How returns and exchanges should change ecommerce growth decisions

By 1 At Bat Media Team

A Shopify campaign can acquire customers efficiently at checkout and still leave less money available to fund the next order. Refunds, exchanges, replacement shipments and unsellable returned products can change the result after the advertising report has already declared a winner.

Before increasing spend, connect those outcomes to the original orders, compare groups at a similar stage of resolution, and include the costs the business actually bears. Then decide whether the issue belongs in the ad, the product page, the offer, the product or fulfilment. Cutting media everywhere can be as unhelpful as ignoring the problem.

This guide gives established ecommerce teams a practical way to make that decision. It is an operating framework, not a recommendation to restrict customers' rights or change a return policy without appropriate review.

Start with the question the business needs answered

There are several useful questions, and they require different views.

Finance may need to know how much was refunded this week. Operations may need to know how many parcels are coming back. The growth team needs to know whether the customers acquired through a particular offer produced enough contribution after their orders were resolved.

A calendar-period refund total can help with the first question. It cannot answer the last one on its own. This week's refunds may relate to last month's acquisition, while this week's new buyers have barely had time to receive their products.

Keep two views beside each other:

• A processing-date view for current workload, refunds and cash planning.
• An original-order view for evaluating the products, offers and acquisition decisions that generated those outcomes.

Neither view should overwrite the other. Agree which question each report answers before comparing it with advertising spend.

Shopify's current sales reporting distinguishes sales reversals from physical returns. Reversals can include refunds, cancellations and other order adjustments; they are recorded when processed. Its sales reports also do not establish cash received. Use the relevant payment records when the decision concerns cash. Shopify sales-report definitions

Keep the original order connected to its outcome

Build a small order-level reconciliation before creating another blended dashboard. Preserve the original order ID and line-item identifiers, then link each refund, exchange, replacement and returned parcel to them.

At minimum, capture:

• Original order date, product and variant, quantity, discount and selling price.
• New or existing customer under the brand's documented definition.
• The offer and acquisition grouping used for the analysis, including unattributed orders.
• Fulfilment and delivery dates where available.
• Request date, reason, physical return status and resolution date.
• Refunded amount, any exchange top-up and any return fee collected.
• The extra shipping, handling, payment and product costs actually borne.
• Whether returned inventory was recovered into sellable stock, written off or remains unassessed.

These are analysis fields, not a reason to circulate names, addresses or customer-service transcripts widely. Share the minimum detail needed for the decision.

Do not use every new fulfilment or replacement order as evidence of another buying decision. An exchange can preserve the original purchase; a no-charge replacement can resolve a problem. Keep their operational records, but distinguish them from an independently paid repeat purchase.

The same caution applies to exchange top-ups. If a customer swaps a $100 item for a $120 item and pays the difference, record the $20 incremental payment and the associated costs. Do not call the entire $120 a fresh repeat sale without reconciling the original item and any reversal.

Separate the outcomes before interpreting the rate

A single return-rate percentage can hide different problems. Use categories that lead to different actions.

A size exchange with the replacement kept may point toward a size guide, fit photography or product-selection problem. A goodwill refund after late delivery belongs in a fulfilment and promise review. A cancellation before dispatch may need a different investigation from a product that arrived damaged. A product refunded without a physical return does not produce recovered inventory.

Record customer-selected reasons as evidence to investigate, not a complete diagnosis. A broad label such as “not as expected” needs a sample of the product information and service conversation before the team concludes that the ad caused it.

Define the denominator too. Orders with any physical return, units physically returned, refunded product value and total sales reversals measure different things. Label each measure and preserve the number of eligible orders or units beside the percentage.

For Shopify specifically, the documented AOV calculation excludes post-order adjustments. A rising checkout AOV therefore needs a separate check of what remained after resolution. Do not rename the platform's AOV field or assume it already answers that question. Shopify AOV definition

Compare orders with a fair observation window

Choose an observation window from the brand's actual delivery times, return policy and processing lag. There is no universal number of days that makes every category mature.

An item delivered in two days and a made-to-order item delivered in six weeks should not be judged as though their buyers have had the same opportunity to use or return them. Seasonal extended-return windows can also leave a promotion unresolved long after its media campaign ends.

For each group, show the order-date range, the observation cutoff and how many cases remain open. When comparing an older offer with a new one, first compare both at the same elapsed age. Keep a later, more mature view of the older offer available separately.

If the latest group is immature, the team has three honest options:

1. Hold the next large budget increase while the material uncertainty resolves.
2. Continue a bounded test with an explicit spend limit and review date.
3. Use a clearly labelled forecast based on relevant historical outcomes, while showing observed results separately.

Do not blend a forecast into an observed result and then call it realised profit. If no suitable history exists, a narrower test is often more defensible than a precise-looking estimate.

The same-age cohort comparison extends this principle to customer contribution and acquisition payback.

Calculate the money available after resolution

For a campaign decision, define a contribution measure with finance rather than treating gross sales, platform ROAS or product margin as interchangeable with profit.

