Returns, Refunds, and Damage Allowance: Pricing for Loss Rates

Why returns and refunds must be priced as a probabilistic cost

Returns and refunds are not “rare exceptions” in ecommerce; they are a predictable loss rate that reduces profit in a measurable way. The key pricing mistake is treating returns as a separate operational problem instead of a per-unit cost that should be embedded into your price model. Because you do not know which specific unit will be returned, you model returns as an expected (probabilistic) cost spread across all units sold.

Think of it like this: if 10% of units are returned and each return costs you an average of $12 in losses and handling, then every unit you sell should “carry” an expected $1.20 return cost (10% × $12), even though 90% of units won’t be returned.

Core definitions (use these consistently)

Return rate (RR)

Return rate is the percentage of shipped orders/units that come back to you.

• Unit-based: returned units ÷ shipped units
• Order-based: returned orders ÷ shipped orders

Use unit-based when products vary in quantity per order or when you need SKU-level accuracy.

Refund rate (RefR)

Refund rate is the percentage of shipped orders/units that result in money refunded to the customer. Most returns lead to refunds, but not all refunds require a return (e.g., “keep it” refunds, lost packages, partial refunds).

Continue in our app.

You can listen to the audiobook with the screen off, receive a free certificate for this course, and also have access to 5,000 other free online courses.

Or continue reading below...

Download the app

• Unit-based: refunded units ÷ shipped units
• Revenue-based: refunded revenue ÷ gross revenue

For pricing, unit-based is often easier to convert into an expected cost per unit, but revenue-based is useful when refund amounts vary (partial refunds, discounts, bundles).

Restocking success rate (RSR)

Restocking success rate is the percentage of returned units that you can resell at full price (or at a known recovery value).

• RSR = resellable returned units ÷ total returned units

Many businesses split this into buckets: “resell as new,” “open-box,” “refurbished,” “liquidation,” “dispose.” That gives a more accurate recovery value.

Disposal/damage costs (DDC)

Disposal/damage costs are the costs incurred when a returned unit cannot be resold (or is resold at a steep discount). This includes:

• Write-off of product value (inventory loss)
• Disposal fees or recycling fees
• Refurbishment/repair costs
• Repackaging materials
• Quality inspection labor
• Inbound return shipping you pay (if you offer free returns)

In practice, the biggest driver is usually the recovery value of the returned item: how much cash you can recover from it (full-price resale, discounted resale, liquidation, or zero).

Build an expected returns cost per unit (ERC)

The goal is a single number you can add to your per-unit cost model: Expected Returns Cost per Unit (ERC). The simplest form is:

ERC = Return Rate × Average Return Loss

To make it actionable, define what “Average Return Loss” includes and compute it from components you can measure.

Step 1: Map the return outcomes and cash flows

For each returned unit, you typically face:

• Refund amount (usually the item price paid, sometimes minus a fee)
• Return logistics cost (label cost, inbound shipping, handling)
• Processing cost (inspection, repack, customer service time)
• Recovery value from the returned item (resell, open-box, liquidation)

Model the loss on a returned unit as:

Return Loss per Returned Unit = Refund Amount + Return Handling Costs − Recovery Value

Then:

ERC = RR × (Refund Amount + Return Handling Costs − Recovery Value)

Step 2: Use a recovery-value approach (recommended)

Instead of trying to guess “damage cost,” estimate what you recover on average from a returned unit.

Example recovery buckets:

• Resell as new: recover ~100% of item price (but may still incur repack/inspection)
• Open-box: recover 70–90% of item price
• Liquidation: recover 5–30% of item price
• Dispose: recover 0% (and may pay disposal fees)

Compute expected recovery value:

Expected Recovery Value = Σ (Probability of bucket × Recovery in that bucket)

Then plug into ERC.

Step 3: Separate “return rate” from “refund-without-return”

Some refunds happen without a physical return (lost package, “keep it” policy, goodwill). Treat these as a separate loss rate because there is no recovery value and often no inbound return shipping.

Expected Refund-Without-Return Cost per Unit (EFC) = (Refund-without-return rate) × (Average refund amount)

Total expected loss per unit from reverse logistics events:

Total Expected Loss per Unit = ERC + EFC

Worked example: expected returns cost per unit

Assume a product sells for $60. You observe:

• Return rate (RR) = 12%
• Refund amount per return = $60 (full refund)
• Return handling costs (label + processing) = $8 per return
• Returned item outcomes: 50% resell as new (recover $60), 30% open-box (recover $45), 20% dispose (recover $0, plus $2 disposal fee)

Compute expected recovery value:

Expected Recovery = 0.50×60 + 0.30×45 + 0.20×0 = 30 + 13.5 + 0 = $43.50

Compute average return loss per returned unit (include disposal fee as part of handling costs or subtract from recovery):

Return Loss = Refund (60) + Handling (8) + Disposal fee expected (0.20×2=0.40) − Recovery (43.50)

Return Loss = 60 + 8 + 0.40 − 43.50 = $24.90

Expected returns cost per unit sold:

ERC = RR × Return Loss = 0.12 × 24.90 = $2.99 per unit

Interpretation: if you ignore returns, you are overstating profit by about $2.99 on every unit sold (on average). That is a pricing problem, not just an operations problem.

