Is your store leaking revenue?
Find out exactly where you're losing sales — takes 2 minutes.
Why a derived ceiling beats a round number every time
What Is Marginal ROAS?
Most ROAS targets are a feeling.
Marginal ROAS (also called IROAS, or incremental ROAS) is the true return on the next dollar of ad spend once you strip out margin costs and channel incrementality. It differs from average ROAS, which is Revenue ÷ Spend across all history, and from break-even ROAS, which is only the floor (1 ÷ Profit Margin). Marginal ROAS is the ceiling: the derived number that tells a channel exactly where to stop chasing a higher target.
Our break-even ROAS guide covers the floor: the minimum return you need to avoid losing money on a sale. That post answers "what is the lowest ROAS I can accept?"
This post answers a different question: what is the right ROAS to run at, once margin and incrementality are both accounted for? That number is rarely round, and it is almost never the number most founders are running.
Here is how the three metrics actually differ:
| Metric | What It Measures | How It's Derived | Question It Answers |
|---|---|---|---|
| Average ROAS | Total return across all spend to date | Revenue ÷ Ad Spend | How has this campaign performed so far? |
| Break-even ROAS | The floor: minimum return before you lose money | 1 ÷ Profit Margin | What is the lowest ROAS I can accept? |
| Marginal ROAS (IROAS) | The ceiling: true incremental return on the next dollar | Contribution margin ratio, discounted by the channel's incrementality factor | What target should this channel actually run at? |
In every ad audit we run, we hear a version of the same line: "there's such a wide range of answers... a lot of those ROAS numbers come from a feeling. Vibes. 'Four is bigger than three, let's go for four.'" A systematized target replaces that vibe with two inputs.
What Are the Two Inputs to a Real Marginal ROAS Target?
Two numbers, combined, set the target.
A marginal ROAS target is built from two inputs: the contribution margin ratio after all variable costs, and the channel's incrementality factor from geo holdout testing or benchmarks. Combining both produces the marginal IROAS target, the ROAS at which one more dollar still produces contribution margin at or above target.
Input 1: The Contribution Margin Ratio
This is the contribution margin left over after every variable cost of delivering the sale is backed out, not just COGS:
- Cost of goods sold
- Cost of delivery / shipping
- Pick and pack fees
- Payment processing
The result is a ratio: how much of every revenue dollar actually stays as margin. This is the same input used for break-even ROAS, but here it is only half the equation.
Input 2: Channel Incrementality
Not every dollar of "attributed" revenue would have been lost without the ad. A Facebook or Google campaign retargeting a customer who was already going to buy gets credit it did not earn.
Incrementality is measured two ways:
- Geo holdout tests, run specifically for your brand: pause ads in one region, compare revenue against a matched region running as normal.
- Benchmarks, when a fresh test is not available yet.
This factor discounts channel-reported revenue down to what the channel actually caused.
Combining Both
Here is the math with the inputs stated as assumptions, illustrative only, not a benchmark for your store:
- Assumed contribution margin ratio: 35%, after COGS, delivery, and pick/pack
- Assumed channel incrementality factor: 0.70, meaning only 70% of platform-attributed revenue is judged truly incremental
- Break-even ROAS on true (incremental) revenue: 1 ÷ 0.35 = 2.86x
- Marginal IROAS target on platform-reported ROAS: 2.86 ÷ 0.70 = 4.1x
Change either input and the target moves. A brand with a 50% margin and a 0.9 incrementality factor would land closer to 2.2x. The arithmetic is what matters here, not this specific pair of numbers.
Why Isn't a Higher ROAS Target Always Better?
This is the rule that changes how you run every campaign.
A ROAS target above the derived marginal number is not safer, it is more expensive. If the correct target is 3.0 based on margin and incrementality math, running the campaign at 4.0 leaves profitable volume unbought. A higher target does not mean more profit; it means a tighter filter on spend that was already contribution-margin positive.
The core rule, and it is worth reading twice: "Four is not necessarily better than three. If three is the right number based on the math that folds in cost of delivery and the incrementality of the channel, then three is the right number. It's not four. If you're running a campaign at four in that case, you are absolutely leaving volume on the table."
Raising a target above the marginal number does not make the business more profitable. It suppresses spend that would still have produced margin at target. You buy a prettier report and pay for it in lost volume.
Google Ads' own guidance on cross-channel bidding makes the same point from the platform side: "narrow targeting often comes from using average ROI to determine whether a performance segment is valuable," and average values don't reflect the cumulative value of marginal conversions.
Is Your ROAS Target Legitimate?
Most targets fail this four-question test on the first question.
A ROAS target is legitimate only if it passes four checks: derived from contribution margin after all variable costs, discounted by a channel incrementality factor, set at the marginal number rather than rounded up for safety, and aligned at both the channel and campaign level. Any "no" answer means the target is a guess, not a target.
Run every channel target through this checklist:
- Was it derived from unit economics? Contribution margin after COGS, delivery, and pick/pack, not gross margin.
- Was it discounted by incrementality? A geo holdout result, or a stated benchmark assumption, not raw platform-reported ROAS.
- Is it the marginal number? Not a round number added "for safety."
- Is it aligned across levels? Channel target and campaign target both set from the same math.
If any answer is no, the target is a vibe. Rebuild it before using it to judge any campaign as over or under target.
Does this sound like your store? Find out where you're leaking revenue: take the free Revenue Score. 3 minutes. Free. No pitch.
What Is IROAS, and Why Does a High IROAS Mean You Should Spend More?
IROAS is not a fancier version of ROAS. It is the honest version.
