The short answer
A returns reserve is a provisional estimate of revenue that may later be reversed. It can make a recent campaign forecast more realistic when returns arrive slowly. Keep it separate from actual refunded revenue: the reserve is an estimate, and its accuracy depends on the historical cohort and assumptions you use.
Choose a comparable mature cohort
Use orders that have had enough time for the return process to develop. Match product mix, geography, promotion, and return terms where practical. A rate measured as refunded revenue divided by original revenue differs from a rate measured as returned orders divided by orders.
Calculate a revenue-based reserve
Illustrative inputs: a mature comparable cohort had $20,000 of eligible original revenue and $1,600 of refunds, giving an observed revenue refund rate of $1,600 / $20,000 = 8%. A new cohort has $10,000 of original revenue and $2,000 in advertising spend. Assuming the same rate, expected eventual refunds are $10,000 × 8% = $800.
Provisional revenue after expected refunds is $9,200 and provisional ROAS is $9,200 / $2,000 = 4.6×. The raw ratio is 5×. Neither the 8% assumption nor the 4.6× forecast is a benchmark or a guaranteed outcome.
Avoid counting actual refunds twice
If $300 of the expected $800 has already been refunded, a simple remaining reserve is $800 − $300 = $500, provided the original estimate still applies. Starting from revenue already net of $300 and subtracting the full $800 again would double-count those known refunds.
Model costs separately
Return shipping, restocking, unrecovered fees, and recovered inventory value affect contribution differently. A revenue reserve alone is not a complete profit forecast. Keep each component explicit instead of assuming every returned dollar has the same cost impact.
As the cohort matures, replace estimated values with actual outcomes and record the forecast error. If product mix or return terms changed materially, show a range of assumptions rather than carrying an old rate forward without review.
Separate remaining reserve from refunds
- Gather: Choose a comparable mature cohort and calculate its revenue-based refund rate.
- Produce: Show original revenue, actual refunds, remaining reserve and provisional net revenue separately.
Decision boundary: Do not subtract a full expected-refund estimate again after already deducting known refunds.
Next reading
Examples are illustrative, not client results or benchmarks. The calculations use the inputs shown; real-world decisions depend on your records and assumptions. Report a correction.