The State of Shopify Return Policies 2026
We analyzed the return policies of 110 real stores: 52 national brands and 58 small Shopify stores, the latter compiled line by line through a production policy engine. This is what online stores actually promise, where their policies go silent, and what that silence costs when support gets automated.
By Ryan Dusterhoft, Founder and CEO of Resolvas
Published September 1, 2026 · Updated September 2026
Everyone can read Sephora's return policy. Nobody has data on what the other million Shopify stores promise, because nobody reads 58 small-store policies line by line. We did, the hard way: we captured each store's actual policy pages and compiled them through the same policy engine our support agent runs in production. The engine only records a rule when the store's own words state it, so every number below is what the policies say, not what we assume.
Three findings stand out. Small stores give customers less time and less help than the big brands they imitate. The average policy is thousands of words long yet answers surprisingly few operational questions. And the questions it skips are exactly the ones a support ticket forces someone to answer within the hour.
The average policy: 3,900 words, 10 enforceable rules
The median policy corpus in our small-store sample runs about 3,900 words across roughly three pages (returns page, FAQ, terms). Compiled down to rules a support system could actually enforce, the median store yields just 10 concrete fields: a window, a shipping stance, an exclusion list, and a handful of conditions. Three stores in the sample, about 5%, compile to zero enforceable rules; their entire returns section amounts to "email us and we'll sort it out."
That gap between word count and rule count is the study's central finding. Policies are written as prose reassurance, not as decisions. The length hides the silence.
Return windows: small stores give you less time
Among the 84% of small stores that state a window at all, the spread looks like this:
| Stated return window | Share of small stores (n=49 stating one) |
|---|---|
| 7 to 15 days | 33% (16 stores; 14 days alone accounts for 10) |
| 21 to 28 days | 4% (2 stores) |
| Exactly 30 days | 53% (26 stores) |
| More than 30 days | 10% (5 stores; one 60, one 90, one full year) |
The 30-day norm holds at both ends of the market: 53% of small stores that state a window pick 30 days, and our June 2026 survey of 52 national brands found 52% at 30 days (Lululemon, Gap, Sephora, Allbirds, and most of the apparel mainstream). The difference is the tail. At national brands the tail runs long: 45 to 100-day windows at ASOS, Revolve, Bose, and Casper, and effectively unlimited at Patagonia, Nordstrom, and IKEA. At small stores the tail runs short: 37% of stated windows are under 30 days, with 14 days the second most common choice. 16% of small stores never state a window at all.
Who pays return shipping: the market splits in half
This is where small stores and national brands genuinely diverge:
| Return shipping stance | Small stores (n=58) | National brands (n=52, June 2026) |
|---|---|---|
| Customer pays | 48% | ~25% |
| Merchant pays (free returns) | 17% | ~37% |
| Customer pays unless our fault | 9% | ~33% (conditional or fee-deducted) |
| Policy never says | 26% | rare |
Free returns are a big-brand luxury. At small stores the modal answer is "you pay the postage," only 22% offer a prepaid label anywhere in their policy, and a quarter never address the question. Fees follow the same pattern: 12% of small stores charge a restocking fee (5% to 20% of the item), and 17% charge a flat return fee between $4 and $40.
The silence problem
For each policy we recorded not just what it says but what it never answers. These are the silence rates across the 58 small-store sample, in descending order of operational pain:
| Question a support ticket will eventually force | Policies that never answer it |
|---|---|
| Damaged item: does the customer return it or keep it? | 84% |
| Refund destination: original payment or store credit? | 84% |
| Condition evidence: is a photo required? | 72% |
| Refund timing: before or after the item comes back? | 55% |
| Are exchanges offered at all? | 50% |
| Who pays return shipping? | 26% |
| Any return window stated | 16% |
The exchange row deserves a second look: 29 stores say yes to exchanges and 29 say nothing, but not one store in the sample says no. Silence is doing the work of policy. Every one of these unanswered questions becomes a judgment call made ticket by ticket, usually by whoever happens to be answering email that day, and increasingly by an AI agent that was never told the answer either.
