ROAS (Return on Ad Spend)

What ROAS means

Return on Ad Spend is the most widely used measure of paid-media efficiency: for every dollar you put into advertising, how many dollars of revenue come back? A ROAS of 4.0 means $4 of revenue per $1 of spend. It's simple, universal, and — taken at face value — frequently misleading, because "revenue" and "spend" can each be defined three different ways. The version of ROAS you're quoting matters far more than the number itself.

The ROAS formula

ROAS = Revenue ÷ Ad Spend

Where it gets slippery is the numerator. Whose revenue, attributed how, over what window? Those choices produce three very different versions of the same metric.

The three versions of ROAS

1. Platform ROAS — revenue a single ad platform (Meta, Google) attributes to itself, divided by spend on that platform. Useful for steering within a channel, but inflated by view-through credit, cross-device matching, and overlapping attribution windows. Summing platform ROAS across channels double-counts buyers.
2. Blended ROAS — total revenue (all sources) ÷ total paid spend. It can't double-count and doesn't depend on click tracking surviving iOS, so it's the honest top-line scorecard. It's identical to MER. Full detail in blended ROAS.
3. Incremental ROAS (iROAS) — incremental revenue from a holdout test ÷ spend on the tested channel. It's the only version that proves causation, stripping out conversions that would have happened anyway. See incrementality.

A single channel can show a 6.0x platform ROAS, contribute to a 4.0x blended ROAS, and have a 1.5x incremental ROAS — all true at once. Knowing which one you're looking at prevents very expensive mistakes.

ROAS vs ACoS, and the SaaS view

ROAS has an inverse cousin: ACoS (Advertising Cost of Sale) = ad spend ÷ revenue, expressed as a percentage — common in retail-media and Amazon contexts. A 4.0x ROAS equals a 25% ACoS. For SaaS, "revenue per ad dollar" is less meaningful on a first-order basis because revenue recurs; SaaS teams lean on CAC (customer acquisition cost) and the LTV:CAC ratio instead, since a single signup's value only emerges over its subscription lifetime. If you run subscriptions, underwrite acquisition with CAC and LTV:CAC; if you sell one-time orders, ROAS (paired with margin) is the natural lens.

Break-even ROAS (illustrative)

ROAS doesn't tell you if you're profitable on its own — margin does. Break-even ROAS ≈ 1 ÷ contribution margin %. As an illustrative example, a product with a 40% contribution margin needs a ROAS above 1 ÷ 0.40 = 2.5x just to cover the cost of goods and variable costs on ad-driven sales. A 3.0x ROAS is healthy at 40% margin and underwater at 25% — which is why ROAS without margin context is meaningless.

Common mistakes

• Summing platform ROAS across channels. Each platform claims overlapping buyers, so the sum overstates real revenue. Use blended ROAS for the total.
• Treating ROAS as a profit metric. It ignores margin entirely; pair it with break-even ROAS and payback period.
• Confusing attributed with incremental. A high platform ROAS on retargeting often hides low incremental lift.
• Ignoring the attribution window. Longer windows mechanically raise ROAS without creating value.

How Admaxxer measures ROAS

Admaxxer shows all three versions of ROAS where they belong: platform ROAS for in-channel steering, blended ROAS / MER computed in real time against your actual store revenue as the source of truth, and incremental ROAS from built-in holdout testing when you need causal proof. Reading them side by side — with margin and payback in view — turns ROAS from a number that can mislead into a decision you can trust.