Ecommerce Returns Analytics: Why Your Return Rate Hides the Real Problem

Your ecommerce return rate tells you that returns happened. It does not tell you what to fix. Ecommerce returns analytics is the work of breaking that blended return rate into diagnostic cuts, so an operator can see whether the problem sits with a customer cohort, a specific SKU, a likely cause, a seasonal swing, or the recovery value lost after an item comes back. A blended figure like a 22% return rate is a warning light, not a diagnosis: it cannot say whether the shift came from new customers who don’t yet trust your sizing, from a handful of SKUs a merchandising team should have pulled months ago, from a fulfillment mistake, from a seasonal swing that one month exaggerates, or from returned inventory sitting too long before it recovers value. Skip that split, and a fix aimed at the average risks solving a problem that was never actually there.

For ecommerce operators, retail brand managers, merchandising teams, fulfillment and reverse logistics personnel, and product quality teams, that distinction changes real decisions about assortment, fit content, warehouse process, and resale recovery. Komar’s event data from Cahoot’s August 2026 Ugly Talk series shows what this looks like in practice. Before Carve Designs did combined fit and purchase-path work, its new-customer return rate ran around 35% and its repeat-customer rate around 25%, a wide gap sitting quietly inside whatever single return-rate number appeared on Carve’s dashboard. After that work, new-customer returns fell to roughly 20% and repeat-customer returns to about 12%. A brand watching only the blended average would have seen one figure move and had no way to know which customers, or which fix, actually mattered.

The rest of this guide walks through those five cuts—customer cohort, SKU, return reasons, seasonality, and recovery value—and then separates diagnosis from the outcome KPIs a team tracks once it knows what it is fixing.

Key takeaways

• A blended return rate is a warning light, not a diagnosis. It shows scale, not cause.
• Splitting return rate by new versus repeat cohort often reveals two different businesses in one number, as Carve Designs’ shift from 35%/25% to 20%/12% shows.
• Ranking return rate by SKU usually finds a small number of products, not a random cross-section of shoppers, driving most of the volume.
• Reason codes are a starting clue, not a verdict. Customers often mislabel why they’re actually returning an item.
• Seasonality and recovery value, what a returned unit is worth once sellable again, both change what the same headline rate means.
• Diagnosis identifies the problem. Outcome KPIs measure whether the fix works, and the two belong in separate conversations.

Headline view	Diagnostic cut	What decision it changes
Company-wide return rate	New vs. repeat customer cohort	Acquisition, onboarding, and fit-confidence investment
Overall return volume	Return rate by SKU / variant	Product, sizing, grading, and assortment decisions
Top return reason	Customer reason plus inspection evidence	Product quality, content, fulfillment, returns fraud and refund fraud prevention controls

| Monthly average | Comparable periods / seasonality | Inventory, staffing, and campaign planning | | Resale / restock rate | Recovery per unit net of cycle time | Routing, processing speed, markdown exposure |

A blended return rate tells you scale, not cause

U.S. retailers were on pace to process $849.9 billion in returned merchandise in 2025, with an estimated 19.3% of online purchases sent back, according to NRF and Happy Returns research. Another benchmark shows 92% of global shoppers return up to 30% of online purchases, which underscores how routine returns are in e-commerce. That figure confirms returns are a permanent, material line item in ecommerce. It says nothing about any single brand’s problem, because a market-wide aggregate blends every category, business model, and customer base into one number. For most retailers, headline return rates still miss context such as why customers bought in the first place, and 88% of customers check return policies before purchasing.

Apparel makes the point clearly. Fit can’t be verified until a product is on a body, so apparel and footwear carry some of the highest return rates in ecommerce. Coresight Research, sponsored by sizing-technology vendor 3DLOOK, put the U.S. online apparel return rate at 24.4%, with size and fit cited by 53% of surveyed brands and retailers among top return reasons, and estimated 2023 online apparel returns at roughly $38 billion, with about $25.1 billion in processing costs attached. A beauty or home-goods brand won’t see anything close to that rate, and comparing rates across categories without adjusting for that is a common reporting mistake; understanding why ecommerce return rates are rising helps teams avoid drawing the wrong conclusions from a single benchmark.