The odd rules are load-bearing
36% of small stores carry at least one value-gated rule: a threshold where the policy changes because the order is worth more. We found gates at $50, $99, $150, $350, $500, and one jewelry store requiring signature confirmation on orders over $2,000. 9% mention a signature requirement outright. These rules exist because someone got burned once, and they are precisely the rules a generic support tool has never heard of.
The keep-it economy is bigger than expected too: 34% of the sample is keep-it or guarantee-shaped somewhere in its policy. Refund the supplement, do not mail it back; satisfaction guarantees on first orders; "no need to return" language for low-value or perishable goods. At the national-brand tier the same shape shows up at about 13% (Seed, Olipop, Ritual, Thrive). It is a rational answer to reverse-logistics costs, and it is also, unmanaged, an invitation to serial refunders.
What happens when support meets these policies
This section is simulation, and we label it that way deliberately. We ran 100-ticket simulated months (a realistic mix: order status, returns, damage claims, exchanges, cancellations, billing, plus a share of bad-faith attempts) against 30 of these stores' real compiled policies, with an AI support agent doing the work and synthetic customers on the other side. Real policies, synthetic people. The numbers measure the policies and the agent, not consumer behavior.
The spread is the finding. The same agent, the same ticket mix, and a 27-ticket gap between the most and least automatable store, driven mostly by the silences in the table above. When the policy never says whether a damaged item comes back, the ticket stalls to a human. When it never names a refund destination, the safe move is to ask the owner. Under a conservative $50 auto-approval cap, the median store still finished 47 tickets outright and queued 15 more at the cap for a one-click approval, handing the owner about 40 written-up decisions instead of 100 conversations. Median simulated handle time saved: roughly 8 hours per 100 tickets.
The practical reading for a store owner is not "buy software." It is: answer the seven questions in the silence table, in writing, on your policy page. Your support gets faster whether a human or an agent runs it, because the judgment calls stop being invented per ticket. Our guides on preventing return fraud and return-item chargebacks cover the two most expensive silences in depth.
Methodology
Small-store tier (n=58). We captured the live return policy, FAQ, shipping, and terms pages of 58 small and mid-size Shopify stores in August 2026, then compiled each corpus through the production policy engine that powers Resolvas. The compiler records a field only when the store's own words state it, with a per-field evidence trail; anything it cannot ground stays a gap. "Silent" in this study means the compiler found no answer in the store's own pages. Stores in this tier are reported in aggregate and not named.
National-brand tier (n=52). A June 2026 hand survey of 52 national-brand policies across roughly 15 categories (Lululemon, Gap, Sephora, Patagonia, Apple, IKEA, Nordstrom, and similar), bucketed on the same dimensions. Brands are named because the policies are public and widely read.
Simulation tier (n=30 stores x 100 tickets). Simulated support months run against real compiled policies with synthetic customers, published per-store as audit reports with full transcripts. Simulation results are labeled as such everywhere they appear and measure policy automatability, never consumer behavior. Percentages are rounded to whole numbers; the underlying dataset was generated 2026-09-01.
Return policy statistics FAQ
What is the average return window for Shopify stores in 2026?
30 days is the median and the mode: 53% of small Shopify stores that state a window use exactly 30 days, and 52% of national brands do the same. Small stores skew shorter (37% of stated windows are under 30 days), while big brands extend longer, to 60, 90, or unlimited days.
Who pays for return shipping at most online stores?
At small Shopify stores, the customer: 48% state the customer pays, 17% offer free returns, and 26% never say. National brands invert it, with roughly 37% offering free returns.
How common are restocking fees?
12% of small Shopify stores charge a restocking fee of 5% to 20%, and another 17% charge a flat return fee between $4 and $40. Among national brands, 71% charge no restocking fee at all.
What do most return policies fail to specify?
Damage handling and refund destination: 84% of small-store policies never say whether a damaged item must be returned or kept, and 84% never say whether refunds go to the original payment method or store credit. Half never mention exchanges either way.
Are these findings based on real stores?
Yes. Every policy statistic comes from the captured public policy pages of 110 real stores (58 small Shopify stores, compiled through a production policy engine, and 52 national brands surveyed by hand). Only the support-outcome section is simulation, and it is labeled as simulation wherever it appears.
Want to see what your own policy compiles to, and what an agent could resolve under it? Resolvas is on the Shopify App Store, and your first 10 resolved tickets are free.