IROAS (incremental ROAS) reports channel performance through incrementality-adjusted revenue instead of raw platform-attributed revenue. An IROAS number sitting far above target is not inflated or too good to be true; it is a signal of headroom and underinvestment, meaning the channel can absorb more budget while still clearing the marginal target.
Report every channel through IROAS, not raw ROAS. Revenue, orders, cost per acquisition, and ROAS all get run through the same incrementality factor, so the numbers stay comparable to each other.
An IROAS well above target is not a trophy. It is a signal to open the budget, because the channel is proving it can absorb more spend and still clear the marginal bar.
Why Do Brand Search Campaigns Need a Different ROAS Target Than Non-Brand?
Not every click is the same customer.
Brand search campaigns need a much higher ROAS target than non-brand acquisition because brand search mostly captures customers who were already going to buy, giving it very low incrementality. Non-brand acquisition campaigns reach new customers with genuinely higher incremental value, which earns them a lower, more permissive target.
| Campaign Type | Customer Cohort | Incrementality | Target IROAS |
|---|---|---|---|
| Non-brand / acquisition | New customers | Higher | Lower target (illustrative, near the ~4.1x worked example above) |
| Brand search | Existing and returning customers | Very low, given a heavy haircut | Much higher target: same margin, but a lower incrementality factor pushes the ratio up |
Because brand search mostly reaches people who were already coming, it gets a large incrementality haircut and, as a result, a much higher required target. Reporting brand and non-brand PPC spend against the same flat ROAS number is a category error. If a brand campaign still clears even that high bar, there is real uncaptured reactivation demand behind it.
What Should You Do When Actual ROAS Is Above, At, or Below Target?
Once you know the target, the decision is mechanical.
When actual IROAS sits above target while spend is below pace, the channel is underinvested and budgets should open. When actual IROAS sits at target, spend is correctly tuned. When actual IROAS sits below target, the account is overspending past efficiency and should pull back or fix targeting.
| Situation | Action |
|---|---|
| Actual IROAS above target, spend below pace | Underinvesting. Open budgets and loosen the bid target toward the marginal number. |
| Actual IROAS at target | Correctly tuned. Spend to the budget; do not tighten for vanity. |
| Actual IROAS below target | Overspending past efficiency or wrong target. Pull back or fix targeting. |
| Target set tighter than the marginal number | Loosen it. A tight target suppresses profitable spend. |
| Budget too small to reach pace | Raise the budget. A correct target still caps volume if the budget can't fund it. |
The two most common causes of "IROAS way over target while underspending" are a target set tighter than the marginal number, and a budget too small to fund it. Check both before assuming the channel is simply out of room.
How Do You Set a Marginal ROAS Target Without Running a Geo Holdout Test?
Most Malaysian and Singaporean stores never run a formal holdout.
Without a fresh geo holdout, you can still set a marginal ROAS target by applying a documented incrementality benchmark to your contribution margin ratio, as long as you explicitly state the assumption. This produces a working target instead of a guess, and it should be revisited once real test data becomes available.
Three steps if you cannot test today:
- Use a benchmark incrementality factor, sourced from a prior test on a comparable brand or a published channel study, rather than assuming 100% incrementality.
- State the assumption in writing. Note which benchmark you used and why, so the target can be re-derived later.
- Revisit once you can test. A geo holdout is still the most reliable read; treat the benchmark-based target as provisional, not permanent.
If you find yourself distrusting the incrementality read entirely, the fix is not to override it with a gut number. Go run a fresh geo holdout. Operating with the best available information, even an imperfect benchmark, is still more defensible than a target picked because it sounded ambitious. This mirrors the underlying law of diminishing returns: past a certain spend level, every channel's true incremental output shrinks, whether or not you have measured it yet.
We build this exact target math into every ROAS audit we run for growing DTC brands, because a single round-number target cannot tell a founder whether the next dollar is worth spending.
Frequently Asked Questions
What is marginal ROAS?
Marginal ROAS, also called IROAS, is the true return on the next dollar of ad spend after accounting for contribution margin and channel incrementality. Unlike average ROAS (total revenue divided by total spend) or break-even ROAS (the minimum floor), marginal ROAS is a derived ceiling that tells you the exact target a channel should run at.
Is a higher ROAS target always safer?
No. A ROAS target set above the derived marginal number does not increase profit. If the correct target is 3.0 based on margin and incrementality math, running at 4.0 filters out spend that was already contribution-margin positive. Google Ads' own guidance confirms average-based targets can be misleading for this reason.
Why does brand search need a different ROAS target than non-brand ads?
Brand search mostly captures customers already planning to buy, so it has very low incrementality and needs a much higher ROAS target to justify the spend. Non-brand acquisition reaches genuinely new customers with higher incremental value, earning a lower target, often in the 3x to 4x range depending on margin.
What should I do if my IROAS is far above target?
A channel running well above its marginal IROAS target is underinvested, not performing too well to trust. Treat it as a signal to open the budget and loosen the bid target toward the marginal number, since the channel is proving it can absorb more spend while still clearing target.
How do I calculate a marginal ROAS target without a geo holdout test?
Apply a documented incrementality benchmark to your contribution margin ratio, state the assumption explicitly, and treat the resulting target as provisional. This produces a working number instead of a guess. Revisit the target once you can run a real geo holdout test for your own brand.
Keep Reading
Free Download
We audited 10 Shopify stores doing $1M–$20M. Here are the 15 leaks every one had.
Real screenshots (names blurred). Most operators have 8 of the 15 and don't know it.
From 10 audits across SG, MY, AE — fashion, beauty, electronics, food.
PDF in your inbox in 30 seconds. 1 email. No follow-up unless you ask.
