Skip to main content
← All articles
COD & RTO · November 2026

Apparel Sizing and COD Returns: Fixing Fit Before Dispatch

12 minutes read
WHERE A SIZE ERROR GETS CAUGHT · CHEAPEST TO DEAREST1 · Product pagegarment measurementscost: near zero2 · Order timesize-risk flagcost: a message3 · Pre-dispatchconfirm size, edit ordercost: a dispatch day4 · Return legboth freight legs paidcost: fullWas the reason size, fit, or spec?Size → exchangesale kept, one leg already sunkSpec or quality → refundfix the listing, not the policyGate 1 is where the volume is. Gate 4 is where most brands are still spending.
Summary

In apparel, a return has two distinct origins: the customer never wanted the parcel, or the customer wanted it and it did not fit. COD controls address the first and do nothing for the second — so a brand that has fixed verification and still has a return problem is usually looking at a sizing problem wearing a COD costume. This covers how to separate size error from fit error from spec error, how to read variant-level return data you already have, how to build a size chart from garment measurements, and why routing size returns to exchange rather than refund is the single highest-value policy change in the category.

Apparel is the category where COD and returns compound worst, and the reason is structural rather than behavioural. Every other category sells a product whose fit is not in question: a phone case either goes on the phone or it does not. Apparel asks the customer to predict, from photographs, whether a specific garment will fit a specific body — and then, on a COD order, lets them settle that question at the door with no money committed. Two independent sources of return risk arrive in the same parcel, and brands routinely spend a year attacking one of them while the other produces most of the volume.

Two risks, one parcel

It is worth separating them properly, because they respond to completely different controls and they are measured in different places.

Intent risk is the question of whether the customer wanted the order at all. It shows up as refusal at the door, as unreachable phone numbers, as parcels that come back unopened. Verification, risk scoring, partial advance and courier allocation all attack this, and they work.

Fit risk is the question of whether the garment that arrived matches the body it was ordered for. It shows up after delivery, as a return request, or at the door when the customer has learned to open the parcel before paying. No amount of order confirmation touches it. A customer who enthusiastically confirms their order on WhatsApp and then discovers the trousers are two inches short is a verification success and a fit failure.

On COD the two fuse in a way that is easy to misread. A COD customer who suspects the size is wrong has a free option: refuse at the door and pay nothing. A prepaid customer with the same suspicion has already paid and has to go through a return. So the same underlying fit problem surfaces as an RTO on COD and as a post-delivery return on prepaid — which means a brand looking only at RTO rate will attribute fit failures to COD behaviour, and will keep tightening verification on a problem verification cannot reach.

The practical test is cheap: compare your return reasons on prepaid orders with your refusal pattern on COD orders in the same catalogue. If prepaid returns are dominated by size, your COD refusals are too, whatever reason the courier recorded.

Three errors that all look like “size issue"

"Size issue” is the most-used return reason in Indian apparel and the least useful, because it covers three failures with three different owners.

Size error. The customer chose the wrong size from a correct chart. They would have been fine in the next size up. This is a guidance problem: your chart, your fit notes, your size recommendation. It is the error most worth attacking because the customer still wants the product.

Fit error. The customer chose the right size, the garment matches the chart, and it still does not sit right — the cut does not suit that body, the rise is wrong, the shoulder is narrow for the chest. This is a product and merchandising problem. No chart fixes it; better photography, on-model shots at multiple body types, and honest fit language reduce it.

Spec error. The garment is not what the listing described. Fabric weight, colour, sheerness, length, finish. This is a listing-accuracy problem, and it is the one you should want to find, because it is fully within your control and it also drives the reviews that suppress future conversion.

If your return form offers one “size/fit issue” option, you cannot tell these apart, and the three have opposite remedies. Splitting that single option into “too small”, “too large”, “right size but the cut did not suit me” and “not as described” costs nothing and is the highest-leverage data change available in apparel. The first two point at your chart. The third points at the product. The fourth points at your listing.

The data is already in your order table

Before buying any fit technology, read what your own variants are telling you. The analysis that matters is return rate by variant, not by product — and most brands only ever look at the product level, which averages the signal away.

For each SKU, lay out units sold and units returned for every size. Four patterns appear, and each has a specific fix.

Returns concentrated in the smallest sizes usually means the garment runs small: customers who correctly picked their usual size received something tighter than expected. The fix is a chart correction or a “runs small, consider one size up” note, not a policy change.

