Returns-adjusted ROAS calculator: compare refund costs

This returns-adjusted ROAS calculator helps ecommerce teams review one order cohort. Enter refunds, recovered stock and return costs to compare two scenarios. See net-sales ROAS beside contribution after ads, inspect the calculation and export the results. No store connection, account or email is needed.

The refund pressure ledger

What remains in the original-order cohort?

Fictional USD example. Change the assumptions; expected fractional return counts are not observed orders.

Current contribution after ads2,840.00
Current net-sales ROAS4.5
Alternative contribution after ads2,010.00
Alternative net-sales ROAS4
Gross-revenue ROAS5
Break-even refund rate after ad spend44.2169%

Read the two return measures separately

A gross revenue-to-ad-spend ratio can look comfortable while the same orders leave little contribution after refunds. This calculator keeps the original order cohort fixed and changes the expected share of orders that return. It then shows two different questions: how much sales revenue remains after refunds, and how much contribution remains after the variable costs and advertising you entered.

Net-sales ROAS is not profit. It divides modeled net sales by ad spend. Contribution after ads also subtracts product cost, outbound shipping, payment fees, other variable costs, reverse shipping and restocking. Recovered product cost and explicitly refunded payment fees are credited separately. Fixed overhead, tax, financing and future repeat purchases are outside this model.

Use the comparison to inspect a decision, rather than treating a green result as a prediction. The inputs are assumptions you control. The page does not connect a store or an advertising account, retrieve orders, infer a normal return rate or recommend a universal acquisition target.

How to use the cohort ledger

  1. Choose a consistent group of original orders and the ad spend associated with those orders. Use a cohort with enough time for its usual return window to develop.
  2. Enter AOV after discounts and before refunds. Keep tax treatment consistent across revenue and costs. Do not combine tax-inclusive revenue with tax-exclusive costs and call the remainder profit.
  3. Enter the share of orders expected to return and the share of an order's value refunded when that happens. A partial refund can be modeled without pretending the whole sale disappeared.
  4. Add reverse shipping and restocking costs. Credit recovered product cost only when the returned inventory has a recoverable value under your accounting assumptions.
  5. Compare the current rate with an alternative rate. Export the three rows to keep the assumptions alongside the result.

The no-return row uses the same original orders, AOV, ad spend and original costs. It is a reference scenario, not a claim that returns can be eliminated. Expected return counts can be fractional because they express a probability applied to a cohort. Actual observed orders are whole numbers.

Worked example with fictional USD inputs

The starting example contains 100 original orders at a 100 USD AOV and 2000 USD of ad spend. Product cost is 30 USD per original order, outbound shipping 8 USD, other variable cost 2 USD and the payment fee 3% plus 0.30 USD. Those original variable costs total 43.30 USD per order. They are counted for all 100 orders, once each.

At a 10% return rate,10 expected returns receive a full 100 USD refund. Modeled net sales become 9000 USD. Each return adds 5 USD of reverse shipping and 2 USD of restocking cost, while 80% of the 30 USD product cost is recovered:24 USD per return. No payment fee is assumed refunded in this example.

The resulting variable cost is 4160 USD:4330 USD of original costs, plus 70 USD of reverse handling, less 240 USD of recovered product cost. Contribution before ads is 4840 USD, and contribution after ads is 2840 USD. Net-sales ROAS is 4.5 times, while the original gross revenue divided by ad spend is 5 times. Neither ratio alone describes the 2840 USD remainder.

At the alternative 20% rate, net sales become 8000 USD and variable costs 3990 USD. Contribution after ads falls to 2010 USD. The lower modeled cost does not mean returns are helpful: it reflects the inventory credit, while the lost revenue is larger. The table keeps those movements visible instead of subtracting an unexplained percentage from a headline ROAS.

Method and assumptions

For each scenario, expected returns equal original orders multiplied by the return-rate share. Refunds equal expected returns multiplied by AOV and by the refunded-value share. Net sales equal original order revenue minus those refunds. These are cohort calculations. They need not match a report that posts a refund on the date the money moved, especially when its original order belongs to an earlier period.[1]

Original variable cost per order includes COGS, outbound shipping, other variable cost and the entered payment fee. Return handling adds reverse shipping and restocking. Recovery credits the chosen portion of product cost, not the retail price of the returned item. A fee credit is included only if you enter it; processor and payment-method behavior must be checked rather than assumed.[2]

The break-even refund rate solves the linear contribution equation after the entered ad spend. If that threshold is negative, the no-return case already fails to cover advertising under these assumptions. A threshold above 100% lies outside the possible range of a return share. If a return does not reduce contribution under the entered recovery and refund assumptions, the tool shows no decreasing-rate threshold and asks you to inspect those inputs.

When ad spend is zero, ROAS is unavailable instead of infinite. When contribution before ads is zero or negative, a positive break-even ROAS is unavailable. Those cases still have readable sales, cost and contribution rows.

What this model cannot decide

Do not mix a count-based return rate with a value-based refund rate without checking how order values vary. The model uses one average order and one refund fraction. A return-heavy expensive product can behave differently from the rest of the cohort. Separate materially different products or fulfillment methods into their own runs.

The recovered-cost field is an economic assumption, not a cash deposit from the warehouse. Unsellable goods, inspection failures, markdowns, storage and a delayed resale can reduce recovery. Chargebacks, exchange orders and customer-service labor need their own consistent cost treatment; they are not automatically discovered here.

Use presell page math to inspect the earlier funnel, A/B testing presell pages to plan a comparison, and product bundle landing pages to make the sold offer explicit. The calculator exports a reusable decision sheet; it does not forecast demand or replace your books.

Frequently asked questions

Is returns-adjusted ROAS the same as profit?

No. Net-sales ROAS divides net sales by ad spend. Contribution also subtracts the modeled variable costs. Fixed overhead, taxes and future repeat purchases remain outside this tool.

Which return rate should I enter?

Use a consistent, sufficiently mature original-order cohort. Do not divide this month’s refunds from several older cohorts by this month’s orders and treat that as a cohort return probability.

Why can the break-even rate fall outside0–100 %?

It is an algebraic threshold under your assumptions. A negative rate means even the no-return case cannot cover this ad spend; a rate above100% is outside the feasible domain, not a realistic performance forecast.

Sources

  1. Shopify · net sales: formula and examples (2026-10-27)
  2. Stripe · refund fee behavior (2026-10-27)

More free tools

Go further

Guides that use these numbers

All articles

Testing & growth12 min read

Presell page math: when the extra step pays for itself

A presell funnel pays when its orders leave more contribution than the same traffic sent straight to the store. Multiply click-through by the purchase rate, then check costs. Keep the same traffic basis, price and cost rules. More clicks or a bigger order alone do not prove a win.

Testing & growth13 min read

A/B testing presell pages: read the result clearly

Read a presell A/B test against the plan you wrote before launch. Keep the same visitor basis, check the product handoff, and separate clicks from purchases. A result below your evidence threshold remains uncertain; it does not prove that the pages perform alike.

Formats12 min read

Product bundle landing pages: make the whole offer clear

A product bundle landing page sells a useful set by naming every item, its quantity and the next step. Show the full offer before the button. Keep the ad, kit contents, price and cart in sync. A larger order helps only if its contribution holds.