by the time a shopper has read the detail the colour swatches are most of a screen above them. Comparing two colourways means scrolling up, tapping, scrolling back down — twice — and the second comparison is the one nobody makes.

The sticky bar's selection line is therefore a button, not a caption. Tapping it opens the same pickers in a sheet over the page, and "Add to bag" with no size chosen opens the same sheet rather than scrolling somewhere. The sheet stops short of full height so the photograph stays visible behind it, which matters most when the thing being changed is the colour.

Both pickers are the same components the buy box uses. A second copy of a variant picker is how "sold out" ends up struck through in one place and greyed in the other, and how one of them quietly stops handling a sparse matrix.

15. Full-screen photography with zoom — Lightbox

Apparel is bought on detail a 900px thumbnail cannot carry: the weave, the stitch density, how a collar actually rolls. A shopper who cannot get closer either takes the risk or leaves, and in a category where returns already run high, "takes the risk" is not the outcome to design for either.

Tapping the photograph opens it full screen with arrows, thumbnails, keyboard control and zoom to 3×. Three decisions that make it work rather than merely exist:

• Zoom is a transform, not a bigger request. The source is already the

largest file there is; re-fetching at higher density would put a spinner in the middle of the interaction meant to answer the question.

• Pan is clamped to the image. Unclamped panning is how a zoomed photo ends

up as an empty field with a sleeve in the corner and no obvious way back.

• Zooming disables the swipe. Sharing one gesture between "look closer" and

"next photo" means neither works — every pan across the image would advance the gallery.

Video opens in the viewer but never zooms, and its frame is a plain box rather than a button: a <video controls> inside a <button> is invalid markup and a scrub bar that cannot be scrubbed.

16. The cart drawer's vertical budget

Everything in the drawer competes with the one thing it is for: seeing what is in the bag. Measured against the classes actually in the file, the header, the suggestion rail and the totals came to 548px of fixed chrome — on a 667px phone that leaves 119px, which is *less than one line item*. A two-item bag where the second item is a rumour.

The suggestion rail was the whole problem. Three 4:5 cards with a name and a price under each is a third of a phone screen given to things the shopper has not chosen, directly above the total they came to check. Laid on their side — 40px thumbnail, name and price beside it, scrolling horizontally — the same three cost about 90px and are no harder to read, because a small thumbnail is plenty to recognise something you were looking at a minute ago.

Chrome drops to 410px, so a 667px screen shows two items instead of none.

The lesson generalises: in a drawer, *rails are the expensive thing*. A card grid is the right shape on a page with a screen to spare and the wrong one in a panel where every block pushes the checkout button further away.

17. Fit warnings in the grid — ProductCard

"Runs small" appears on the card, not just the product page. Someone comparing eight products decides which two to open from the grid, and that is the fact that decides it.

Turning it on

Everything above is data. Nothing needs a code change.

Feature switches live in storefront.trust:

{
  "trust": {
    "payments": ["Visa", "Mastercard", "Amex", "PayPal", "Apple Pay", "UPI"],
    "repairs": true,
    "showCertifications": true,
    "showFitFeedback": true,
    "showSocialProof": true,
    "socialProofThresholds": { "bought": 25, "saved": 20 }
  }
}

If you only do three things

1. Publish garment measurements per size. It is the single highest-leverage

field in an apparel catalogue and most stores do not have it.

1. Put returns and a dated delivery estimate next to the buy button. Same

copy you already have, moved.

1. Collect fit feedback after delivery — one question, three options — and

publish the distribution. It compounds: every order improves the next shopper's decision.

What this theme deliberately does not do

• No countdown timers on evergreen products.
• No fake viewer counts or invented "only 2 left" when there are forty.
• No pre-checked marketing opt-ins.
• No card fields. Payment is handed to a provider — see CHECKOUT.md.

These lift a single session's numbers and cost the second purchase. For a store selling £200 coats that people are meant to keep for a decade, that is a bad trade.

Sources

• Ecommerce Return Rates in 2026 — Richpanel
• Ecommerce Return Rates 2025 — Rocket Returns
• Why Is Fashion E-Commerce Return Rate So High — Koozee
• Apparel Product Page Examples — ConvertCart
• Product Page Highlights — SplitBase
• Trust Signals That Boost Conversion — CrazyEgg
• Ecommerce Trust Signals — Metricuno
• Understanding Apparel Certifications — Mughal Apparel
• OEKO-TEX vs GOTS — Hemptique