Returns concentrated in the largest sizes often means the opposite, or that your grading is inconsistent at the top of the range — the pattern was scaled uniformly when it should have been re-cut.

Returns flat across all sizes is not a sizing problem at all. Flat distribution points at fit, fabric or spec: everyone is equally disappointed regardless of what they ordered.

Returns concentrated in one middle size is usually a production or labelling fault in one cut batch, and it is worth checking physical stock before touching the listing.

Two refinements make this far more useful. Separate first-time buyers from repeat buyers of the same product line, because a repeat buyer already knows how your garments run and their return rate is close to your floor — the gap between the two groups is the size of your guidance problem. And look at the exchange direction: if a measurable share of returns are exchanges upward, you have a chart that reads small, and that is a one-line fix.

Build the chart from garment measurements

Most size charts in Indian D2C are body-measurement charts: a table saying that size M corresponds to a 38-inch chest. These fail in a predictable way, because the customer has to know their own chest measurement, measure it correctly, and then trust that your M is cut for that body with the ease you intended.

Garment-measurement charts ask less. They state the flat dimensions of the actual garment — chest across at the underarm, length from high point of shoulder, shoulder seam to seam, sleeve length, waist, hip, inseam — and let the customer compare against a garment they already own and already like. That comparison is the one customers can actually perform, in their own home, with a garment whose fit they know. It removes the guesswork step that produces most size error.

Three things make such a chart work rather than merely exist. Measure real production samples rather than reproducing the tech pack, because the two diverge and the customer receives the former. State your tolerance openly — a half-inch to an inch of variance is normal in cut-and-sew production, and saying so prevents a return from being filed as a spec error. And publish the chart per style, not per brand: a single site-wide chart is a promise your oversized shirts and your slim trousers cannot both keep, and a shared chart is itself a return cause.

If you sell a style in a cut that differs from the category norm — oversized, cropped, relaxed, tapered — say it in the title and in the first line of the description, not only in a fit note three scrolls down. Customers who order an oversized tee in their usual size and receive an oversized tee have not made a size error; they have been under-informed about the cut.

Product page mechanics that reduce guessing

A few patterns consistently reduce size error, and none require new technology.

Put the model’s height and the size they are wearing next to every on-model photograph. This is the single most-used reference point customers have, and omitting it forces them to guess scale from the image. Show at least one garment on more than one body where you can; a single model shot teaches nothing about how the cut behaves.

Make the size guide open in place rather than navigate away — a customer who leaves the product page to find a chart frequently does not come back to it. Where you have real exchange data on a style, say so plainly in the fit note. “Customers most often exchange this style for one size up” is more persuasive than any chart, and it is a fact you already own.

Show out-of-stock sizes rather than hiding them. A customer who cannot find their size and orders the adjacent one is manufacturing a return for you, and they will usually tell you afterwards that they knew it might not fit.

The multi-size COD order

Customers in uncertain sizing order two sizes of the same garment intending to keep one. On prepaid this is a known pattern and the economics are tolerable, because the customer has paid for both and you are refunding one. On COD it is a different proposition: you ship two units, carry both forward legs, and collect for one at best.

Worth noticing that this behaviour is itself a signal. A customer buying two sizes is telling you your size guidance failed on that product. Before restricting the behaviour, look at which styles it clusters on — it is rarely uniform, and the styles where it concentrates are the ones whose charts need work.

Three ways to handle it, in increasing severity. Treat a multi-size cart as a verification trigger, so the order routes through confirmation where you can ask which size they actually want and offer the chart. Treat it as a prepaid-only cart, which is the honest version — the customer is explicitly buying optionality, and optionality on your freight is a thing you can reasonably decline to fund while keeping the sale available prepaid. Or allow it on COD above a margin threshold where the keep-one economics still work, and nowhere else.

What not to do is block it silently with a rule the customer cannot see. A cart that fails at payment with no explanation converts a size-uncertain buyer into a lost one, and they were a buyer.

Route size returns to exchange, not refund

This is where the largest recoverable value sits, and it is a policy decision rather than a technical one.

A size return is the one return reason where the customer still wants your product. They have chosen you, received the parcel, and the only thing wrong is a number on a label. If that interaction ends in a refund, you have paid both freight legs, taken back a unit, and lost the sale. If it ends in an exchange, you have paid the same freight and kept the revenue. Same cost, entirely different outcome — and the thing that decides which one happens is whether exchange is the default path or a thing the customer has to ask for.