Cahoot’s breakdown of the average ecommerce return rate by category is a useful ceiling check for whether an ecommerce business sits inside a normal range. Its companion piece on how return rate affects profit margins explains why that blended number is dangerous to plug directly into a margin model. Resources on crafting the perfect ecommerce returns program pick up from there, but neither answers the operator’s real question: what inside that number needs to change.

Carve Designs shows why customer cohorts must be separated

New and repeat customers are not the same population wearing the same size. A new customer is guessing at fit for the first time and may be buying on impulse or as a gift. A repeat customer already knows how a brand’s sizing runs and buys with more confidence, and that matters because return behavior can shape customer lifetime value, not just the immediate return rate. Blending their return rates into one company-wide figure erases that difference, along with the clearest signal in the data.

Carve Designs’ cohort numbers show the gap in practice. Per the Komar event deck presented by Jay Harris at Ugly Talk NYC, new-customer returns ran around 35% and repeat-customer returns around 25% before Carve’s combined fit and purchase-path work; afterward, the same cohorts fell to roughly 20% and 12%. About one in five Carve shoppers now completes the brand’s proprietary swim-fit quiz before buying, and the event deck reports higher average order value and lower returns among that group, without isolating an exact reduction attributable to the quiz alone. More personalized post-purchase and return experiences based on customer differences can strengthen customer loyalty, and an exceptional returns program turns those operational choices into a retention asset. The larger cohort-level gain came from pairing that guidance with changes across the purchase path, not one feature working in isolation, and a smooth return experience can increase customer lifetime value.

It’s worth being direct about what 35% means here. As Jay Harris put it, a 2% return rate and a 35% return rate can represent completely different business models. Bad SKUs, poor sizing charts, weak construction, and design choices that don’t match how customers use a product can all push a headline number up. Treating 35% as simply too high, without asking why, misses the point as much as treating it as acceptable would. The cohort split turns that number into a question worth answering, especially when longer windows can improve customer confidence and when policy choices need to match customer expectations if you want retention, not just a lower rate.

The SKU leaderboard tells you what the company average cannot

A company-wide return rate can sit at a stable, unremarkable number while a small set of products quietly drives most of the volume behind it. The Komar deck put this bluntly: “Repeat offenders are SKUs, not shoppers.” The recommendation is straightforward: rank return rate by SKU, not just by category or channel, and use return analytics to track product, variant, order, customer, and reason level data so you can investigate the styles that keep coming back rather than assuming the problem is spread evenly across the catalog.

This doesn’t mean every brand has the same concentration pattern; there’s no fixed share of returns that always sits in the worst-performing SKUs. What’s consistent is the habit: a SKU-level leaderboard turns a vague “returns are up” conversation into a specific one about a style, size grade, fabric, or listing, with the key metrics needed to understand product performance. Cahoot’s guide to why ecommerce returns run high covers the product, content, and fulfillment drivers that tend to concentrate in a handful of SKUs, pointing merchandising toward a sizing correction, a copy update, or pulling the SKU from the assortment.

Return reason codes are clues, not verdicts

Most returns platforms ask a customer to select a reason or submit a return request: wrong size, changed mind, item not as described, defective. Those codes are useful as a starting point and unreliable as a final answer. Jay Harris described selected reason codes as frequently wrong in his own operating experience, summarized on the Komar deck in one line: “Reason codes lie. Garments don’t.” A customer who feels awkward admitting a product wasn’t as flattering as expected may select “changed my mind” instead, and one returning a defective item may just pick whichever option sits first in the list. That’s an attributed operating observation, not a published industry statistic, but it argues for treating a reason code as a clue, not a number to report at face value.

