What is ROAS?

A campaign spends A$4,000 in a month and the platform attributes A$14,000 of revenue to it. ROAS is 14,000 ÷ 4,000 = 3.5. If the gross margin on that revenue is 40%, the campaign returned A$5,600 of gross profit on A$4,000 of spend — a gross profit of A$1,600, not A$10,000.

Return on ad spend divides the revenue a platform attributes to a campaign by what that campaign cost over the same window. It is a ratio, which means it moves whenever either side moves, and the two movements look identical in a report.

That matters more than it sounds. A campaign whose revenue held steady while its spend fell has a higher ROAS and a smaller business behind it. A campaign that doubled its spend and its revenue has an unchanged ROAS and twice the profit. Neither of those facts is visible in the ratio alone, which is why ROAS is a diagnostic rather than a target: it tells you the efficiency of the spend, never the size of the outcome.

The two inputs are both softer than they look. Attributed revenue is whatever the platform decided to assign to the campaign, inside whatever attribution window the account is configured for. Change the window from 1-day click to 7-day click and 1-day view and the same campaign, over the same days, reports a different ROAS. Ad spend is firmer, but it is still the platform's own figure in the account's own currency, and a business running accounts in three currencies has three ROAS figures that cannot be averaged without a stated exchange rate.

The useful question is not "is 3 a good ROAS" but "what ROAS does this business need". That figure is arithmetic, not opinion:

Break-even ROAS = 1 ÷ gross margin

At a 50% gross margin, every dollar of revenue carries fifty cents of gross profit, so a dollar of ad spend needs two dollars of revenue to come back level: break-even ROAS 2.0. At 40%, it is 2.5. At 25%, it is 4.0. At 15% — a distributor, a marketplace reseller, a grocery line — it is 6.67, and a campaign running at a ROAS of 5 is losing money while its dashboard glows.

This is why a published average ROAS is close to useless as a target. Two businesses with the same ROAS and different margins are in completely different situations, and the ratio does not know the difference. Shopify's own guidance on the metric declines to name a benchmark figure for precisely this reason, saying an acceptable return depends on profit margins, operating costs and advertising goals.

Break-even is a floor, not a goal. It ignores fixed costs entirely: the warehouse, the staff, the software, the founder's salary. A business at exactly break-even ROAS is paying for its advertising and nothing else. The target ROAS a business actually needs is break-even plus whatever contribution the rest of the P&L requires, which is a finance question rather than a media one.

It double-counts across platforms. Each platform attributes conversions using its own model and its own window. A customer who saw an Instagram ad, searched the brand, clicked a Google ad and bought will appear in both platforms' reported revenue. Add the two ROAS figures together and the total revenue exceeds what the business actually took.

It counts demand it did not create. Branded search is the clearest case: someone who already decided to buy searches the brand name, clicks the ad above the organic result, and converts. That revenue lands in the campaign's ROAS whether or not the ad existed. Analytic Partners' comparison of siloed web-analytics ROAS against modelled incremental contribution puts the overstatement for paid search at two to ten times, and estimates that thirty-five cents of every dollar is misallocated when brands optimise on siloed channel metrics.

It has no memory. A campaign that acquires customers who order again for two years and a campaign that acquires one-time buyers can report identical ROAS in the month they ran. Only lifetime value separates them, and no ad platform can see it.

It punishes prospecting by construction. Retargeting a warm audience always returns a higher ROAS than reaching a cold one, because the revenue is nearly there already. An account optimised purely on ROAS drifts toward the audiences closest to purchase, shrinks its reach, and then wonders why volume stopped growing.

A few habits make the ratio far more useful:

Pair it with an absolute. ROAS with contribution profit beside it answers both the efficiency question and the size question. A drop in ROAS alongside a rise in profit is usually a business scaling correctly.

Segment before you conclude. An account-level ROAS is an average over campaigns doing completely different jobs. Prospecting, retargeting and branded search have structurally different returns, and blending them produces a figure that describes none of them.

Hold the window still. Comparing this month's ROAS to last month's is only valid if the attribution settings did not change in between. They frequently do, quietly, when a platform updates its defaults.

Watch what the ratio does when spend moves. The interesting number is not ROAS at current spend but how ROAS behaves as spend increases. Every account has a point where the next dollar returns less than the last one, and finding it is the actual job.

The instinct is to cut spend, which does raise the ratio — a smaller, more efficient account is still a smaller account. Before cutting, work through the cheaper moves.

Start with measurement rather than media. Confirm the attribution window has not shifted, that conversion tracking is firing on every purchase, and that the revenue values passed to the platform are order totals rather than line items. A surprising share of sudden ROAS collapses are tracking failures.

Then work on the denominator's quality. Excluding the placements, search terms, audiences and geographies that consistently return least raises ROAS without shrinking the account's reach. This is subtraction rather than reduction, and it is almost always available.

Then look at the numerator. Average order value moves ROAS as directly as spend does, and it is often easier to move: bundles, thresholds, a better-chosen bestseller in the ad. A ROAS that fell while conversion count rose is a mix shift toward cheaper products, and that is a merchandising decision wearing a media costume.

Only then adjust budgets, and adjust them against your own break-even figure rather than against a number somebody published about a different business.

Return on ad spend is attributed and media efficiency ratio is not. Return on ad spend divides the revenue a platform believes it caused by the spend on that platform, which makes it the figure that guides allocation and the figure that inherits every weakness of the attribution beneath it — most visibly that platform claims overlap, so summed they frequently account for more revenue than the business took. Media efficiency ratio divides all revenue by all advertising spend, crediting nothing to anyone. It cannot be double-counted, and it moves with things advertising did not do: a strong organic month, a returning customer base, a press mention. Use both. Allocate with the attributed figure, control with the blended one, and treat a widening gap between them as a measurement problem to investigate rather than a discrepancy to reconcile away.

• Fix the attribution window before you touch the campaign. A ROAS computed over a 1-day click window and a ROAS computed over 7-day click, 1-day view are different numbers about the same spend, and moving budget between them is moving budget on noise.
• Raise the denominator's quality before you cut it. Excluding the placements, audiences and search terms that convert least raises ROAS without lowering spend, and it does so without shrinking the account.
• Look at the order value, not only the order count. A ROAS that fell while conversions rose is a mix shift into cheaper products, which is a merchandising decision rather than a media one.
• Compare against the break-even figure your own margin implies, never against a published average. A 2.5 ROAS on a 60% margin is profitable and a 3.5 ROAS on a 25% margin is not.