Skip to main content

Right of Return, Explained: How ASC 606 Books Sales Your Customers Can Send Back

Published 9 min readMike ThriftMike Thrift
Right of Return, Explained: How ASC 606 Books Sales Your Customers Can Send Back
On this page

Roughly one in five dollars of online sales comes back. The National Retail Federation estimated that 19.3% of U.S. online sales were returned in 2025 — nearly $850 billion across all of retail. If you book the full sale price as revenue the day you ship, about a fifth of your reported revenue belongs to customers who have not decided to keep anything yet. When the returns arrive, you get to unwind revenue you already reported, already paid tax on, and possibly already paid commissions against.

ASC 606 fixes this by treating a right of return as variable consideration: at the point of sale you recognize revenue only for what you expect to keep, record a refund liability for what you expect to give back, and recognize an asset for the inventory you expect to recover. This guide walks through all three pieces with real numbers, how to estimate expected returns, how to update the estimate each period, and the one look-alike — customer acceptance — that works completely differently.

Why Returns Are a Pricing Problem, Not a Later Adjustment​

Under the old rules, many sellers booked the full sale at shipment and dealt with returns when they happened, sometimes with an allowance account smoothing the edges. ASC 606 takes the opposite view: the refund you expect to pay is part of the transaction price itself — step three of the five-step model — so it is estimated up front, at the same moment you recognize revenue.

The practical difference is timing. The old way front-loads revenue and back-loads the pain: a strong December followed by a brutal January when the holiday returns land. The ASC 606 way matches the two in the same period. Your December revenue is already net of the January returns you can reasonably predict, which is also the number your lender and your tax return would rather see.

One important boundary: a right of return is not a separate performance obligation. You do not carve the return privilege out as its own deliverable and allocate part of the price to it. It only shrinks the transaction price of the goods you transferred.

The Three Entries at the Point of Sale​

When control of the goods transfers to a customer who may return them, ASC 606 requires three simultaneous recognitions:

  1. Revenue for the consideration you expect to keep — the sale price minus expected refunds.
  2. A refund liability for the consideration you expect to return to the customer.
  3. A right-of-recovery asset for the goods you expect to get back, with a corresponding reduction to cost of goods sold.

Take a concrete example. You sell 100 units at $50 each — $5,000 of gross sales. Each unit cost you $30. Based on your history, you expect 8 units to come back within the 30-day return window.

At shipment you record:

  • Revenue of $4,600 (the 92 units you expect to stay sold)
  • A refund liability of $400 (8 units × $50)
  • A recovery asset of $240 (8 units × $30 cost), so cost of goods sold is $2,760 for the 92 units instead of $3,000 for all 100

Notice what did not happen: you did not book $5,000 of revenue and hope. The $400 of shaky consideration sits on the balance sheet as a liability, and the $240 of inventory you expect to see again sits on the balance sheet as an asset rather than flowing through cost of goods sold.

How to Estimate Expected Returns​

The estimate uses the same two methods as all other variable consideration under ASC 606:

  • Expected value — the probability-weighted average across possible outcomes. Best when you sell many similar units and have return history: if 8% of units came back over the last two years, 8% of this sale is your starting point.
  • Most likely amount — the single most probable outcome. Best for binary or lumpy situations, like one custom machine the buyer either keeps or rejects.

Whichever method you choose, apply it consistently to similar contracts, and ground it in evidence: your own return history by product and season, adjusted for anything that makes this sale different — a new product with no track record, a looser return policy, a holiday-season spike.

Then apply the constraint: include variable consideration in the transaction price only to the extent it is probable that a significant reversal of revenue will not occur when the uncertainty resolves. In plain terms, when you genuinely cannot predict returns — a brand-new product in a new market with no comparable history — you recognize less revenue now, possibly none, until the return window closes and the uncertainty lifts. The constraint is the standard's answer to wishful forecasting.

Measuring the Right-of-Recovery Asset​

The recovery asset is measured at the former carrying amount of the inventory — what the goods cost you — less any expected costs to recover them and any expected decline in their value. Returned goods are often worth less than they were on the shelf: restocking labor, inspection, repackaging, damage, or a product that can only be resold as open-box at a discount.

