Skip to content
Faisal Mahmud TusherDigital marketing, SEO & analytics
with Faisal Tusher

E-commerce Marketing › Troubleshooting

Online Store Return Rate Rising

Why an online store's return rate rises and how to fix it: missing size detail, photos that mislead, campaign promises, gift orders and refunds counted late.
Troubleshootingintermediate9 min readUpdated

First, confirm the rise and find where it sits

Start with the calendar. The return rate entry explains that returns belong to the orders they reverse and arrive after them, so only order cohorts whose return window has closed give a final figure. Line up the periods being compared and mark which are still open. Then confirm that refunds reach the records you are reading: the GA4 e-commerce tracking tutorial notes that refunds usually happen in the back office, which makes the refund event the one most implementations skip. A store whose refunds were switched on, switched off or batched differently this season can see its rate move without a single extra parcel.

With the measurement settled, cut the rise five ways. By product and by variant, since a single size or colour can carry most of the change. By campaign, because the orders from one promotion behave differently from orders that arrived without one. By the gift flag set at checkout. And by coded return reason, which the returns policy and experience guide asks every store to collect as a short list rather than free text. The cut that concentrates the rise tells you which row of the table to start from.

Diagnosis table

Diagnosis

SymptomCheckLikely causeFix
Returns concentrate in particular sizes, coded as too small or too largeReturn reasons by variant; whether the size guide opens beside the selector and gives measurementsSizing and fit information missing where the choice is madePut measurements and a fit note beside the size selector
Returns coded as not as pictured, clustered on certain colours or itemsThe product's images against the real item in daylight; whether each variant has its own imagePhotographs that misstate colour, scale or detailReshoot the gaps with colour checked and a scale reference
The rise belongs to orders from one campaignAd and email claims beside the product page and the product itselfA campaign promising something the product page and product do not deliverCorrect the claim and make the page state the same thing
The rise appears after a gifting season and fades afterwardsReturn rate with gift orders split from personal ordersGift orders returned by recipients and read together with every other orderReport gift returns apart and make exchanges easy for recipients
Returners leave with a refund although many wanted another size or colourThe options shown in the returns journey and their orderA refund-only returns journey where an exchange would have kept the saleOffer the exchange first, on top of the right to a refund
The rate jumps, or recent periods look better than older onesReturn window status of each cohort; refund events against the order systemReturns read before the return window closes, or refunds never recordedReport closed cohorts only and send refunds with the original order identifier

Fix, verify and prevent each cause

Fit information the shopper never saw

When the coded reason is size and the returns pile up on a few variants, the page let the buyer guess. The product page optimisation guide calls a size guide that opens in place, with measurements rather than labels and a note on how the item fits relative to expectation, the most effective returns-reduction element a page can carry, and puts fit feedback from reviews beside the selector.

Fix: for the products the rise points to, add the measurement buyers need most and a plain fit note, reachable without leaving the selector.

Verify: among closed cohorts after the change, size-coded returns for those variants fall and the share of other reasons does not grow to replace them.

Prevent: make the size guide a required field when a new style is listed, so no product goes live with a selector and nothing beside it.

Images that set the wrong expectation

"Not as pictured" is the returner saying the screen and the parcel disagreed. The product photography brief tutorial treats colour accuracy, a consistent scale reference and detail shots as standing lines in every brief, and starts each brief from return reasons, so the photographs answer the questions that previously went unanswered. It also records that each variant needs its own image, which Google's product data specification reflects in how a feed carries a variant's image.

Fix: reshoot only what the returns name, comparing the shade with the real item in daylight and adding the missing scale or close-up.

Verify: that reason's share of returns falls for the reshot products once their cohorts close.

Prevent: add the returns export to the brief's inputs each season, so the next shoot begins from the newest complaints rather than last year's shot list.

A campaign that promised more than the product keeps

When the rise belongs to one promotion, compare what its ads and emails said with the page and the product. The guide to merchandising the store for a campaign asks product pages to continue the campaign's message where the decision is made, and the returns policy guide reads returns concentrated in a campaign as a pointer to what that ad promised.

Fix: withdraw or reword the claim the product only partly meets, and state the same limit on the product page so later buyers decide on accurate terms.

Verify: returns from the corrected campaign's later orders move back towards the product's usual pattern, with the campaign-coded reason thinning out.

Prevent: have whoever owns the product approve every claim in campaign copy against its specification before the ads go live.

Gift returns mixed into the store's figure

A recipient who did not choose an item returns it for reasons no product page could fix. The guide to selling gifts online asks stores to read the Return Rate on gift orders separately for that reason, using the gift option at checkout as the record that an order was a gift, and to offer recipients a gift receipt or packing-note reference for exchanges.

Fix: split gift orders out of the headline rate and give recipients an exchange route that does not involve the giver.

Verify: the personal-order rate, read alone, shows whether anything beyond seasonal gifting changed.