Shopify’s own documentation treats this as a first-class case. Its returns and exchanges help page defines an exchange as “Send a customer an alternative item as part of the return, such as a different size or color”, alongside a return, defined as “Receive an item back from a customer, with optional return shipping labels and tracking”, and states that “You can manage returns and exchanges from the Orders page in your Shopify admin”. The capability is there; the question is whether your flow puts it in front of the customer first.

Practical design points. Ask the reason before offering the resolution, so a “too small” answer can lead with the next size up rather than with a refund button. Hold the replacement unit when the exchange is raised rather than when the return is received, because the most common reason an exchange collapses into a refund is that the size went out of stock while the parcel was in transit. And make a same-style, same-price exchange free where you can afford it, because the alternative is a refund that costs you the same freight and the whole order value.

One honest caveat: exchange only works if the replacement is genuinely available. A brand running thin on size depth will promise exchanges it cannot fulfil, and a failed exchange is worse than a clean refund. Check your depth by size before making exchange the headline policy.

Where COD controls still belong

None of this means the COD-side levers are irrelevant in apparel — it means they should be pointed at the intent half of the problem and at the orders where fit risk is highest.

Size-uncertainty is a legitimate input to your COD rules. A first-time buyer, ordering a style with a known exchange-upward pattern, at the edge of your size range, on a high-value basket, is a different proposition from a repeat buyer reordering a size they have already kept. The first deserves a confirmation step that includes the size question; the second deserves to be left alone, and on most apparel catalogues the second group is larger than the controls assume.

The pre-dispatch window is also the last cheap moment to fix a size error. An order confirmed in a conversation where the customer mentions they are between sizes can still be edited before packing — and an edit at that point costs a dispatch day, against both freight legs plus a unit cycle if you catch it after delivery.

Where COD King fits: COD rules by product, order value and pincode, plus confirmation before dispatch, are the mechanisms behind the multi-size restriction and the size-check conversation described above. The chart, the return-reason taxonomy and the exchange policy are yours, and they are where most of the gain in this category actually comes from.

The return leg needs its own QC

Apparel is the category where returned stock most often cannot go back on the shelf, and the losses here are easy to under-count.

Set a written grading standard before volume arrives — resellable, needs cleaning or pressing, discountable, write-off — and grade every unit against it rather than by the judgement of whoever opened the box. Record the grade against the original order, because that is what turns “we get returns” into a per-style loss figure you can act on. A style with a high return rate and a high resellable share is an annoyance; a style with the same return rate and a low resellable share is losing you inventory, and the two should not share a strategy.

Photograph the unit at intake on any style where wear-and-return is a pattern. And handle the time dimension: in a seasonal category, a unit that comes back in good condition six weeks later is still a markdown, and counting it at full cost overstates your recovery.

What to track

Five numbers tell you whether the sizing work is paying.

Return rate by variant, with the size distribution visible, because the shape of that distribution is what names the problem. The split of your return reasons across size, fit and spec, which is only available if you split the form. Exchange share of size returns, which is the direct measure of whether the policy change landed — it should move within weeks of making exchange the default, and if it does not, the flow is still burying it. Resellable share of returned units by style, which separates expensive returns from cheap ones. And the gap between first-time and repeat buyer return rates on the same style, which is your best single estimate of how much of the problem is guidance rather than product.

Review these per style rather than per catalogue. Sizing problems are not evenly spread, and a catalogue average will hide the four styles doing most of the damage — which, in most apparel catalogues, is roughly what you will find when you look.

The short version

Apparel carries two independent return risks, and COD controls only address one of them; a brand with good verification and persistent returns is usually looking at a sizing problem. Split “size issue” into size, fit and spec before anything else, because the three have different owners and opposite fixes. Read return rate by variant rather than by product, and use the size distribution to name the cause. Publish garment measurements per style, measured from production samples, with your tolerance stated, so customers can compare against something they already own. Put model height and worn size on every on-model image, and show out-of-stock sizes rather than pushing customers into the adjacent one. Treat multi-size COD carts as a guidance failure first and a policy problem second. Make exchange the default resolution for size returns, because the freight is already spent and the customer still wants the product. Grade returned units against a written standard and record the grade against the style. Then measure the first-time versus repeat buyer gap, which is the clearest number you have for how much of this is your chart rather than your product.

Quoted definitions of returns and exchanges are from Shopify’s Returns and exchanges help page, checked 20 November 2026. Platform behaviour changes; verify against the current page. No return-rate or sizing benchmarks are quoted here because we could not source any to a named organisation we would stand behind — the patterns described are mechanisms to test against your own variant data.