ROAS: knowing whether an ad really pays

ROAS in one formula

“ROAS” is searched 12,100 times a month in France. The formula fits on one line: ROAS = revenue attributed to advertising ÷ ad spend. It is expressed as a ratio (3) or as a percentage (300%).

Worked example (fictional shop): you spend €1,000 on Google Ads and the campaign brings 30 sales at €60, that is €1,800 of revenue. The ROAS is €1,800 ÷ €1,000 = 1.8.

ROAS is useful because it reads instantly, campaign by campaign, ad by ad. It is dangerous because it ignores what the product sold costs you: a campaign with a high ROAS can lose money on a low-margin product.

Break-even ROAS: the only figure that tells you whether you are making money

Break-even ROAS is the ROAS below which every sale from advertising loses you money. It is calculated from your margin, after product cost, delivery and payment fees: break-even ROAS = 1 ÷ margin rate.

Back to the example: the product sold at €60 costs €24, delivery €6. That leaves €30 per sale, a margin rate of 50%. The break-even ROAS is 1 ÷ 0.50 = 2.0. With a ROAS of 1.8, the campaign loses money: 30 sales × €30 of margin = €900, minus €1,000 of advertising, a loss of €100.

Do not forget payment fees in the margin. At Stripe, for example, the listed rate in France for standard European Economic Area cards is 1.5% + €0.25 per transaction; premium and international cards cost more.

1. 1
   Work out the margin per sale: selling price incl. or excl. VAT depending on your accounts, minus product cost, the delivery you pay, payment fees and, where relevant, packaging.
2. 2
   Divide by the price: that gives the margin rate.
3. 3
   Invert it: 1 ÷ margin rate gives the break-even ROAS.
4. 4
   Compare: every campaign below this threshold loses money on the first sale; above it, it makes money.

ROAS, ROI, acquisition cost: do not mix them up

• ROAS: revenue ÷ ad spend. It measures how efficient the advertising is, not the profit.
• ROI (return on investment): (profit − investment) ÷ investment. It takes costs into account; it is the figure that tells you whether the money invested pays.
• Customer acquisition cost (CAC): spend ÷ number of new customers. In the example, €1,000 ÷ 30 = €33.33 per customer. “Coût acquisition client” (customer acquisition cost) is searched 90 times a month in France.
• Customer lifetime value: what a customer brings in across all their orders. A customer who buys again justifies a higher acquisition cost than their first order alone.

Rule of thumb: a first-order campaign can tolerate a ROAS close to break-even if your customers often buy again; it must clear it comfortably if your customers only buy once.

Improve ROAS without spending more

ROAS rises in two ways: more sales for the same spend, or more revenue per sale. The most profitable levers are almost always on the website, not in the campaign:

• Fix the buying journey: one click to the payment, no mandatory account, costs shown early. Every paid visitor who drops out lowers the ROAS.
• Increase the basket: bundles, a complementary product offered at the right moment.
• Send each ad to the page that matches the search exactly, not to the home page.
• Cut the searches that spend without selling: the search terms report shows what people actually typed.
• Scale back advertising where organic search already wins the query.

Measure accurately: the traps of reported ROAS

The ROAS an ad platform reports depends on how it attributes sales. Google Ads, Meta and your analytics tool can each claim the same order. Before deciding, compare your total real sales with the total attributed by all the platforms: if the latter is higher, some sales are counted twice.

Measure purchases with a unique transaction ID: Google Analytics states that the transaction_id parameter helps you avoid duplicate events for a purchase.

1. 1
   One reference figure: the real sales recorded by your payment tool or your shop.
2. 2
   Clean purchase events in Google Analytics 4, with transaction_id and value.
3. 3
   A long enough period: one week of a campaign is not enough to conclude on a product that sells slowly.
4. 4
   A hold-out test: pausing a campaign for a few weeks and watching whether total sales fall tells you what it really brought.