Prevent: keep the gift flag mandatory at checkout and keep the split permanent on the returns report, not only at peak.

A journey that only refunds

Many returns are the right product in the wrong size or colour, and a journey that leads straight to a refund turns each of them into a lost sale. The returns policy guide puts the exchange first in the journey and store credit second, always on top of the customer's right to a refund where that right applies, and treats the journey itself as a retention moment. Whatever the store offers must not suggest fewer rights than customers have; the guide frames this as general direction, not legal advice.

Fix: show the exchange option before the refund and dispatch the replacement as early as the store safely can.

Verify: among returned orders, the share kept as exchanges rises, and the net rate of revenue lost to returns falls even if the count of parcels does not.

Prevent: review the journey's option order whenever the returns platform or policy changes.

A rate read too early or without refunds

If the rise vanishes when you compare closed cohorts only, or appears the week refund tracking changed, nothing happened to customers. The GA4 e-commerce tracking tutorial records, from Google's developer guide to measuring ecommerce, that a full refund is sent with the original transaction identifier and a partial refund with only the refunded items.

Fix: send refunds from the order system with the original identifier and report only closed cohorts, or label the open ones as provisional.

Verify: a closed period reconciles with the order system's refund list.

Prevent: log every change to refund tracking or the return window in the measurement change log, so a future jump can be checked against it in minutes.

A prevention routine for returns

Returns are a delayed verdict on decisions made weeks earlier, so the routine has to look backwards and forwards at once. Each month, read the closed cohorts by product, variant, campaign and gift flag, with the coded reasons beside them, and name an owner for any product whose reasons cluster. Before each campaign, check its claims against the specifications of the products it features. Before each shoot, hand the photographer the latest return reasons for the range. The tutorial on coding and reading return reasons builds the coded reasons this routine reads, from the first list to the count against units sold.

Read returns in money as well as in parcels. Ad platforms keep a returned order's value, so a campaign can top the ROAS table while costing more than it earns; when the question becomes profit, the page on a high ROAS that is not profitable takes over from this one, and Contribution Margin after returns is the figure that settles it. Finally, keep the policy, the feed and the page saying the same thing, so that a return never begins with a promise the store did not make. The other store symptoms, from smaller baskets to stalled restocks, sit together in the e-commerce troubleshooting section.

Illustrative example — not a real result

Tuftmallow Workwear traces a spring jump in returns

Tuftmallow Workwear, an invented workwear brand, sees its order return rate reported at 19% for April against 13% a year earlier. The analyst first checks the cohorts: the return window is 30 days, so April's figure is final but May's is not, and refund events have been sent since January, so both years are comparable. Split by product, one trouser fit accounts for 41% of April's returns, and 68% of those are coded "too small". A field jacket in sage green carries another 17%, mostly coded "not as pictured": the studio images show it greener than the cloth. Orders from the spring campaign return at 24% against 14% for other orders, and their returners often cite the waterproofing, which the ads called "fully waterproof" although only the shoulders are treated. Gift orders are 3% of April and change nothing. The brand adds an inside-leg measurement and a "cut slim; size up if between sizes" note to the trouser, reshoots the jacket in daylight with a colour card, and rewrites the campaign line as "showerproof shoulders". It also moves the size exchange above the refund in the returns journey. For June's closed cohort the rate is 14%, and exchanges rise from 22% to 47% of returned orders.

Common mistakes

Comparing a closed month with one still inside its return window. Rewriting the whole returns policy when the rise belongs to one style. Reshooting the range when the returns name one colour. Reading gift and personal orders as one group. Judging a campaign on platform revenue that still includes the parcels coming back.

Frequently asked questions

Is a higher return rate always a bad sign?

Not on its own. The returns policy guide treats a generous policy as a trade: it can lift conversion while also lifting the number of orders that come back. A rise that follows a deliberate change to the window or to free returns may be the price of extra sales, and the test is contribution margin per order after returns rather than the rate alone. A rise with no policy change behind it, concentrated in a few products or one campaign, is the kind this page diagnoses.

Should we count returns by order or by item?

Decide once and state it beside the figure. The return rate entry explains that an order rate counts an order as returned if anything in it came back, while a unit rate counts items and is the better reading when shoppers buy several sizes to try. Switching between the two between reports produces a rise or fall that never happened, so a sudden jump is worth checking against the counting basis before anything else.

How soon after a campaign can we read its returns?

Only once the return window for its orders has closed. Returns arrive days or weeks after the sale, so a rate calculated on recent orders is incomplete and usually low; the return rate entry asks for rates on cohorts whose window has closed, or labelled as provisional. Comparing a closed month with an open one makes the older period look worse, which is a common reason a recent campaign seems to have fewer returns than it will.

Terms used on this page

Where this page is used

Also related

Sources

  1. Measure ecommerce (GA4 developer guide) — Google for Developers
  2. Product data specification — Google Merchant Center Help

Related