ROAS and iROAS both measure return on ad spend. The difference: ROAS counts all conversions that happened while your ads were running. iROAS counts only the conversions your ads actually caused.
For channels with high organic intent, retargeting, brand search, loyalty campaigns, the gap between the two can be large enough to completely change how you evaluate a campaign.
iROAS = incremental revenue generated per euro of incremental ad spend
The formulas
ROAS = Total attributed revenue / Ad spend
iROAS = Incremental revenue / Ad spend
Where:
Incremental revenue = (conversion rate of test group − conversion rate of control group) × audience size × average order value
The key distinction is in the numerator. ROAS uses attributed revenue, every conversion the platform claims credit for. iROAS uses only the revenue that would not have existed without the advertising.
A worked example
A brand runs a retargeting campaign. They split the retargeting audience into two groups for a four-week holdout test: 10,000 people in the test group who see the ads, and 2,000 in the holdout/control group who don't.
At the end of four weeks:
- Test group conversions: 420 (4.2% conversion rate)
- Control group conversions: 70 (3.5% conversion rate)
- Average order value: $80
- Ad spend during the test period: $8,000
ROAS calculation (using platform-attributed conversions):
The platform attributes 420 conversions to the retargeting campaign.
ROAS = (420 × $80) / $8,000 = $33,600 / $8,000 = 4.2x
iROAS calculation:
Incremental conversion rate: 4.2% − 3.5% = 0.7%
Incremental conversions: 0.7% × 10,000 = 70
Incremental revenue: 70 × $80 = $5,600
iROAS = $5,600 / $8,000 = 0.7x
The platform reports 4.2x ROAS. The actual incremental ROAS is 0.7x, below break-even. The campaign is losing money in incremental terms, despite showing an excellent attributed number.
Why the gap exists
In this example, 350 of the 420 conversions in the test group (83%) happened regardless of the ads. Those people were already going to buy. The ads only caused 70 additional conversions, the same rate as the control group's organic baseline.
In last-click or platform-native attribution, all 420 conversions get credited to the retargeting campaign. The platform shows a 4.2x ROAS because it can observe that those 420 people saw a retargeting ad and then converted. It cannot observe, and has no incentive to measure, how many of them would have converted anyway.
This is the structural problem with retargeting measurement. You're showing ads to people who already visited your site and expressed purchase intent. Many of them are going to convert on their own. The ads intercept that organic journey and take credit for it.
When ROAS and iROAS are close
The gap between ROAS and iROAS is largest for high-intent audiences: retargeting, brand search, existing customer campaigns, and abandoned cart flows. These audiences have high organic conversion rates, so a large portion of their attributed conversions would have happened without the ads.
For cold prospecting campaigns, targeting people who have no prior relationship with your brand, organic purchase intent is low. Most conversions from this audience are genuinely incremental. ROAS and iROAS will be relatively close, though not identical.
As a rough heuristic:
- Cold prospecting: iROAS is typically 60–90% of attributed ROAS
- Warm audiences (lookalikes, engaged): iROAS is typically 40–70% of attributed ROAS
- Retargeting: iROAS is typically 15–50% of attributed ROAS
- Brand search: iROAS is often below 30% of attributed ROAS
These are generalizations. Run your own tests, the numbers vary significantly by brand, category, and audience size.
How to get iROAS without running an experiment
Running a formal holdout test every time you want an iROAS estimate isn't practical at scale. Once you've run a test for a given channel or campaign type, you can use that result as a calibration factor.
If your retargeting holdout test showed 20% incremental lift (meaning 20% of attributed conversions were additional):
Estimated iROAS ≈ Attributed ROAS × 0.20
So if your current retargeting campaign shows 5x ROAS, your estimated iROAS is approximately 1.0x.
This calibration is imprecise. Incrementality rates change as audiences change, creative changes, and market conditions shift. Plan to re-test your major channels every six to twelve months to keep calibration factors current.
An alternative for brands investing in media mix modeling: a well-calibrated MMM can serve as a proxy for incremental ROI at the channel level, since it models business outcomes as a function of spend rather than attributing individual conversions. Use the MMM's marginal ROI estimates alongside your holdout test results to triangulate.
What iROAS threshold to target
The right threshold depends on your margins, business model, and the stage of the customer relationship you're driving. A common starting framework:
| iROAS | Interpretation |
|---|---|
| Above 3x | Strong incremental efficiency, scale with confidence |
| 1.5x–3x | Healthy, maintain current investment |
| 0.8x–1.5x | Marginal, optimize creative, audience, or frequency before scaling |
| Below 0.8x | Losing money incrementally, reduce spend significantly or pause |
A campaign with iROAS below 1x is, in incremental terms, destroying value: you're spending more in ad costs than the additional revenue you're generating. Even if the attributed ROAS looks good, you'd be financially better off not running the campaign.
The caveat: some campaigns don't need to be incremental in the short term. Brand-building campaigns, subscriber acquisition for a newsletter, or retention campaigns targeting at-risk customers may have strategic value that doesn't show up cleanly in a four-week incrementality test. Be intentional about when you apply an iROAS filter and when you're measuring something else.