What does iROAS mean?
iROAS, or incremental return on ad spend, answers a deliberately narrow question: of all the revenue we recorded, how much was genuinely caused by the advertising? It is the return that would have evaporated had the campaign never run. Standard ROAS makes no such distinction. It takes every conversion an ad platform attributes to a campaign, divides the revenue by the cost, and reports the ratio - regardless of whether those buyers were ever in doubt.
That difference matters because most paid media sits on top of demand that already exists. A shopper who searches your brand name was already intending to buy. A loyal customer reordering a consumable would have returned anyway. Retargeting a basket-abandoner who comes back that evening often captures a sale that was always going to close. Reported ROAS counts all of these as wins; iROAS counts none of them, because the advertising did not change the outcome. The canonical one-line definition lives in our iROAS glossary entry - this article is about what that definition means in practice, and how it changes the decisions you make.
How does reported ROAS overstate performance?
Attribution is fundamentally a crediting system, not a causal one. When a platform records a conversion, it is recording that an ad was seen or clicked somewhere along the path to purchase - not that the ad moved the purchase. The two are easily conflated. Channels that intercept high-intent moments, such as branded search and lower-funnel retargeting, are structurally good at being present at the point of conversion and structurally bad at proving they caused it. They therefore post the most flattering reported ROAS while frequently delivering the weakest incremental return.
The practical consequence is over-investment in demand capture and under-investment in demand creation. A channel showing a headline 6.0 ROAS may have an iROAS closer to 2.0 once the organic baseline is stripped out; a brand-building channel showing a modest 1.8 reported ROAS may be doing most of the real work of generating new customers. This is the same logic that separates the efficiency of your next pound of spend from your blended historic figure - explored in depth on marginal ROAS vs average ROAS. Incrementality and marginal thinking are two lenses on the same problem: averages hide where the money actually works.
How is incrementality measured?
Because no ad platform reports iROAS directly, it has to be estimated. The most rigorous approach is experimental: hold out a randomly selected group from exposure and compare its conversion rate to the exposed group. Geo holdouts switch a channel off in matched regions; audience holdouts suppress ads for a slice of users; conversion-lift studies measure the gap between test and control. The difference between the two groups is the lift - the revenue that only appeared because the advertising ran.
Live experiments are not always practical, so most teams supplement them with response and media-mix modelling. These methods estimate the baseline a channel would have earned with zero spend and attribute only the uplift above it, while accounting for the fact that advertising effects decay over time and flatten as budgets grow. Two modelling concepts do most of the heavy lifting here: adstock captures how impact carries over after exposure, and the Hill function describes how incremental response curves towards a ceiling. In practice the strongest programmes run periodic experiments to calibrate ongoing models, then plan against the resulting incremental curves rather than against attributed totals.
Using iROAS in budget planning
Budgets are decided at the margin, so the only question that matters when reallocating spend is: where will the next pound create the most genuinely new revenue? iROAS reframes the entire planning exercise around that question. Instead of ranking channels by attributed average - which rewards demand harvesting - you rank them by the incremental return of additional spend, and you keep feeding a channel only while that incremental return stays above your target. This is exactly the forecast-led discipline behind ElenIQ’s positioning: forecast before you spend, rather than reconcile after the fact.
It also explains a pattern every performance marketer recognises. Scaling a winning channel so often disappoints because incremental return falls as spend rises - each extra pound reaches a less responsive audience, the dynamic described by saturation curves and budget ceilings. A channel can have a strong reported ROAS and a near-zero incremental return on its next increment at the same time. Planning against incremental curves rather than averages is the difference between scaling profitably and pouring money into a saturated channel because its blended figure still looks healthy - a distinction that matters as much for e-commerce as it does for lead generation.
A worked example
Imagine a retailer running two channels. Branded search spends £20,000 a month and is attributed £160,000 in revenue - a reported 8.0 ROAS. A prospecting channel spends £40,000 and is attributed £120,000, a reported 3.0 ROAS. Judged on those headline figures, the obvious move is to cut prospecting and pour more into branded search. That is precisely the trap.
Now suppose a geo holdout reveals that 80% of the branded-search revenue would have arrived anyway - customers who searched the brand because the prospecting channel had already made them aware of it. The incremental revenue from branded search is therefore around £32,000, giving an iROAS of roughly 1.6 - barely above breakeven. The prospecting channel, by contrast, is found to drive almost entirely net-new demand: its iROAS sits close to its reported 3.0. The ranking inverts completely. The channel that looked twice as efficient is in fact the weaker investment at the margin, and the “underperformer” is what makes the whole account work.
The planning conclusion is not “switch off branded search” - it remains a cheap way to defend existing demand - but “stop crediting it with growth it does not create, and protect the upper-funnel spend that feeds it.” This is the kind of reallocation you can model before committing budget rather than discover after a quarter of misdirected spend. You can pressure-test the same logic on your own numbers with the marginal ROAS calculator, then turn historic data into forward-looking incremental forecasts with ElenIQ’s ad spend forecasting tool.
Frequently asked questions
What does iROAS mean?
iROAS stands for incremental return on ad spend. It measures only the additional revenue that advertising genuinely created - the sales that would not have happened without the spend - rather than all the revenue an ad platform claims credit for.
Why does reported ROAS overstate performance?
Reported ROAS divides all attributed revenue by spend, including conversions from people who would have bought anyway through brand searches, repeat purchases, or existing demand. Because that baseline revenue is counted as if the advertising caused it, reported ROAS is almost always higher than the true incremental return.
How is incrementality measured?
Incrementality is measured by comparing an exposed group to an unexposed control - through geo holdouts, audience holdouts, or conversion-lift studies - or by response and media-mix modelling that estimates the baseline a channel would have earned without spend and attributes only the uplift above it.
How do you use iROAS in budget planning?
Plan against the incremental return of the next pound of spend rather than a blended historic average. Shift budget towards channels whose incremental return stays high as they scale, and cap channels whose incremental return collapses once they saturate, even if their reported ROAS still looks strong.