Practical step-by-step: implement this in a pricing sheet

Step 1: Track the minimum data fields

• Units shipped (by SKU)
• Units returned (by SKU)
• Refunded units and refunded amount (by SKU if possible)
• Return label cost (average or actual)
• Processing cost estimate (minutes × labor rate, or flat)
• Disposition outcome (resell/open-box/liquidation/dispose) and recovery amount

Step 2: Calculate rates

• RR = Returned Units / Shipped Units
• RefR = Refunded Units / Shipped Units
• RSR = Resellable Returned Units / Returned Units

Step 3: Calculate average return handling cost

Start with a simple flat estimate per return and refine later:

Return Handling Cost = Avg label cost + Avg processing cost + Avg repackaging cost

Step 4: Calculate expected recovery value

If you have bucket data:

Expected Recovery = Σ (Bucket share × Avg recovery in bucket)

If you do not have bucket data yet, use a conservative temporary recovery assumption (e.g., “we recover 50% of item price on returned units”) and replace it once you have 30–100 returns of data.

Step 5: Compute ERC and add it to your per-unit cost model

Return Loss = Avg refund amount + Return handling cost − Expected recovery

ERC = RR × Return Loss

Use ERC as a line item in your pricing model so your target margin is achieved after expected returns.

Category benchmarks (use as a starting point, not a rule)

Return rates vary heavily by category, sizing/fit uncertainty, and customer expectations. Use benchmarks only to set an initial assumption and to sanity-check your own data.

When in doubt, start conservative: assume a higher return loss until your data proves otherwise.

How to price new products with limited data

Use a staged assumption model

New SKUs often have no return history, so you need a method that avoids underpricing while you learn.

• Stage 0 (pre-launch): use category benchmark RR and conservative recovery (lower recovery, higher handling)
• Stage 1 (first 50–200 orders): compute RR weekly, but smooth it (small samples swing wildly)
• Stage 2 (after 200–500 orders): switch to SKU-specific RR and recovery buckets

Apply smoothing so early data doesn’t whipsaw pricing

A simple approach is a weighted average between benchmark and observed:

RR_est = (w × RR_benchmark) + ((1−w) × RR_observed)

Where w starts high (e.g., 0.7) and declines as order volume grows (e.g., 0.3 after a few hundred orders).

Use leading indicators to adjust assumptions early

• High “size/fit” questions or reviews mentioning fit → raise RR assumption
• High damage complaints → raise disposal/damage bucket share and handling costs
• High “not as described” feedback → raise RR and reduce recovery value (more opened/used)

Policy choices (free vs paid returns) and how they influence pricing

Your return policy changes customer behavior and your cost structure. Pricing must reflect the policy you choose.

Free returns (you pay label; often free refunds)

Typical effects:

• Higher RR: lower friction increases “try-and-return” behavior
• Higher handling cost per return: you pay inbound shipping
• Potentially higher conversion rate: which may offset some cost

Pricing implication: increase ERC via both higher RR and higher handling cost. If you do not raise price (or improve recovery), free returns can silently erase margin.

Paid returns (customer pays label or a fee is deducted)

Typical effects:

• Lower RR: some customers keep the product
• Lower cost per return: you may not pay inbound shipping
• Customer experience trade-off: may reduce conversion or increase support tickets

Pricing implication: ERC decreases, but you may need to invest elsewhere (better product pages, sizing tools) to protect conversion.

Restocking fees

A restocking fee reduces refund amount and can discourage casual returns. Model it explicitly:

Avg refund amount = Item price × (1 − Restocking fee rate)

Be careful: some platforms, payment providers, or local regulations restrict restocking fees or require specific disclosures.

Exchange-first policies

If you encourage exchanges instead of refunds, you may reduce cash refunds but still incur reverse logistics costs. Model exchange outcomes as:

• Return shipping + processing still happens
• Replacement shipment cost may occur
• Revenue may be retained (good), but costs rise (also true)

In your model, treat exchanges as a separate bucket with its own average cost.

Common modeling pitfalls (and how to avoid them)

Pitfall 1: Using only return rate and ignoring recovery value

A 10% return rate can be manageable if 90% of returns are resold at full price, but disastrous if most are disposed. Always model recovery.

Pitfall 2: Treating refunds as 100% loss

Refund is a cash outflow, but the item may come back and be resold. The true loss is refund + handling − recovery.

Pitfall 3: Mixing order-level and unit-level rates

If customers often buy multiple units per order, order-level return rate can understate unit-level impact. Choose one level and stay consistent.

Pitfall 4: Ignoring “refund without return” events

Lost packages, chargebacks, and goodwill refunds can be a meaningful loss rate. Track and model them separately because recovery is usually zero.

Template formulas you can copy into a spreadsheet

Assume these inputs per SKU:

• RR = return rate
• RNR = refund-without-return rate
• P = average item price paid (net of discounts)
• H = average return handling cost (label + processing + materials)
• Rec = expected recovery value per returned unit
• Disp = expected disposal fee per returned unit (optional; can be included in H)

ReturnLoss = P + H + Disp − Rec

ERC = RR × ReturnLoss

EFC = RNR × P

TotalReverseLogisticsCostPerUnit = ERC + EFC

Add TotalReverseLogisticsCostPerUnit as a line item in your pricing model so your target profit accounts for expected losses from returns and refunds.