Ecommerce ROAS: How to Measure Paid Ads Without Losing Sight of Profit

For ecommerce, calculate ROAS from a consistent revenue basis, then subtract matched non-ad costs and ad spend to test contribution. Reconcile refunds once, keep unrecovered fees visible and compare a net-revenue result only with a net-revenue threshold.

Ecommerce return on ad spend is simple to calculate but easy to misread. An ad account can report a strong number while the store loses money after discounts, refunds, product cost, payment fees, fulfillment and shipping subsidies. The useful question is not merely “How much revenue did the platform attribute?” It is “How much economically valid revenue and contribution did the advertising create?”

This guide builds that answer from the order level upward. It treats platform ROAS as an optimization signal, not a complete profit statement, and shows how to establish a break-even threshold you can defend.

The basic ecommerce ROAS formula

ROAS = attributed revenue ÷ advertising spend. If a campaign spends $5,000 and receives credit for $20,000 of revenue, reported ROAS is 4.00x. That means four dollars of attributed revenue for each advertising dollar. It does not mean four dollars of profit.

The revenue definition matters. A platform may report the checkout value captured at purchase time. Your store later records cancellations, partial refunds and full returns. For business decisions, reconcile attributed revenue with retained net revenue. Use the ROAS calculator to check the arithmetic, but keep the numerator consistent whenever you compare periods or campaigns.

Build the target from contribution margin

Start with revenue after discounts and expected returns. Subtract costs that rise when an order is placed: cost of goods sold, card and marketplace fees, pick-and-pack charges, packaging, outbound shipping paid by the store, and an allowance for return processing. What remains before advertising is contribution.

Contribution margin = contribution before ads ÷ net revenue. The advertising break-even point is then:

Break-even ROAS = 1 ÷ contribution margin.

If an average order produces $100 of net revenue and $60 of non-ad variable cost, contribution before ads is $40 and contribution margin is 40%. Break-even ROAS is 1 ÷ 0.40 = 2.50x. Spending $40 to acquire that order consumes all $40 of contribution; fixed payroll, software and overhead are still unpaid. The distinction between gross margin and contribution margin explains why gross margin and usable contribution margin should not be substituted casually.

Reconcile one order cohort before judging ROAS

Use one acquisition cohort, one currency and a documented reporting cut-off. Match refunds to the original orders even when the refund happens in a later month. A calendar-month refund total can include last month's customers, so subtracting it from this month's newly acquired revenue can mix populations. This is a planning reconciliation, not a live tracker or an integration with your store.

Retained revenue = captured order revenue − adjustments not already reflected in that capture. Write which discounts, cancellations, refunds, taxes and shipping amounts the starting value includes. Do not subtract discounts again when the purchase event already used the discounted price; do not subtract an actual refund and an allowance for that same refund.

For this example, revenue excludes sales tax and customer-paid shipping. Actual business policies may differ, but the revenue and cost basis must stay consistent. Shopify defines net sales from gross sales less discounts and sales reversals; that does not mean every advertising event starts from Shopify gross sales.

The platform's captured $40,000 divided by $10,000 spend reports 4.00x. Reconciled $36,000 divided by the same spend is 3.60x. Subtracting the $2,000 discount again from the captured path would incorrectly produce $34,000 and 3.40x. The discrepancy is a definition error, not evidence that a campaign improved or declined.

Keep refunded revenue separate from unrecovered costs

Suppose the matched cohort has $16,000 of product cost after any documented inventory recovery, $1,200 of payment/marketplace fees actually retained by providers, $3,600 of outbound shipping, packaging and fulfillment, and $800 of incremental return handling and labels. These hypothetical non-ad costs total $21,600. Use records for your provider and returns policy; do not assume every refund reverses a payment fee or makes the original fulfillment cost disappear.

• Contribution before ads: $36,000 − $21,600 = $14,400.
• Contribution margin on retained revenue: $14,400 ÷ $36,000 = 40%.
• Contribution after $10,000 ads: $14,400 − $10,000 = $4,400, or 12.22% of retained revenue.
• Net-revenue break-even ROAS: $36,000 ÷ $14,400 = 2.50x.
• Captured-revenue break-even on the same economics: $40,000 ÷ $14,400 = 2.78x.

Use the pre-ad contribution and break-even calculator on a per-original-order basis: for a hypothetical 100-order cohort, enter $360 retained revenue and $216 non-ad costs (divide each cohort total by 100). The result is $144 contribution per original order, 40% and 2.50x. This normalization is not the AOV of only the kept orders. Compare its 2.50x threshold with the net-revenue 3.60x result, not the differently defined captured 4.00x. The captured 2.78x threshold is only equivalent when the same cohort's retention and cost assumptions apply. Positive contribution is not net business profit: overhead, taxes and other excluded costs remain.

Find the remaining refund headroom, not a generic safe ROAS

A sensitivity test makes the exposure visible. Hold ad spend at $10,000 and non-ad costs at $21,600 solely to isolate further revenue reversals. This is not a prediction that costs stay fixed when returns rise. Additional labels, unrecovered inventory or fee changes must be added separately.

