ROAS Contribution Margin: How 11x Still Loses Money

An 11x ROAS looks like a win. It can also lose money on every order, and it often does. The gap between a strong return on ad spend and a shrinking bank balance has a name. It is ROAS contribution margin: the profit left after every cost the ad platform never sees.

Reported return on ad spend answers one narrow question, revenue over spend. Founders quote it to investors as though it settled a much broader one.

This guide walks the arithmetic step by step, with figures you can swap for your own. The claim underneath it is uncomfortable and well documented. In September 2026, Search Engine Land published an account of an 11x that lost money per order, because the account “was taking credit for sales it didn’t cause, at a margin it didn’t have” (Search Engine Land). The reporting was honest. The losses were real. Run the same math and you can find the same hole in yours.

What an 11x ROAS actually measures

Return on ad spend is reported conversion value over ad cost. Both numbers come from the platform, and both flatter it. Google and Meta each count purchases the other influenced, and each counts sales that would have happened with no ad at all (AdsX). Nobody is lying. The ratio answers a narrow question while the room hears a broad one.

Two leaks turn 11x into a loss. Reported revenue is inflated. Honest revenue still sits above a stack of costs the ratio ignores. Fix them in that order.

Step one: strip reported revenue back to what was incremental

Start with incrementality, the share of sales your ads actually caused. Platform attribution credits ads that merely witnessed a purchase. Branded search is the clearest case. When someone types your name into Google and clicks the ad sitting above your own organic result, the platform books a paid sale even though you already owned that customer. Seresa documents branded-search ROAS reading 12x for exactly this reason, Google counting demand you had already earned (Seresa).

Retargeting runs the same trick. It claims buyers who were already on their way back, and platforms overstate its return by two to five times (AdsX).

Then there is double counting. Sum the ROAS each channel reports and the same order gets sold several times over, because every platform a shopper touched claims the sale as its own. In one reconciliation, $558K of platform-attributed revenue came down to $290K of actual revenue, a blended 1.92x (AdsX). The correction is blended ROAS: total revenue divided by total ad spend, so no conversion counts twice. It is a cruder number. It is also a more honest one. Incrementality testing goes further and tells you which spend caused growth rather than which pixel fired last. On our illustrative account, an 11x platform figure lands near 6x once branded and retargeted demand comes out.

Step two: build the ROAS contribution margin the report ignores

Take the reported revenue entirely at face value and the account can still lose money. ROAS is a revenue ratio, and profit lives below the line it draws. Contribution per order is the selling price minus cost of goods, minus shipping and fulfillment, minus payment fees, minus discounts, minus the expected cost of refunds (Trendtrack). None of those appear anywhere in a ROAS report.

The arithmetic on a discount-heavy catalog

Here it is, per order:

• Net revenue after a promo code: a $58 list price with an average 25% discount leaves $43.50.
• Cost of goods: $30.00 per unit, a 48% gross margin on the list price before discounts.
• Shipping and fulfillment: $9.00 to pack and ship free-shipping inventory.
• Payment processing: 2.9% plus $0.30 on Shopify Payments or Stripe, about $1.56.
• Refunds: at a 15% return rate, outbound shipping and the retained processing fee amortize to roughly $4.00 per order.
• Ad cost at 11x: $43.50 of conversion value over an 11x ratio is $3.95.

Add the costs and you spend about $48.51 to deliver $43.50. That is a loss near $5.01 on every order, at an 11x ROAS, before a single line of overhead. The pattern is not exotic. Saras Analytics shows a 4:1 ROAS on a 30% gross-margin product yielding roughly 5% contribution after ad cost, barely above break-even, while the same 4:1 on a 70% margin product returns 45% (Saras Analytics). Same ratio. Different business.

This is why a headline ROAS is a ROAS vanity metric when it travels without its costs. It climbs fastest on exactly the orders you should sell least: deep-discounted, thin-margin units the platform loves because they convert. An 11x on the wrong catalog mostly counts the thin-margin orders the platform found easiest to sell.

Step three: bid to contribution, not revenue

The fix is one change of target. Stop feeding the platform reported revenue. Feed it contribution, and bid to profit. Your gate is break-even ROAS, which is simply one divided by your contribution margin (Trendtrack). A 30% contribution margin needs 3.3x just to stand still, so a proud 3x is already underwater. Even a 3.8x has lost four figures on a single campaign when it sat below that line (marjn). Profit on ad spend, or POAS, is the same instinct written as a metric: optimize for the money you keep, not the money you report.

The Search Engine Land account did precisely this. The team rebuilt conversion values around contribution. It split brand from prospecting so it stopped harvesting credit it had not earned, then set volume against sell-through. Blended ROAS fell. The reporting got uglier. Contribution per order turned positive, and cash started arriving during the season instead of at the clearance sale (Search Engine Land).

There is a timing pressure here too. Ad platforms keep removing the manual levers that let you cap spend, which we covered in Google Ads automation removing manual controls. As bidding moves inside the model, the value you send it becomes the main thing you still control. Send it revenue and it chases revenue. Send it contribution and it chases profit.

Measurement literacy outlives any channel

The specific trap is ROAS. The habit is general. A reported metric is a claim from one system under one definition, not a fact about your business. We made that case about analytics in GA4 measurement reliability, and about the death of traffic as a north star in demand generation measurement. ROAS is the same failure in financial form.

So run the two passes on your own account this quarter. Discount the reported revenue to what was incremental, then subtract the full contribution stack until you can state profit per order. If the number you have been quoting survives that, quote it with pride. If it does not, you have found something better than an 11x: a metric you can trust. Encode the arithmetic once, in the reporting layer rather than a monthly spreadsheet, and do it before the next season rather than after it.