Say two of your eight expected returns are display-sensitive items you will have to mark down by $10 each, and recovery costs run about $2 per unit. The asset is not $240; it is $240 minus $20 of markdown minus $16 of recovery costs, or $204. The $36 shortfall is cost of goods sold right now, not a surprise later.

Two presentation rules matter here. First, the recovery asset and the refund liability are measured independently and presented separately — the asset is not netted against the liability, even though both arise from the same estimate. Second, the asset is subject to impairment testing like any other asset: if returned goods turn out to be worth less than you measured, you write the asset down.

Updating Both Every Period​

The estimate is not set-and-forget. At the end of each reporting period you update your assessment of expected returns and adjust revenue, the refund liability, the recovery asset, and cost of goods sold accordingly. If returns are running hotter than you predicted, revenue comes down and the liability goes up; if the return window closes quietly, the leftover liability becomes revenue.

Continuing the example: the 30-day window lapses and only 6 units actually came back, not 8. You paid out $300 in refunds and restocked $180 of inventory. The remaining $100 of refund liability — the two units you braced for but kept — is now revenue, and the $60 of recovery asset attached to those two units becomes cost of goods sold. The true-up entries move both pieces to their final homes, and nothing lingers on the balance sheet.

This periodic update is also where seasonal sellers get the most value. If your January return rate runs triple your September rate, a single annual estimate misstates both months. Updating the estimate each period — or maintaining separate rates by season — keeps each month's revenue honest.

Right of Return vs. Customer Acceptance​

These two look alike and work nothing alike, so confirm which one your contract actually grants:

  • Right of return affects the transaction price (step three). Control has transferred — the customer has the goods — and revenue is recognized for the portion you expect to keep.
  • Customer acceptance affects whether control has transferred at all (step five). If the customer must formally accept the goods — a trial period, an installation sign-off, a subjective approval — you generally recognize no revenue until acceptance happens or the right lapses. The whole sale waits.

A 30-day money-back guarantee on shipped goods is a right of return: recognize net revenue at shipment. A two-week free trial after which the customer decides whether to buy is closer to acceptance: recognize nothing until they say yes. Misclassifying acceptance as a return pulls revenue forward that the standard says has not been earned yet.

The Bookkeeping in Practice​

Here is the full cycle from the example above as plain-text ledger entries. First the sale, with 8 of 100 units expected back:

2026-01-15 * "Sold 100 units at $50, expect 8 back"
  Assets:Checking                              5000.00 USD
  Income:Sales                                -4600.00 USD
  Liabilities:Refund-Liability                 -400.00 USD
  Assets:Right-of-Recovery                      240.00 USD
  Assets:Inventory                             -240.00 USD
  Expenses:COGS                                2760.00 USD
  Assets:Inventory                            -2760.00 USD

Then the window lapses with only 6 actual returns — $300 refunded, $180 restocked, and the remainder trued up:

2026-02-28 * "Return window lapsed, 6 units returned"
  Liabilities:Refund-Liability                   400.00 USD
  Assets:Checking                               -300.00 USD
  Income:Sales                                  -100.00 USD
  Assets:Inventory                               180.00 USD
  Assets:Right-of-Recovery                      -240.00 USD
  Expenses:COGS                                   60.00 USD

The second entry is the discipline made visible: the liability account must zero out, the recovery account must zero out, and the $100 of extra revenue plus $60 of extra cost of goods sold land in the period the uncertainty resolved. A balance-sheet report at any date shows exactly how much of your cash is still owed back to customers. If you are new to plain-text ledger mechanics, the Beancount documentation walks through accounts, postings, and balance reports from first principles.

Four mistakes account for most of the errors in this area. First, booking gross revenue at shipment and expensing refunds as they arrive — the pre-606 habit that overstates every strong month. Second, netting the recovery asset against the refund liability instead of presenting both. Third, measuring the asset at selling price rather than inventory cost, which quietly recognizes profit on goods sitting in a returns warehouse. Fourth, freezing the estimate at sale and never updating it, so the liability on the books describes a forecast from months ago instead of what you know now.

Your last useful step is a small one: pull your last twelve months of sales and returns by product line, compute the actual return rate for each, and compare it against whatever rate — or zero — your books currently assume. That gap, multiplied by this month's sales, is the refund liability missing from your balance sheet.

Source: https://beancount.io/blog/2026/11/01/right-of-return-explained-asc-606-refund-liability-recovery-asset-guide

Published: November 1, 2026