First, rebuild one month's return from the ledger
Choose a recently closed month and lay four figures side by side. From the order system or ledger, take net revenue after refunds, discounts and cancellations for the sales marketing claims. From finance, take the agreed contribution marginRevenue left after every cost that varies with each sale. for those products. From the ledger, take every marketing cost booked in the month: media, agency fees, production, tools and, if the company's convention includes them, people. And from each platform, take its attributed revenue and media spend. The platform ratio is the last two; the business's return is margin on net revenue, less all marketing cost.
The distance between the two tells you where to look. If platform revenue summed across channels is larger than net revenue the business took, the problem is over-crediting or returns. If revenue agrees but contribution after all costs is thin, the problem is the target or the costs left out. If both look fine but profit is still falling, ask whether the spend is buying sales that would have happened anyway. The ROI vs ROAS comparison describes the bridge between the two figures; this page diagnoses why the bridge shows a loss.
Diagnosis table
Diagnosis
| Symptom | Check | Likely cause | Fix |
|---|---|---|---|
| Campaigns hit target yet each order adds little or no contribution | The margin the target was derived from, and whether it includes delivery, fees and returns | Target below break-even for the margin | Re-derive the target from contribution margin, by product line |
| Platform revenue well above what reaches the bank | Refund, discount and cancellation rates for the products and channels concerned | Revenue counted before returns and discounts | Use net revenue in targets; adjust or discount platform value |
| Platform totals summed across channels exceed actual sales | Each platform's attribution settings; overlap against analytics and orders | Platforms over-claiming | Report against one business standard; test incrementality |
| Highest ROAS sits in retargeting and brand search, and those budgets have grown | Share of spend on audiences already close to buying | Spend on existing demand | Cap capture spend; judge it on holdout results |
| Channel returns positive, programme return negative | Non-media marketing costs in the ledger for the same period | Fees, tools and people left out | Add all marketing costs to the return finance sees |
| Finance and marketing disagree about whether marketing "works" | Which return figure each meeting uses, and what it includes | ROAS used where ROI is needed | Report ROI on contribution for programme decisions |
Fix, verify and prevent each cause
A target set without the margin
A return target only means something against the share of each sale the business keeps. Targets copied from another account, set on gross margin, or set once and never revisited can approve orders that lose money on every sale, while every campaign appears to be performing.
Fix: re-derive the break-even return (Break-even ROASMinimum ROAS to avoid a loss.) and a target above it from contribution margin, by product line where margins differ, using the method in the contribution margin and break-even guide.
Verify: campaigns that previously met the old target are re-sorted; those below the new line lose budget, and contribution after media rises.
Prevent: keep the target table with finance's margin inputs and the date they were set, and refresh it whenever prices, costs or return rates change. When the spend that falls below the new line is in Google Ads, the guide to cutting a Google Ads budget removes the pounds earning least first and sequences budget and target changes so learning periods do not stack.
Revenue taken before refunds, discounts and cancellations
Platforms record the value a tag or upload sends at the moment of conversion. Google's documentation on conversion tracking describes recording the actions people take after interacting with an ad and the value assigned to them; unless the business adjusts that value afterwards, a refunded order, a cancelled subscription or a discount applied later still counts in full. Fashion, footwear and promotion-heavy categories feel this most.
Fix: measure refund, cancellation and discount rates by product and, if possible, by channel, and either send net values or apply the expected deduction to targets, as the contribution margin guide describes for returns and discounts.
Verify: platform revenue, adjusted, falls into line with net revenue from the order system for the same orders.
Prevent: reconcile platform value with net revenue monthly and give discount campaigns their own, stricter targets. For a store, the guide to the returns policy and returns experience shows how coded return reasons and recorded refunds let channels be read on revenue kept.
Platforms claiming sales they did not cause
Each platform credits conversions it can connect to its own ads. Meta's help centre describes attribution settings as the window after someone clicks or views an ad within which a conversion is credited to it; a purchase made by someone who merely saw an ad, or who clicked one on another platform too, can be counted by more than one platform.