The correction is inspection. When merchandise physically comes back, someone can look at the item itself: is it worn, damaged, mismatched to the order, or genuinely defective? Carrier tracking and return processing time help validate what happened across the entire process, from initiation to completion. Pairing the stated reason with what the item shows closes the gap between what a shopper says and what happened. Cahoot’s guide to reducing returns using customer feedback goes deeper on turning that combined data into prevention work, and platforms like Return Prime’s Shopify-focused returns solution can operationalize those insights for smaller brands without in-house reverse logistics.

Kulfi shows how returns data can expose a product defect

Return data isn’t only an operations signal; it’s a product-development signal, and Kulfi Beauty’s experience shows why. Speaking on Ugly Talk, Kulfi’s Gabrielle Kerins described a lip product whose packaging had passed quality control before launch. Consistent return feedback on that one product eventually revealed the real problem: the packaging was temperature-sensitive and leaked under certain conditions, something a standard QA pass hadn’t caught. Kulfi used that pattern to redesign the packaging rather than treating the returns as ordinary buyer’s remorse, illustrating how convenient drop-off networks like Happy Returns’ reverse logistics solution can surface recurring issues quickly when feedback and inspection data flow back to product.

That sequence only works if return data reaches product and quality teams, not just the returns desk. A defect showing up as a handful of “damaged” or “wrong item” codes each week can look like noise in a dashboard and a clear pattern once someone maps it back to a single SKU and root cause, using real-time visibility to spot return trends and customer behavior patterns faster. Treating returns data as an input to product development, not just a cost center, is what turned Kulfi’s leaky packaging into a fixed product in a data-driven way that supports smarter decisions.

Seasonality can make the same headline rate mean something different

A return rate is not a fixed characteristic of a brand. It moves with the calendar, and a trailing average can smooth away the exact months where the economics spike. Gabrielle Kerins made this point directly on Ugly Talk: return rate is not static. Manish Chowdhary added the operator framing on the same panel: an annual average can hide the specific months where returns jump, whether from a holiday gifting surge, a size-run change between seasonal collections, or a spike in first-time buyers from a new marketing channel.

The fix isn’t to distrust monthly reporting; it’s to compare like periods against like periods. A December return rate should be measured against last December, not the trailing twelve-month average, and when seasonality and channel mix move together, the comparison should also be broken out by sales channels. A spike tied to a product launch should be evaluated against that launch’s own cohort. What drives a spike in one category during one season won’t generalize to every brand or month, which is why the comparison has to be specific rather than assumed.

Recovery value changes while the returned item is in motion

A returned unit’s value isn’t fixed at the moment a customer requests a return. The value of returned items decays the longer that item takes to travel back, get inspected, and become sellable again, so recovery value and cycle time have to be measured together for better inventory management decisions by product category.

McKinsey’s research on apparel returns management found the difference between a retailer’s least and most expensive return channel averaged $5 to $6 per unit, and that in-store processing could save up to 18 days compared with warehouse processing, improving the odds an item resells at full price rather than at a markdown. Full return costs also include shipping and restocking labor, so tracking the complete operational cost supports more rational return window and return policy choices, especially when weighing the true cost of offering free returns. Those days are the gap between an item back on a shelf at full margin and one reaching a liquidator after a season has turned, with consequences for markdown exposure and the broader supply chain. Cahoot’s breakdown of the hidden economics of a return walks through the full cost stack this section only touches.

Use five return analytics diagnostic numbers before changing policy or product

Pull these five measurements before rewriting a return policy, redesigning a product, or restructuring reverse logistics. Together, they make up the diagnostic scorecard Komar presented at Ugly Talk NYC, useful as a working method rather than a universal Cahoot KPI standard every brand must adopt.

• Return rate split by customer cohort, especially new versus repeat, to separate an acquisition and fit-confidence problem from a retention problem.
• Return rate by SKU, ranked, to identify products carrying disproportionate return exposure instead of assuming volume spreads evenly.
• Recovery per returned unit, net of cycle time, to capture what an item was actually worth once it became sellable again, not what it was worth on the day it shipped.
• Reverse cost per unit versus forward cost per unit, to show how policy choices affect both cost recovery and customer experience instead of treating the return journey as a fixed cost.
• Share of returns the brand caused, covering wrong item, damage, lateness, misleading copy, fit or shade guidance, and product or packaging defects.
• Exchange-versus-refund mix, because exchanges retain revenue that refunds surrender and often correlate with higher satisfaction, creating a more positive experience.