A useful structure starts with product revenue after discounts and refunds, then subtracts the attributable product and variable servicing costs, and finally acquisition spend. Keep shipping revenue and shipping costs consistent with the agreed definition. Treat taxes, currencies, gift cards and store credit according to the underlying records rather than mixing them into product revenue casually.

Product-cost recovery deserves particular care. If the returned item is inspected and restored to sellable inventory, the accounting treatment can differ from a damaged item written off. Do not assume every requested return restores the product cost. Do not subtract the original cost twice when a separate write-off line is already included.

Likewise, a refund may leave payment fees or outbound shipping costs behind. Use the actual provider and fulfilment records; the economics should not assume costs were refunded merely because the customer received money back.

This operating contribution measure still excludes any overhead not explicitly allocated. It is not after-tax profit, a bank balance or a financial-statement substitute.

A worked example where cheaper acquisition leaves less contribution

The following example is fictional. It illustrates the reconciliation, not a client result or industry benchmark. Both offers have 100 acquired customers, the same observation window, no unresolved cases, one currency and no repeat purchases in the comparison. Taxes and customer shipping charges are excluded.

Offer A generates $12,000 of product revenue after discounts but before refunds. Offer B generates $11,500. Acquisition costs $3,000 for A and $2,500 for B, making observed acquisition cost $30 and $25 per customer respectively.

After resolution, the economics are:

The product-cost row already incorporates the agreed treatment of recovered and unrecoverable stock. Additional servicing costs exclude that product cost, avoiding a second deduction.

Offer B acquired customers for $5 less each but left $1,175 less contribution across the group. Its contribution after acquisition was $15.25 per customer versus $27 for A.

That finding supports investigating B before expanding it. It does not establish that its advertising caused every refund. Product mix, fit, discount depth, fulfilment and customer expectations could all differ. Nor does it establish which offer has higher lifetime value; this example contains no later purchase evidence.

Route the finding to the part of the journey that can fix it

Start with the largest explainable loss, not the team with the most visible dashboard.

If a creative execution attracts buyers expecting a feature the product does not provide, compare the actual ad, landing page and customer reason before replacing that message. If exchanges cluster around one size or variant, improve the selection information and inspect the product before suppressing a whole channel.

If late-delivery refunds cluster around an advertised arrival promise, bring operations into the offer review. Media changes alone cannot create fulfilment capacity. If the economics deteriorate mainly during a discount, compare the margin and product mix under that offer before assuming the audience became worse.

For each intervention, write down:

• The evidence supporting the suspected cause.
• The exact ad, product, variant, page or offer affected.
• One proposed change and its accountable owner.
• The expected improvement and the customer-experience measure that must not deteriorate.
• A review date with enough observation time to assess outcomes.

Avoid making a return policy harsher simply to improve a dashboard. Any policy change needs appropriate legal and commercial review, and should be evaluated alongside conversion, complaints, retention and customer trust. Reducing preventable disappointment is a more useful starting point than making a valid resolution difficult.

Make the next budget decision explicit

The weekly review should end with a specific decision: expand, maintain a bounded test, correct a defined issue, or pause the affected offer. Record the observation cutoff, unresolved cases and costs still estimated.

The media owner should know what can change immediately. The product or operations owner should know what must be fixed before further traffic arrives. Finance should know which costs and adjustments remain provisional. These are proposed responsibilities to assign, not assumed job titles at every brand.

For a promotion still inside its return window, schedule the mature readout when the campaign is planned. That prevents the team from celebrating the checkout result and never revisiting what happened afterward.

Use the ecommerce growth feedback loop to connect the resulting action to creative, conversion, retention and the next spending decision.

Questions growth teams should settle before scaling

Does a high return rate mean paid media is poor quality

Not by itself. Check product and offer mix, customer expectations, delivery outcomes and attribution limits. Compare like-for-like groups before assigning responsibility to a channel or agency.

Should exchanges count as repeat purchases

Keep exchanges identifiable as resolutions of the original purchase. Measure additional paid value separately where it exists. Report independently paid repeat purchases using a consistent definition so service activity does not inflate customer retention.

Can we use last month refunds divided by this month sales

That ratio may describe processing-period pressure, but it mixes different order groups. Use original-order linkage and a stated observation window when judging the acquisition or offer that generated the outcomes.

How should an agency report this

Ask for a reconciled view of acquired customers, retained revenue, relevant variable costs, unresolved cases and the next action. The agency may need finance and operations data to complete it. A platform report alone cannot fill those gaps.

Connect acquisition to the order customers actually keep

A growth plan becomes more useful when the advertising decision survives the return, exchange and service-resolution process. Keep original orders linked, separate observed outcomes from forecasts, and route the largest supported problem to an owner who can change it.

If your team needs paid media, Shopify conversion and retention decisions connected to these commercial outcomes, talk with 1 At Bat Media about the scope and evidence needed. The first question is which decision the business cannot make reliably today.