iROAS (Incremental ROAS)

How to calculate iROAS

Divide extra revenue by ad spend to get iROAS (incremental return on ad spend). Extra revenue is the lift from the test, not the ads manager's conversion column. Put the method and the window on the same slide as the number.

Let's take a drink brand as an example. They sell a low-sugar sports drink for active women over 50, mostly from their own website. The drink brand spends $10,000 on Instagram in four weeks. Meta (Facebook and Instagram) credits those ads with $40,000 of orders, a 4x ROAS (return on ad spend: credited sales divided by spend). The drink brand also holds Meta off in five Mountain West states. Those states still sell drink. Scaled to the same size as the exposed regions, extra sales are $12,000. $12,000 ÷ $10,000 = an iROAS (incremental return on ad spend) of 1.2.

Let's take a drink brand as an example. They sell a low-sugar sports drink for active women over 50, mostly from their own website. Meta's 4x ROAS (return on ad spend) and the test's 1.2 iROAS (incremental return on ad spend) can both be true. 4 is credited. 1.2 is extra. The other $28,000 of credited sales would likely have happened anyway (women who already drink the sports drink, or who had the case in the cart).

Why iROAS exists

In-platform ROAS (return on ad spend: credited sales divided by spend) is easy and biased. iROAS (incremental return on ad spend) is harder and closer to the decision you actually have to make: is the next dollar extra? Let's take a drink brand as an example. They sell a low-sugar sports drink for active women over 50, mostly from their own website. If the drink brand's iROAS is below 1 after a fair test, the channel is buying sales they already had. Spending more does not grow the brand. It rents the same customers.

Use iROAS (incremental return on ad spend) when the spend decision is large: keep brand search, scale retargeting, cut a prospecting channel. Do not use iROAS to pick a headline in an ads manager. That job is still ordinary ROAS (return on ad spend), with all of its bias.

How to read iROAS

iROAS (incremental return on ad spend) is only as good as the incrementality design behind it (who was held out, for how long, what counted as a sale). Say the method (geo holdout, user holdout) and the window. Do not average iROAS with last-click ROAS (return on ad spend). They answer different questions.

An iROAS (incremental return on ad spend) below in-platform ROAS (return on ad spend) is normal. Let's take a drink brand as an example. They sell a low-sugar sports drink for active women over 50, mostly from their own website. The drink brand's 1.2 iROAS next to Meta's 4x ROAS is the expected gap, not a broken test. If they match, the test may be too small, too short, or not actually holding anyone out. Treat a perfect match as a warning, not a win.

Common iROAS mistakes

• Renaming platform ROAS (return on ad spend) to iROAS (incremental return on ad spend) without a control group. Let's take a drink brand as an example. They sell a low-sugar sports drink for active women over 50, mostly from their own website. The drink brand's Meta 4x is not incremental until someone was held out.
• Killing brand search because iROAS (incremental return on ad spend) looks low on a short test. Let's take a drink brand as an example. They sell a low-sugar sports drink for active women over 50, mostly from their own website. Women already looking for the drink will buy in the holdout too. That is the point of brand, and a two-week test will punish it.

If you need a daily number, use platform ROAS (return on ad spend) and say it is credited, not extra. If you need the spend decision, wait for the test and report iROAS (incremental return on ad spend) with the method attached. Do not average iROAS with last-click attribution.