Real ROAS calculator: is your ROAS real?

Calculate your adjusted ROAS (estimate)

Spend, margin and what each platform reports. Everything updates as you drag.

Total monthly ad spend (EUR)

Contribution margin before marketing

Break-even ROAS at this margin: 2.5×

Margin is what is left per order after cost of goods, shipping, payment costs, returns and discounts. Not gross margin (contribution margin). Keep revenue and margin on the same footing (VAT, returns).

Source of reported ROASAttribution window

Use one source, one period and the same window for every channel. These two fields only label the result.

Channels

Shares are relative weights. Set a channel to 0 if you don't run it. The example account is invented.

Retargeting€25k / month · 25% of spend
Reported ROAS
6.0×
Branded search€20k / month · 20% of spend
Reported ROAS
8.0×
Prospecting / creative€35k / month · 35% of spend
Reported ROAS
2.0×
Email / CRM€5,000 / month · 5% of spend
Reported ROAS
12.0×
Affiliate / coupons€10k / month · 10% of spend
Reported ROAS
5.0×
Creators / social€5,000 / month · 5% of spend
Reported ROAS
2.5×

Adjusted ROAS (estimate)

~55%

of reported revenue, as estimated

Reported ROAS5.0×
Adjusted (estimate)2.2× to 3.3×

Blended range around 2.8×: the estimate plus or minus a fifth, a design choice and not a confidence interval. Factor 0.55× on reported ROAS, margin 40%. Break-even ROAS at this margin: 2.5×.

Underwater: Retargeting, Branded search, €45k/month of your spend. At these assumptions this channel earns less margin than it costs. Treat it as a question to test, not an instruction to switch it off.

€10kContribution after ad spend. Not profit: agency costs, creative production, tools and team are not in it.

Break-even ROAS calculator

Illustrative estimate from stated assumptions. It does not measure your account. A test with a control group, or a marketing-mix model calibrated with tests, is how you find out what your ads actually did.

Reported vs. adjusted ROAS, channel by channel

Sorted by the gap between reported and adjusted revenue. Bars show reported ROAS; the band is the adjusted ROAS (estimate) plus or minus a fifth.

Reported and adjusted ROAS by channel

Branded search
€20k / month
Biggest gap
8.0× → 2.5×
2.0×–3.0× underwater
Retargeting
€25k / month
6.0× → 1.8×
1.4×–2.2× underwater
Email / CRM
€5,000 / month
12.0× → 6.6×
5.3×–7.9×
Affiliate / coupons
€10k / month
5.0× → 2.5×
2.0×–3.0×
Creators / social
€5,000 / month
2.5× → 3.0×
2.4×–3.6×
Prospecting / creative
€35k / month
2.0× → 3.1×
2.5×–3.7×

Illustrative estimate from stated assumptions. It does not measure your account. A test with a control group, or a marketing-mix model calibrated with tests, is how you find out what your ads actually did.

What does the calculator do?

You enter monthly spend and reported ROAS for each channel. You add your contribution margin before marketing. It returns three things per channel: an adjusted ROAS (estimate) as a range, the contribution left after ad spend, and a flag when the channel looks underwater.

"Reported ROAS" means the conversion value the platform credits to ads, divided by what you spent. An analytics vendor's explainer defines ROAS this way¹. The credit follows the platform's rules and attribution window. Underwater means adjusted revenue times margin comes to less than spend.

How does it work, and what does it assume?

The whole model is one line: adjusted ROAS (estimate) = reported ROAS × a fixed factor for the channel type. No tests, no account history.

The factor is an incrementality factor. One testing vendor uses the term for incremental conversions from a test divided by conversions the platform reported². An agency's published channel analysis multiplies platform ROAS by such a factor to estimate incremental ROAS³.

