A hundred people open one of your product pages today. Three of them buy. A few days later, two parcels come back.

One customer keeps what they ordered. Marketing paid for all one hundred.

I want to be honest about that arithmetic before I build anything on it, because two returns out of three purchases sits well above any published average. What it describes is a single bracketed order: a shopper who could not tell which size would fit, ordered three, kept one and sent the rest back.

Where those numbers actually sit

Fashion sites convert somewhere between 1.5% and 3% of sessions, depending on price tier and device. Accessible fashion sits at the top of that band, contemporary labels nearer the bottom, and mobile converts at roughly half the rate of desktop despite carrying most of the traffic. Three buyers from a hundred page views is a good day, not a bad one.

Returns are where the range gets uncomfortable. Apparel runs 20% to 40% across most published benchmarks, against about 19% for ecommerce as a whole. Germany is its own case: online fashion returns there run 40% to 50%, with 44% quoted often enough to treat as a working figure. If you sell fashion into the German market, roughly every second garment you ship is coming back.

And bracketing is now normal behaviour rather than an edge case. Around 63% of consumers say they order multiple sizes with the intention of returning most of them, and in UK fashion specifically the figure sits near 62%, almost identical between men and women.

So the three-buy, two-return picture belongs to a specific shopper: the one whose page never answered the question.

Two numbers, one asset

Here is what makes the product page unusual inside a large retailer.

Conversion belongs to marketing. It has a budget, a dashboard and someone who defends it weekly. Returns belong to operations and finance, and get read at the end of a quarter, months after the decision that caused them. The two numbers live in different rooms, in different reporting cycles, under different names.

They meet in exactly one place. The page.

Every other lever moves one of them. More paid traffic moves conversion volume and does nothing for returns. A stricter returns policy moves the return rate and quietly costs you conversion. The page is the only asset in the chain that touches both, and in most companies I walk into it is managed as a production cost to be brought down rather than a commercial lever to be tuned.

Why flattering the product moves both numbers the wrong way

This is the part that took me longest to accept.

A photograph that oversells does its job at the moment of purchase. Better light, a more generous drape, a colour lifted half a stop, a garment pinned at the back so it falls the way it never falls on a body. Conversion goes up. Everyone in the room is happy, because that is the number being watched.

Then the parcel arrives and the customer meets the actual product.

Across published return-reason research, "looked different than expected" and "not as described" sit consistently in the top three drivers, at roughly a fifth of all returns each. Fit sits above them as the single largest cause in apparel. Those are not separate problems. A photograph that flatters is a photograph that misinforms, and the bill arrives three weeks later in a different department's budget.

So flattery raises conversion and raises returns together. Showing the product as it actually is raises returns performance and holds conversion. Those are the only two directions available, and the second one is the only one where the two numbers finally move apart.

What the shopper is actually asking

Peer-reviewed work makes the mechanism plain. Chrimes and colleagues, looking specifically at how fashion product page design affects a consumer's appraisal of fit, found that fit is the most important consideration during garment appraisal and simultaneously the primary reason for online returns, because shoppers do not have enough information to judge it.

Read that as an operational instruction rather than a finding. The shopper is asking one question: will this fit me. When the page answers it, they order one size. When the page stays quiet, they answer it themselves by ordering three.

Bracketing gets filed as customer behaviour. It starts as a page that handed the customer the work.

What "as it actually is" looks like in production

Most of it comes down to a handful of decisions that are almost always missing.

One image in scale. Not a crop, not a styled flat lay. A frame where the garment's size is legible against something known.
Model height and size worn, on the page. A body with stated measurements settles a question that no size chart settles. This one line of text does more for the return rate than most reshoots.
A zoom that shows the material. Enlarging pixels answers nothing. The buyer is asking what this fabric is, so the shot has to be lit for surface rather than for shape.
Fit facts beside the image, not below the fold. Measurements, size worn, fit tendency. If it sits in paragraph four, it does not exist.
Sequence as a decision. Order the gallery by what the buyer needs next, not by the order the files came off the card. This is the single most frequent correction we make on incoming catalogues.

Notice that none of these make the product look worse. They make it look accurate, which is a different thing, and accuracy is what stops the parcel coming back.

How to prove it on your own catalogue

Industry benchmarks set the hypothesis. Your assortment produces your number. If someone tells you a photography change will lift conversion by a specific percentage on your catalogue before seeing it, they are quoting somebody else's data.

The test we run with clients is deliberately dull:

Pick one category with enough volume to reach significance in a season. Shoot it to the standard above. Leave a comparable category on the current standard. Agree the baseline on the retailer's own commercial data before anything changes. Then measure both categories on three things: time to online, conversion, and return rate.

One variable, one category, one season, measured on numbers the client already reports to their own board. If the change does nothing, you find out cheaply. If it works, you have a figure that belongs to your business rather than to a case study.

Want to know what your own product pages are costing you on both numbers?

Book a pilot on one category

The part worth remembering

A hundred people opened the page. You paid for all of them. Three bought, and how many of those three keep what they ordered was decided before the parcel ever shipped, by a page that either answered the fit question or left it open.

Conversion and returns look like two problems owned by two departments. They are one asset, read twice.