Reading these five together, rather than one at a time, turns a single return-rate headline into a specific decision about acquisition, product, quality control, fulfillment, or reverse-logistics routing, including whether policy, routing, or store credit can improve recovery.

After diagnosis, manage the outcome KPIs separately

Effective ecommerce returns management separates diagnosis from outcome measurement, and conflating them is how a returns program tracks the wrong thing. The five-number scorecard above explains what’s driving a return-rate change. Once diagnosis points at a cause, outcome KPIs measure whether the response is working.

Cahoot’s guide to the KPIs that actually matter for modern returns management owns that second layer, covering metrics like refund time, share of returns eligible for peer-to-peer resale, and net cost per order. Predictive insights, machine learning, and fraud detection in modern returns management software can automate ecommerce returns management and cut handling costs by up to 40%. Those metrics show whether a returns operation is executing well; they don’t explain why the underlying return rate moved, which is the gap diagnosis closes first.

Cahoot is an end-to-end e-commerce returns management and fulfillment operations suite built around a simple principle: save every penny a returns process doesn’t need to spend while supporting brand reputation and customer satisfaction when execution is strong. A brand that has diagnosed where its returns value is actually being lost is better positioned to use a recovery lever like Cahoot’s Peer-to-Peer Returns or broader returns management software, which routes eligible items toward new demand instead of a full warehouse cycle, on the SKUs and cohorts where routing will matter most to enhance customer satisfaction.

Frequently Asked Questions

What is the best way to analyze an ecommerce return rate?

The best approach is systematic return analytics: split the blended rate into cohort, SKU, cause, seasonality, and recovery-value cuts rather than reacting to the headline number, and track product, variant, order, customer, and reason as core key metrics. Each cut points to a different owner and fix, whether that’s acquisition, merchandising, product quality, fulfillment, or reverse-logistics routing.

Why can an average ecommerce return rate be misleading?

An average blends every customer, product, season, and cause into one figure, so it can stay flat, rise, or fall for different reasons underneath, including shifts across customer segments and changes in customer behavior. A brand can show a stable company-wide rate while one cohort or a handful of SKUs drives most of the actual volume and cost.

Should ecommerce brands track return rate by new and repeat customers?

Yes. Repeat customers already know a brand’s fit and sizing, while new customers are often guessing for the first time. High return rates do not automatically lower customer lifetime value when the return experience is smooth. Carve Designs’ shift from a 35% new-customer and 25% repeat-customer return rate to roughly 20% and 12%, following combined fit and purchase-path work, shows how differently those cohorts can move. Brands can gauge the retention impact with Net Promoter Score.

How do you calculate return rate by SKU?

Divide units returned for a SKU by units of that SKU sold over the same period, then rank every SKU from highest to lowest. The goal is a ranked list showing which products drive disproportionate return volume, so a team can investigate the specific style, size grade, product page, or product descriptions.

Can return reason codes be inaccurate?

Yes. Komar’s Jay Harris described selected reason codes as frequently wrong in his own operating experience, using the shorthand “reason codes lie, garments don’t” to describe the gap between what a customer selects and what inspecting the item actually shows. Authentic negative reviews can also help validate whether fit or quality complaints are isolated or recurring.

Which returns metrics should ecommerce brands track beyond return rate?

Once diagnosis identifies the cause of a return-rate change, brands should also monitor self-service portal workflows, real-time tracking, and how they communicate proactively, alongside outcome metrics like refund time, share of returns eligible for peer-to-peer resale, and net cost per order. Many teams also track fraudulent returns, patterns involving multiple items, and the effect of free return shipping because they shape both cost and customer experience. Cahoot’s guide to the KPIs that actually matter for modern returns management covers that layer in detail. Automation can reduce handling costs by up to 40%.

• Category: Uncategorized
• Tag: Order Fulfillment, Returns, Strategy