On this page
Most merchants treat returns as a store-wide average, but averages hide the truth: a small cluster of SKUs is almost always responsible for the bulk of fit-related returns. If you don't isolate those products, you'll keep tweaking marketing and shipping while the real problem sits untouched on your product pages.
Why a Single Average Return Rate Misleads You
A blended return rate across your catalog tells you almost nothing actionable. A store running at a healthy 8% average can still have five SKUs running at 35-40%, quietly eating margin while hundreds of other listings perform fine. Retail data consistently shows that returns concentrate rather than distribute evenly, and apparel is one of the worst offenders because fit — unlike color or material — cannot be verified from a photo. According to industry analysis compiled by the National Retail Federation, apparel remains among the highest-return categories, and a large share of those returns trace back to size and fit rather than damage or preference.
Treating every SKU the same means you're either over-investing in fixes nobody needs, or ignoring the sizing problem sku that's bleeding revenue every day it stays live.
The Cost of Not Isolating the Problem
Every day a broken-fit SKU stays undetected, you pay three times: the cost of the return shipment, the cost of the restock or write-off, and the lost sale from a customer who won't reorder in "their size" again. Multiply that by weeks of undiagnosed traffic and the number gets large fast — especially on SKUs with high unit volume.
Building Your Sizing-Problem SKU Diagnostic
A reliable diagnostic isolates the specific products whose fit is broken by combining three signals: return rate versus peers, size-purchase skew, and review sentiment. No single signal is enough on its own.
Signal 1 — Return Rate vs. Peer Group
Compare each SKU's return rate not against your store-wide average, but against a peer group of similar products — same category, similar price point, similar fabric. A men's slim-fit shirt should be benchmarked against other slim-fit shirts, not against your loose-fit hoodies. Any SKU sitting more than 1.5x above its peer group median is a candidate worth flagging.
| Signal | What it measures | Red flag threshold |
|---|---|---|
| Peer-relative return rate | SKU return % vs. category median | 1.5x peer median or higher |
| Size purchase skew | Distribution of sizes ordered vs. expected bell curve | Heavy lean to one edge (e.g. XL/XXL) |
| Review sentiment | Frequency of "runs small/large" mentions | 3+ mentions per 20 reviews |
| Exchange-to-return ratio | Exchanges vs. outright returns for size reason | Low exchange rate = customer gives up |
Signal 2 — Size Purchase Skew
Plot the distribution of sizes actually ordered for a SKU. A healthy product roughly follows your store's typical size curve. A sizing problem sku often shows a skew — customers compensating by ordering up or down from what they'd normally buy, followed by a return once the garment arrives. If your data shows a spike toward XL on a product where your store average sits at M/L, that's a strong signal the labeled sizing runs small, not that your customer base changed.
Signal 3 — Review Sentiment Mining
Reviews are a free, high-signal dataset most merchants under-use. Search review text for phrases like "runs small," "runs large," "true to size," and "had to return." A product with three or more size-complaint mentions per twenty reviews is very likely a sizing problem sku, even if its return rate hasn't fully caught up yet — sentiment often leads the return-rate data by weeks, since not every dissatisfied buyer files a formal return.
Turning the Diagnostic Into Action
Once you've built this three-signal view, rank every SKU by a composite score and triage the top decile first. This is the same discipline covered in how to spot fit problems before they become returns — catching the pattern before it compounds into a costly return cycle.
Prioritizing the Worst Offenders
Not every flagged SKU deserves the same urgency. Weight your triage list by return volume, not just return rate — a SKU with a 40% return rate on 20 orders a month matters less than one with a 20% rate on 2,000 orders a month. This is exactly the kind of prioritization outlined in the metrics that predict a high-return product, where volume-weighted scoring consistently outperforms rate-only rankings.
Fixing the Size Chart, Not Just Flagging It
Diagnosis without correction doesn't move the needle. Once a SKU is confirmed as a sizing problem, the fix is usually one of three things: correcting the size chart against actual garment measurements, adding fit guidance copy ("runs small, size up"), or integrating a fit-recommendation tool at the product page so shoppers get a personalized suggestion instead of guessing from a generic chart. Merchants using YourSizer typically see the review-sentiment signal drop within a few weeks of deploying a corrected size recommendation on a flagged SKU, since the tool intercepts the wrong-size purchase before it happens rather than after the return request is filed.
Monitoring After the Fix
A diagnostic isn't a one-time exercise — it's a recurring audit. Re-run the three-signal check monthly, especially after restocks, supplier changes, or fabric substitutions, since any of those can silently reintroduce a fit problem on a SKU that was previously healthy. Products with a documented product-level fit score explained make this recurring check far faster, since the score already aggregates the underlying signals into one trackable number per SKU.
Setting Alert Thresholds
Set an automated alert when a SKU's peer-relative return rate crosses 1.5x for two consecutive weeks, or when size-complaint mentions in reviews cross the 3-in-20 threshold. Catching the drift early is cheaper than catching it after a quarter of write-offs.
Frequently Asked Questions
How many SKUs typically cause most sizing returns?
In most apparel catalogs, a small fraction of SKUs — often under 10% — account for the majority of size-related returns. This concentration pattern holds across categories and price points, which is why a peer-relative diagnostic is far more useful than a store-wide average.
What return rate counts as a sizing problem sku?
There's no single universal number, since acceptable return rates vary by category and price point. The more reliable signal is relative: a SKU sitting at 1.5x or more above its peer group's median return rate, combined with a skewed size-purchase pattern, is a strong candidate regardless of the absolute percentage.
Can review sentiment really predict returns before they happen?
Yes, and it often predicts faster than return-rate data alone. Not every dissatisfied buyer requests a formal return, but many will still mention "runs small" or "runs large" in a review, which means sentiment mining can flag a sizing problem sku weeks before the return numbers fully reflect it.
Do I need special software to run this diagnostic?
You can start with a spreadsheet pulling return reason codes, size-order distribution, and review text, but the process becomes far faster with a dedicated fit-analytics tool that automates the peer benchmarking and sentiment scan, especially once your catalog grows past a few hundred active SKUs.