The assumptions, in plain words:

The factors are illustrative
They are our modelling choices for each channel type. They are not a benchmark, and they are not drawn from ZFD client results.
The range is arbitrary
It is a band of a fifth either side of the estimate. That is a design choice, not a confidence interval.
Direction varies
A measurement vendor says platform ROAS is often higher than incremental ROAS for lower-funnel activity such as brand search, and can be lower for upper-funnel video and reach⁴. A testing vendor reports that, in tests for its own clients, Meta's 7-day click-only reporting understated incrementality on average⁵. So the result is not always lower than your panel.
Margin is our convention
We use contribution margin before marketing. Gross margin gives a more flattering break-even.
Channels are independent
Cutting retargeting may change branded search. The model cannot see it.
Automated campaigns do not fit
Performance Max runs across Google's channels from one campaign⁶. Advantage+ sales campaigns (formerly Advantage+ shopping campaigns) are Meta's automated ecommerce campaigns⁷. Neither maps onto one channel type. Split their spend by judgement and treat the range as wider.

Example (illustrative): A channel reports ROAS 4.0. The factor for its type is 0.5, so adjusted ROAS (estimate) is 2.0. At a 40 per cent contribution margin, break-even is 2.5. The channel is flagged underwater. That is a prompt to test.

Incrementality factor assumed for each channel
Channel	Why	Factor	Example
Retargeting	Mostly reaches people already on their way to buying.	×0.30	6.0× → 1.8×
Branded search	Captures demand that other channels created.	×0.31	8.0× → 2.5×
Prospecting / creative	Under-reports: its effect shows up in other channels.	×1.55	2.0× → 3.1×
Email / CRM	Partly owned demand: some of it would have happened anyway.	×0.55	12.0× → 6.6×
Affiliate / coupons	Often claims the last click on a sale that was happening.	×0.50	5.0× → 2.5×
Creators / social	Mid-funnel; moderately under-reported.	×1.20	2.5× → 3.0×

What can't it tell you?

Incrementality. Incrementality asks what would have happened without the ads. The tool assumes an answer. It does not find one.
iROAS. iROAS is incremental revenue divided by spend, and a trade publication describes it as measured by testing⁸. Our adjusted figure only approximates it. Do not report it to your board as iROAS.
The truth. Academic work comparing observational methods with experiments on Facebook found that the observational methods often did not reproduce the experimental results⁹. A marketing-mix model has the same weakness. The team behind Meta's open-source Robyn tool names calibration with randomised experiments as its main way to reduce that bias¹⁰.
The next euro. The tool uses average ROAS. The next euro of spend earns marginal ROAS, which can sit below break-even when the average looks fine. One measurement vendor makes this argument about ROAS and MER¹¹.
Long-term value. It cannot see repeat purchase, brand effects or organic spill-over. Underwater in prospecting does not prove a loss over a customer's lifetime.
Returns and VAT. There is no returns field. Put returns in your margin, and keep revenue and margin on the same basis.

For ZFD: The range is the question. It is not the answer.

What should you do with the range?

Check each flagged channel against its threshold in the Break-even ROAS & contribution margin calculator.

Next, decide whether a test is worth it. A holdout needs enough volume to detect the effect you care about, a testing vendor explains¹². It also gives up sales in the control group, a trade publication notes⁸, and it takes time. Read when not to run an incrementality test first.

Concepts used

ROAS, MER, blended ROAS, break-even ROAS, contribution margin, iROAS, incrementality factor, attribution window, holdout test, marketing-mix modelling.

FAQ

Is the adjusted ROAS my real ROAS?

No. It is an estimate from fixed assumptions. Only a test, or a model calibrated with tests, shows the real effect on your account.

My blended ROAS looks fine. Why check channels?

In our convention, blended ROAS is total revenue divided by media spend only, and MER divides by all marketing spend. A good average can hide a channel that loses money on every order.

Why a range and not one number?

A single number gets quoted without its assumptions. A range forces the conversation about what you do not know.

Do I need to give an email address?

No. There is no email gate.

Real ROAS calculator: is your ROAS real?

Calculate your adjusted ROAS (estimate)

Channels

Reported vs. adjusted ROAS, channel by channel

What does the calculator do?

How does it work, and what does it assume?

What should you do with the range?

Concepts used

FAQ

Is the adjusted ROAS my real ROAS?

My blended ROAS looks fine. Why check channels?

Why a range and not one number?

Do I need to give an email address?

Sources

Next step