Under those fixed-cost assumptions, $4,400 is the remaining contribution buffer against additional revenue loss, not a forecast or allowable return rate. An extra $600 of return handling would reduce that buffer to $3,800. Do not reuse the original 40% margin once retained revenue or costs change. If contribution is zero or negative, there is no positive advertising allowance and no finite positive break-even target under this model.

Save the cohort dates, revenue definition, actual refunds, outstanding allowance and cost categories in the free ROAS Profitability Workbook. Revisit the same worksheet after the return window matures or a provider fee changes. Direct PDF access does not require opting into ongoing marketing. The workbook is a reusable record, not automated order matching.

Set a profit-aware operating target

Break-even is a floor, not a healthy target. To reserve a desired post-ad contribution margin, use:

Target ROAS = 1 ÷ (contribution margin before ads − desired post-ad contribution margin).

With a 40% contribution margin and a goal to retain 15% of net revenue after advertising, allowable ad cost is 25% of revenue. Target ROAS is 1 ÷ 0.25 = 4.00x. If the desired margin equals or exceeds contribution before ads, the model is impossible: there is no room for acquisition cost. Keep the desired margin and the pre-ad contribution margin on the same retained-revenue basis.

Use low, base and high scenarios

One target can hide uncertainty. Build three scenarios before changing budgets. A low case can use a higher return rate, heavier discounting and costlier fulfillment. The base case should reflect recent mature cohorts. A high case may use a better product mix or higher average order value, but only when supported by observed data.

For example, contribution margins of 30%, 40% and 50% imply advertising break-even ROAS levels of 3.33x, 2.50x and 2.00x respectively. This range shows why a campaign can be safe for a high-margin bundle but destructive for a discounted single item. Segment products into meaningful margin bands instead of assigning one storewide target to every campaign.

Returns, discounts and timing

Return-heavy categories need a lagged view. A seven-day campaign report may look complete before customers have had time to return products. Apply a reasonable expected return allowance by product category and acquisition cohort, then replace estimates with actual results as the window matures. Avoid using a single storewide return rate when sizes, seasonal products or promotional orders behave differently.

Discount codes, loyalty credits, taxes and shipping revenue also require a documented policy. Decide whether each item belongs in net revenue and apply that rule across channels. Consistency matters more than creating a deceptively precise daily metric.

Platform ROAS versus blended ROAS

Ad platforms use different attribution windows and may both claim the same order. Platform ROAS is still valuable for comparing ads inside one system under consistent settings. It should not be summed across platforms as if attribution were exclusive.

Blended ROAS divides total store revenue by total advertising spend. It is less vulnerable to cross-platform double counting, though it includes demand generated by email, organic search, direct traffic and other sources. Review both views: platform data for tactical optimization and backend blended performance for the overall budget. See blended ROAS versus platform ROAS for a structured reconciliation method.

New customers, returning customers and lifetime value

Branded search, retargeting and retention campaigns often show high immediate ROAS because they capture people who already know the store. Prospecting may report a lower first-order result while bringing in incremental customers. Separate new-customer acquisition from returning-customer revenue whenever identity and consented data allow.

Lifetime value can justify a lower first-order ROAS only when repeat contribution is demonstrated by mature cohorts. Do not multiply optimistic revenue forecasts by a margin and call the result guaranteed value. Set a maximum first-order loss, monitor payback time, and update the allowance when retention changes.

What to include, and what ROAS leaves out

A defensible ecommerce review should include ad spend, retained revenue, discounts, returns, product cost, transaction fees, variable fulfillment, packaging and shipping subsidies. It may also include creator fees or campaign-specific production when those costs are required to run the program.

Standard ROAS does not automatically include salaries, software, rent, inventory financing, taxes or other fixed costs. It also does not prove incrementality. A profitable campaign can still have cash-flow problems if inventory is purchased months before customer receipts, and a high attributed result may partly reflect sales that would have occurred without the ad.

Before treating a higher order value as a win, use the same-spend ROAS and contribution scenario planner to compare conversion rate, net order revenue and variable cost together. Keep the resulting assumptions separate from measured experiment outcomes.

A practical weekly and monthly workflow

1. Weekly: review platform ROAS, spend, conversion volume, average order value and obvious tracking breaks using fixed attribution settings.
2. Weekly: flag product groups whose actual mix differs from the margin assumptions used in the target.
3. Monthly: reconcile platform-attributed orders with net store revenue and mature refunds.
4. Monthly: refresh variable costs, return allowances and the contribution margin by product band.
5. Monthly: compare platform ROAS with blended performance and new-customer economics.
6. Before scaling: confirm that the base case remains above the profit-aware target, not merely above advertising break-even.

For campaigns reviewed using profit-on-ad-spend metrics, state whether the numerator is contribution before or after advertising. Keep a written data dictionary so finance, ecommerce and media teams use the same definitions.

Primary references

• Shopify sales-report definitions and sales reversals
• Google Ads conversion adjustments

Google Ads supports restating or retracting eligible recorded conversions; a store refund should not be assumed to update an advertising report automatically. Verify your existing integration and conversion action before making adjustments. This guide does not upload or change conversions. Sources reviewed October 18, 2026; all example amounts are hypothetical.

Frequently asked questions