Platform ROAS divides attributed revenue by advertising spend. True commercial return asks how much incremental contribution the advertising produced after discounts, refunds, cost of goods, fulfilment, payment fees and other variable costs. Shopify revenue, Meta or Google attributed revenue and finance-recognised revenue will rarely match exactly because they use different event timing, identity, attribution and adjustment rules.

Teams should preserve each system for its intended decision. Use ad-platform reporting for campaign optimisation, Shopify for order-level commerce records, analytics for journey analysis and finance data for profitability. Reconcile them into a governed model instead of forcing one source to answer every question.

Project Supply builds ecommerce analytics and attribution systems that connect marketing activity with commercial outcomes: Project Supply AI and Data Analytics

Four metrics commonly called ROAS

Platform-reported ROAS

Advertising platforms calculate return from conversions they attribute to ads under their configured windows, models and identity signals. This metric is operationally useful inside the platform because bidding systems optimise against the same environment. It is not an audited profit measure.

Shopify revenue-to-spend ROAS

A Shopify-centred calculation divides selected Shopify order revenue by advertising spend. It may use marketing reports, tracked sessions or an external model. The result depends on whether revenue is gross or net, which orders and dates are included and how channels receive credit.

Blended ROAS

Blended ROAS divides total store revenue by total advertising spend for the same period. It avoids fighting over channel attribution, but includes organic, direct, repeat, referral and other demand. It is a useful business trend, not proof that paid media caused all revenue.

Contribution return on ad spend

Contribution return divides incremental contribution generated by advertising by advertising spend, or compares contribution after media with a suitable baseline. This is closer to the commercial question but requires reliable cost and incrementality assumptions.

Why the numbers disagree

Attribution windows

Platforms may credit conversions occurring days after an interaction. Shopify and analytics tools may use different lookback windows or session logic. Document click-through and view-through windows before comparing reports.

Attribution models

Last-click, first-click, data-driven and platform self-attribution allocate credit differently. Multiple platforms may each claim the same order. Adding platform-attributed revenue across channels can therefore exceed store revenue.

Identity and consent

Cookies, browser restrictions, consent, logged-in identity, device changes and server-side signals affect matching. A platform may model conversions; Shopify holds the resulting order but may not know every marketing touchpoint.

Order timing

Ad platforms may report by interaction date or conversion date. Shopify records order creation and later adjustments. Finance may recognise revenue under different timing rules. Align date logic before labelling a variance as tracking failure.

Currency and tax

International stores can record presentment currency, shop currency and settlement currency. Reports may include or exclude taxes, duties and shipping. Convert consistently and retain the original currency for audit.

Refunds, cancellations and returns

Initial conversion reports may retain revenue that is later refunded or cancelled. Build adjustment logic and define the delay after which a cohort is considered sufficiently mature for profitability review.

The true-ROAS calculation

Start with eligible net revenue from orders associated with the measured activity. Subtract discounts, refunds and cancellations according to the organisation’s accounting definition. Then subtract variable costs that change with the sale.

A practical contribution expression is: net revenue minus product cost minus variable fulfilment and packaging minus payment fees minus variable shipping subsidy minus expected return or RTO cost minus other order-level variable expenses. Advertising spend is then compared with this contribution.

Do not present one formula as universally correct. Finance should define which costs belong in contribution and how shared costs are treated. Preserve both the formula version and the effective date so historical results remain interpretable.

Worked model without invented benchmarks

Suppose a cohort contains attributed orders, discounts, recognised refunds, product costs, payment fees, variable fulfilment and advertising spend. Calculate gross attributed revenue, net revenue and pre-ad contribution in separate steps. Then subtract media spend to obtain contribution after advertising.

Report the result as currency contribution, contribution margin percentage and contribution return relative to media. Use the organisation’s real values; avoid inserting generic industry margins or target ROAS numbers that ignore the product mix.

Break-even ROAS

Break-even ROAS depends on the contribution margin available before advertising. If only a fraction of each revenue unit remains after variable costs, the required revenue-to-ad-spend ratio must compensate for that structure. Finance should calculate break-even by product or margin band when the catalogue varies materially.

A campaign can exceed platform ROAS targets and still destroy contribution when it attracts heavily discounted, high-return or expensive-to-fulfil orders. Conversely, a lower initial ROAS may be acceptable for an incremental cohort with strong retained value, provided that assumption is measured.

New-customer and repeat-customer economics

Separate new and returning customers. Repeat orders can make retargeting or branded campaigns appear highly efficient even when advertising did not create the underlying demand. New-customer acquisition is usually the more relevant view for growth investment.

For lifetime-value analysis, cohort customers by acquisition period and source, then observe repeat contribution over time. Do not add speculative future value to current results without a documented retention model and confidence range.

