Turning returns data into margin: sizing, quality, and bracketing

Fit. "Too small," "too large," "didn't fit." This is a size guide and product spec problem.

Quality. "Defective," "not as described," "poor quality," "color off." This is a sourcing, QA, and product-detail-page problem.

Bracketing. The behavior described in the opening scene: a shopper deliberately buys multiple sizes or colors intending to keep one and return the rest. This is a policy and pricing problem.

Bracketing (ordering variants of the same item to compare at home, then returning most) is worth defining clearly because it hides inside your "fit" returns. A customer who orders three sizes and returns two will often tag both returns as "wrong size," even though only one was ever going to be kept.

Building the dataset

You need to join three tables that usually live in different systems:

1. Orders (order ID, customer ID, SKU, size, color, price paid, date).

2. Returns (order ID, SKU, return reason, return date, condition on arrival).

3. Product master (style ID, category, garment measurements, supplier, landed cost).

The key trick: group SKUs by style ID, not just SKU. A single dress style has many size and color SKUs. Bracketing only becomes visible when you look at the style level within a single order and customer.

Here is the core logic for flagging bracketing orders.

Once you can flag bracketed style-orders, you can measure how they behave versus single-variant orders: return rate, keep rate, and margin.

Quantifying true landed margin

Gross marginGross marginGross margin is the share of revenue left after subtracting the direct cost of producing goods or services, expressed as a percentage of revenue.View full definition → per unit sold is the wrong number. You want contribution margin per order after returns, sometimes called landed or realized margin.

Build it up cost by cost:

• Revenue kept = price paid on units not returned.
• Landed product cost = cost of goods for all units shipped (returned units still cost you outboundoutboundProactive outreach that pushes your message to targeted audiences through advertising, email, or direct prospecting, initiated by the seller rather than the buyer.View full definition → freight and handling even if the refund is issued).
• Return handling cost = inboundinboundA strategy that attracts prospects organically via valuable content (blog, SEO, social) rather than interrupting them.View full definition → freight plus inspection plus restock or refurbishment. A common planning estimate for apparel is several dollars per returned unit, but measure your own.
• Markdown loss = returned items often re-enter inventory late and sell at a discount, or not at all.

A simple per-order model:

Run this across order types and the pattern usually jumps out: bracketed orders can convert well on revenue while destroying margin, because you paid to ship and process two or three units to earn the margin on one.

For a solid primer on the economics of returns, the National Retail Federation's returns research is a useful free reference that is refreshed annually.

Turning fit returns into a better size guide

Fit returns are the most fixable, and the payoff is durable. The signal is in the size migration pattern: when someone returns a medium as "too small" and reorders a large that they keep, that medium ran small.

Step 1: Build a size-accuracy score per style

For each style, count fit returns split by direction:

• "Too small" returns suggest the garment runs small (customer needs to size up).
• "Too large" returns suggest it runs large.

A style with 80 percent of fit returns tagged "too small" is not a random-fit problem. It is a spec problem. The garment is consistently smaller than the size label implies.

Step 2: Compare against measured garment specs

Pull the actual garment measurements from your product master (chest, waist, inseam) and compare across styles in the same size. You will often find that "size M" varies by several centimeters between suppliers. Customers cannot see this, so they guess, and they guess wrong.

Step 3: Rewrite the guide

Concrete fixes that move numbers:

• Add a "fits small, we recommend sizing up" note on styles with lopsided fit returns.
• Publish flat garment measurements, not just body measurements.
• Standardize specs across suppliers for the same size label.
• Add fit feedback prompts at return so you capture direction cleanly.

Turning quality returns into sourcing decisions

Quality returns cluster by supplier and style, not by customer. That makes them actionable in procurement.

Aggregate return reason by supplier:

• If one supplier's styles show elevated "defective" or "color off" returns, that is a QA conversation with hard data behind it.
• If "not as described" is high, the problem may be your product page, not the garment. Photos, color rendering, and fabric descriptions are cheaper to fix than a supplier relationship.

The discipline here is separating product problems (fix the source) from content problems (fix the page). Both show up as returns. Only the data tells them apart.

Managing bracketing without killing conversion

Bracketing is the trickiest because the customer behavior is rational and often produces a happy keeper. You do not want to punish good customers. You want to reduce the cost of the behavior.

Options, from lightest to heaviest touch:

Reduce the need to bracket. Better size guides and virtual fit tools mean customers order one size with confidence. This is the win-win path: fewer returns, same revenue.

Nudge at checkout. When a customer adds two sizes of one style, a gentle "Not sure of your size? Check our fit guide" can convert a bracket into a single confident purchase.

Segment your policy. Some retailers now identify chronic high-return accounts and adjust terms (for example, charging return shipping) while protecting the majority who return rarely. This is a sensitive area: be transparent, apply rules consistently, and follow the consumer-protection rules in each market you sell in. Do not treat this as legal advice; involve counsel before changing published return terms.

Measure the keeper margin. Some bracketed orders still net out profitable because the keeper is a high-margin item. Segment before you act. The goal is margin, not a lower return rate for its own sake.

A simple decision frame

For each style, place it on two axes: return rate and realized margin.

• High return, low margin: prioritize (fix fit or reconsider carrying it).
• High return, high margin: fix fit, keep selling.
• Low return, low margin: pricing or cost problem, unrelated to returns.
• Low return, high margin: protect and expand.

This frame keeps teams from chasing returns everywhere and lets them attack the styles that actually bleed.

Key Takeaways

• Revenue hides the leak; landed margin exposes it. Always model contribution margin after return shipping, handling, and markdown, not gross margingross marginGross margin is the share of revenue left after subtracting the direct cost of producing goods or services, expressed as a percentage of revenue.View full definition → per unit sold.
• Group by style ID, not SKU, to see bracketing. Multiple size or color variants of one style in a single order is your bracketing signal, and it often masquerades as "wrong size."
• Fit returns are a spec problem you can fix. Lopsided "too small" or "too large" returns point to inconsistent garment measurements; publish flat measurements and standardize specs across suppliers.
• Separate product-quality returns from content returns. "Defective" points to sourcing; "not as described" often points to your product page, which is far cheaper to fix.
• Target margin, not the return rate. Rank styles by return rate and realized margin together, and act on the high-return, low-margin styles first.