Reducing returns from the product page

Returns are a product page problem long before they are a logistics problem. A return happens because the item that arrived was not the item the buyer expected, and the expectation was set on the page. That makes return reduction a merchandising discipline, measured in the same place as conversion, and it is one of the few interventions that raises profit without needing more traffic.

The work is unglamorous and specific: find out why items actually come back, trace each reason to the part of the page that failed, fix that part, and watch the rate on that product group. Here is how to run it.

Get the return reasons into a usable shape first

Most stores collect a return reason and never look at it, because the reason list was written for the warehouse rather than for merchandising. Rewrite the list so each option points at a cause you could act on:

• Too small
• Too large
• Colour or finish differs from the images
• Material or texture not as expected
• Quality below expectation
• Arrived damaged
• Wrong item received
• Changed mind, no fault
• Found it cheaper elsewhere
• Arrived too late
• Ordered multiple to choose from

Add a free-text field and read it. The categories tell you which bucket, the free text tells you which detail. Then compute the return rate by product, by variant and by category, weighted by revenue rather than units, because a high return rate on a low-priced item costs less than a moderate rate on your flagship.

One caution on measurement: return rate should be calculated on orders that have passed the full return window, so a recent month always looks better than it is. Use a cohort by order date, not a count of returns received in the month.

Match each reason to the page element that caused it

Return reason	What the page failed to do	The fix
Too small or too large	Size guidance was generic or absent	Product-specific measurements, fit notes, model reference
Colour differs	One lighting setup, no context	Multiple lighting conditions, a real-world shot, a named colour description
Material not as expected	Composition buried or missing	Composition stated plainly, close-up texture image, weight
Quality below expectation	Images flattered, copy overpromised	Honest photography, specific rather than superlative language
Ordered multiple to choose	Buyer could not decide from the page	Better fit tools, or accept it as the cost of the category
Arrived damaged	Packaging, not the page	Operations, but track it separately so it does not distort the analysis
Changed mind	Often price or delay, not the product	Check delivery estimates and post-purchase communication

Split the operations causes out early. Damaged and wrong item are fulfilment issues, and leaving them in the same pool as fit issues makes the page-level numbers unreadable.

Sizing: the single largest lever in apparel, footwear and furniture

Generic size charts do not work because manufacturers cut differently and buyers know it. What works is measurements of the actual garment or item, in the units of the market you sell to.

Give three things on every size-sensitive product:

1. Product measurements, not body measurements. Chest width laid flat, inseam, sleeve length, or for furniture the assembled dimensions and the boxed dimensions. Body measurement charts ask the buyer to translate; product measurements let them compare against something they already own.
2. A fit note in plain language. “Runs small in the shoulders, size up if you are between sizes” is worth more than a chart. Take these from your own returns free text, because that is where the pattern is already written down.
3. A model or scale reference. The model’s height and the size worn. For homeware and furniture, an object of known size in the frame, or a room shot with dimensions marked.

Where you sell footwear or fitted apparel across regions, show conversions rather than assuming the buyer will find them. A mismatch between the size on the page and the size on the label is one of the most reliable causes of a return.

Photography that prevents returns rather than sells harder

Product photography usually optimises for the click. Return-reducing photography optimises for accuracy, and the two are not in conflict as often as people assume.

• Show the colour under more than one light. Include one image that is deliberately unstyled, in neutral light, next to a familiar object.
• Show scale in every set. In hand, on a body, in a room, next to something common.
• Show the texture close. A macro shot of the weave, grain or finish answers the material question directly.
• Show what is included. Everything in the box, laid out. This removes a whole class of “missing parts” complaints.
• Show the imperfections you know about. Natural variation in wood, leather or stone should be photographed and named, not avoided. Buyers who expected variation do not return it.

Video is the strongest tool for movement, drape and mechanism. Where the return reason is about how something behaves rather than how it looks, a short clip answers it better than any number of stills. Do product videos increase conversion covers the conversion side, and the return side usually moves in the same direction. The wider question of image count and coverage is in how many product images should a product page have.

Copy that sets an accurate expectation

Superlatives create the gap that returns fill. Replace them with specifics.

“Incredibly soft” is a claim the buyer cannot verify and will judge against their own best comparison. “220 gsm brushed cotton, softer than a standard tee, holds shape after washing” is checkable, and the buyer who wants something softer self-selects out. Losing that order is cheaper than shipping it twice.

Three habits:

• State the limitation. Not machine washable, not suitable for outdoor use, requires assembly, needs a specific accessory sold separately. Buyers who find out on delivery return; buyers who find out on the page either accept it or leave.
• Answer the top objection in the first screen. Whatever your support team gets asked most about this product belongs above the fold, not in a tab.
• Use the words customers use. Pull the vocabulary from reviews and support tickets rather than from the supplier sheet.

Let reviews do the work you cannot

Reviews reduce returns when they carry structured attributes alongside the text. Fit rating on a scale, height and usual size for apparel, room type for furniture, skin type for skincare. A buyer who can filter to reviews from people like them makes a better decision than one reading five-star averages.

Ask for the return-relevant attributes at review time, and surface the summary on the page: the proportion who said it fits true to size, the most common comment about colour. Reviews are also the cheapest source of the fit notes recommended above. Displaying them properly has a search benefit as well, and the requirements for review markup are documented rather than a matter of opinion.

Where the policy fits

A generous return policy raises conversion and raises returns. Both are true, and the net effect depends on your contribution margin and on the cost of processing a return, which includes inbound shipping, inspection, repackaging and the share of goods that cannot be resold at full price.

Decide the policy from that arithmetic rather than from what is common in the category. Then state it clearly on the product page and in checkout, because a policy nobody can find creates support load without earning the conversion benefit. If you advertise on Shopping surfaces, return policy information is part of the account setup, and inconsistency between the policy on the site and the policy in the account is worth checking.

Two structural options worth considering before a blanket policy change:

• Free returns on categories where fit is the risk, paid returns elsewhere. Aligns the cost with the cause.
• Store credit offered at a premium over a refund. Keeps the revenue while giving the buyer a real choice.

Measure it as a programme

Set up a monthly view with return rate by category, the reason mix, and revenue lost to returns. Then pick one category per quarter, fix the top reason on its top twenty products by revenue, and read the rate on that group against the rest of the catalogue as a control.

Expect a lag. Returns arrive weeks after orders, so a change made in one month shows in the following one, and a clean read needs a full return window on both sides. Track two numbers together: return rate and conversion rate on the same products. A page change that halves returns and halves orders has not helped. The goal is contribution per session, and both halves belong in the calculation. Where the diagnosis points beyond the page, high return rate sets out the operational side.

---

Sources

• Google Merchant Center Help: Set up your return policies for Shopping ads and free listings
• Google Search Central: Product structured data
• Google Search Central: Review snippet structured data