What Most Stores Don't Know About Their Own Returns
Ask most e-commerce store owners what their return rate is by product. They'll tell you a number. Ask them to break it down by return reason, and things get murky. Ask them to correlate returns with specific product descriptions or images, and you've lost them entirely.
Returns data is almost universally bad in e-commerce — not because owners don't care, but because return reasons are collected inconsistently, stored poorly, and analyzed rarely.
Why Returns Data Gets Dirty
Reason codes are too broad: "Doesn't fit" and "Wrong size" are the same reason but stored differently. "Not as described" could mean anything from a color mismatch to a completely wrong product.
Customer-selected reasons are unreliable: When a customer clicks "Other" or picks the first reason from a dropdown, you're not capturing the real reason for the return.
Data lives in multiple systems: Returns managed in your platform, refunds processed in your payment gateway, return labels generated in a shipping tool — the data is fragmented and never joined.
Stop bad product data from reaching customers — Sohovi flags quality issues before they ship — try Sohovi free.
No consistent product-level tracking: A return is linked to an order, but connecting that order to a specific product variant and then to a specific product page version (with its images and description) is rarely done.
Building a Usable Returns Dataset
Step 1: Standardize reason codes Define 8–12 mutually exclusive reason categories. Map all existing codes to this standard. Going forward, enforce these categories in your returns portal.
Suggested categories: Wrong size/fit, Defective/damaged, Not as described, Changed mind, Arrived too late, Better price found, Gift – unwanted, Wrong item shipped, Other (with required text field)
Step 2: Join returns to product data Export your returns data and join it to your product catalog by product ID. Now you can see: what % of returns for Product X cite "Not as described"? That's a description or image problem. What % of returns for Product Y cite "Wrong size"? That's a sizing guide problem.
Step 3: Track return rate by variant The same base product in different sizes or colors may have wildly different return rates. A shirt that has a 4% return rate in Medium but a 22% return rate in XL has a sizing issue that's visible in the data — but only if you're tracking at the variant level.
Step 4: Enrich with customer-written reasons For every return where the customer wrote a free-text reason, read them. 100 free-text reasons will tell you things that 10,000 clicks on reason codes never will.
What Good Returns Data Unlocks
When your returns data is clean and joined to your product data:
- You can identify which products have description/image problems (high "not as described" rates)
- You can identify sizing issues before you order more inventory
- You can calculate true margin by product (revenue minus returns minus return shipping)
- You can catch defective batches early (sudden spike in "defective" returns for a specific SKU)