Fix: stop summing platform revenue, report against one business standard, and read the comparison of platform and analytics attribution to set it; for the largest budgets, plan an incrementality test before trusting the ratio, and read the iROAS versus ROAS comparison for which figure should govern the budget.
Verify: the sum of credited revenue across channels is reconciled to actual sales, with the gap shown rather than hidden.
Prevent: keep each platform's window and counting settings documented, and note any change as a break in the series. Where most of the claimed revenue comes from Meta, the tutorial on reconciling Meta results with your orders walks Meta's figure down to the orders the business can confirm, line by line.
Budget concentrated on people already about to buy
Retargeting and brand search report the strongest ratios because they reach people who were already close to buying. Much of that revenue would have arrived anyway, so the ratio is high and the profit the spend adds is low. As their budgets grow, programme return on ad spend rises while profit stalls.
Fix: cap remarketing and brand spend as a layer on top of prospecting, as the prospecting versus remarketing comparison sets out, and judge them on holdout tests rather than on their reported ROASRevenue generated per unit of ad spend..
Verify: after the cap, total orders fall less than the platform figures predicted, and prospecting's share of new customers rises.
Prevent: report new and returning customers separately for each channel, and review capture budgets against holdout results each quarter.
Costs outside media left out
Return on ad spend divides by media alone. Agency fees, creative production, tools, affiliate commission and, by some conventions, the team's salaries are all part of what marketing costs, and some sit in places marketing does not see. A programme can be profitable on media and loss-making once those are added.
Fix: map every marketing-related cost and where it sits in the accounts, following the marketing P&L guide, and ask agencies to invoice media and fees separately.
Verify: the return finance sees includes every mapped cost and reconciles to the ledger for the month.
Prevent: keep the cost map agreed with finance and update it when a new supplier, tool or commission arrangement starts.
A channel ratio answering a business question
Return on ad spend is the right number inside a channel and the wrong one for deciding whether marketing makes money. Sent to finance, it is read as a profit ratio, and the disagreement that follows is about which number is right rather than what to do.
Fix: report return on investment on contribution, with all marketing costs, for programme-level decisions, and keep the platform ratio for channel optimisation, as the ROI vs ROAS comparison recommends.
Verify: finance reviews use the investment return, channel reviews use the platform ratio against margin-based targets, and both figures appear in one report with the bridge between them.
Prevent: define both in one metrics dictionary and send each to the meeting it belongs in.
A prevention routine for return and profit
A strong ratio should trigger questions, not only praise. Monthly, rebuild the return for the closed month from the ledger: net revenue, contribution on the agreed margin, all marketing costs, and the platform figures beside them, with the gap between credited and actual revenue stated. In the same review, read refund and discount rates for the main products and any campaign built around a promotion. Quarterly, refresh the margin inputs with finance and re-derive break-even and target returns by product line; check the share of spend on retargeting and brand terms; and confirm each platform's attribution settings have not changed. At least once a year, run a holdout on the largest capture budget to see how much of its credited revenue it actually caused. Keep platform figures in channel meetings and margin-based figures in finance meetings, and never let a summed platform total stand in for company revenue.
An outdoor retailer finds the missing costs
An online outdoor equipment retailer reported a blended return on ad spend of 5.2 for a quarter: £94,000 of media and £489,000 of platform-credited revenue, against a target of 4. Rebuilt from the ledger, net revenue for the same orders was £372,000 after refunds and discounts, because tents and jackets had a 22% return rate and a spring sale had run at 25% off. Summed platform revenue also exceeded orders by £61,000 through overlap. At the agreed 34% contribution margin, the orders contributed about £126,000; agency fees, production and tools added £41,000 to media, so total marketing cost was £135,000 and the programme lost about £9,000. Retargeting and brand search held 46% of spend. The retailer raised targets for returned-heavy lines, capped retargeting at 20% of spend and moved fees into the reported return.
The comparison of gross margin and contribution margin explains why a return target built on gross margin leaves delivery, fees and returns unpaid.
Common mistakes
Summing platform revenue and calling it company revenue. Setting targets on gross margin. Running discount campaigns at full-price targets. Letting retargeting grow because its ratio is the best. Leaving fees and production out of the return. Sending the platform ratio to finance as if it were profit.