Incrementality

Attribution answers which touchpoints receive credit under a model. Incrementality asks what would not have happened without the advertising. Use geo experiments, audience holdouts, platform lift tests or carefully designed time-series methods where feasible.

Incrementality factors should be estimated by channel, campaign type, audience and market rather than assumed globally. Branded search, retargeting and existing-customer campaigns often have different baseline demand from prospecting activity.

Data model

Order fact

Store order ID, customer identity key, creation time, market, currency, gross sales, discounts, refunds, tax, shipping, net sales, product cost and variable expenses. Preserve adjustment timestamps.

Marketing spend fact

Store platform, account, campaign, ad set or group, ad, date, currency and spend. Maintain stable mapping tables because campaign names change and identifiers belong to specific platforms.

Touchpoint fact

Where lawful and available, store session or event time, source, medium, campaign identifiers, click IDs, landing page, consent state and customer or session linkage. Avoid collecting unnecessary personal information.

Attribution output

Keep model, lookback window, credit and model version. Never overwrite historical attribution when the model changes; otherwise reported performance moves without underlying business change.

Cost reference

Maintain product cost, fulfilment, payment and other variable-cost rules with valid-from dates. Cost data is often the weakest part of true-ROAS reporting and needs finance ownership.

Reconciliation process

Choose a closed period and common timezone. Pull Shopify orders and adjustments, platform spend and conversion reports, analytics events and finance cost references. Normalise currencies and identifiers. Compare totals in layers rather than expecting immediate equality.

First reconcile order counts and revenue definitions. Then reconcile spend. Next compare attribution credit. Finally calculate contribution. Classify differences as definition, timing, identity, duplication, missing data or implementation defect.

Need a governed Shopify, GA4 and advertising reconciliation model? Discuss the analytics implementation with Project Supply: Talk to Project Supply

Dashboard design

Executive view

Show net revenue, spend, blended ROAS, contribution before and after advertising, new-customer acquisition cost, refund or return impact and variance to plan. Include data freshness and confidence.

Channel view

Show platform ROAS beside governed attributed ROAS and incrementality-adjusted contribution where available. Never hide the attribution model or window.

Cohort view

Show first-order contribution, repeat contribution, cumulative contribution, payback progress and retention by acquisition cohort. Allow maturity filters so recent cohorts are not compared unfairly with older cohorts.

Data-quality view

Show event coverage, click-ID capture, missing cost, unmatched orders, currency conversion status, refund maturity and source latency. Decision-makers need to know when performance changes are measurement problems.

Decision rules

Use platform ROAS for in-platform bidding diagnostics, but gate budget decisions with marginal spend, new-customer quality, contribution and incrementality evidence. Evaluate what happens as spend increases; average ROAS can conceal weakening marginal return.

Create actions for different failure modes. Strong attributed revenue but weak contribution calls for product, discount, return or fulfilment analysis. Strong blended performance but weak incrementality calls for channel reallocation. Strong contribution but poor platform signals may require measurement improvement.

Governance

Assign finance ownership of contribution definitions, marketing ownership of campaign taxonomy, data ownership of pipelines and models, engineering ownership of instrumentation, and business ownership of decisions. Approve changes through versioned documentation.

Restrict access to customer-level data, apply retention rules and minimise identifiers. Reconciliation tables should support audit without exposing unnecessary personal information.

Implementation roadmap

Days 1–15: definitions and audit

Agree revenue, spend, contribution, customer and attribution definitions. Audit Shopify, platform and analytics tracking. Document gaps and current discrepancies.

Days 16–30: foundation

Build order, spend and cost tables. Normalise timezone, currency and campaign identifiers. Produce a daily reconciliation report before adding complex attribution.

Days 31–60: contribution and cohorts

Add refunds, product costs, variable expenses, new-customer logic and cohort maturity. Obtain finance sign-off on sample orders and monthly totals.

Days 61–90: incrementality and decisions

Design priority experiments, add confidence labels and embed budget-review routines. Train teams to explain differences rather than choosing the most favourable report.

Common mistakes

Avoid adding attributed revenue from several platforms, comparing gross revenue with net finance revenue, ignoring refunds, applying one margin to every product, treating repeat revenue as paid acquisition, changing definitions without versioning, and optimising to last-click ROAS alone.

Another mistake is overengineering attribution before basic order and spend reconciliation works. Trust grows from explainable totals and stable definitions.

Commercial recommendation

Report a metric ladder: platform ROAS, governed attributed ROAS, blended ROAS, first-order contribution and incrementality-adjusted contribution where evidence supports it. Each metric answers a different question.

The goal is not to produce one perfect number. It is to make marketing decisions that remain defensible when finance, operations and customer behaviour are included.

For Shopify attribution, cohort analytics and contribution reporting implementation, contact Project Supply: Talk to Project Supply