Showing the reported vs causal ROAS gap to a team
Present a ROAS gap badly and every channel owner hears a performance review. Four framings that keep the conversation about the instrument rather than the person.
Three platforms, three answers, one budget to defend. Your GA4 export already holds the answer, and the first finding is free and never leaves your browser. Get your first finding free
By Joris van Huët, Founder & CEOUpdated 3 min read
Run the numbers for your store: the free break-even ROAS calculator.
Show a channel owner that their reported ROAS is inflated and they will hear a judgement about their work, because that is what it sounds like. The gap is a property of the measurement, not of the person running the channel, and the presentation has to make that unmistakable.
Four framings that work
| Instead of | Say |
|---|---|
| "Branded search is overstated" | "Branded search is credited for demand other channels created" |
| "Your ROAS is wrong" | "The platform's rule counts touches; we are asking a different question" |
| "We are cutting this" | "We want to test this properly before deciding" |
| "The model says" | "Here is the estimate, the interval and what would change it" |
The second column is not softer, it is more accurate. The platform number is not wrong; it is an answer to a question about touches. Saying so is both kinder and more correct, which is a rare combination.
Lead with the structural explanation
Explain the mechanism before showing any channel's figure. Branded search captures demand created elsewhere. Retargeting reaches people already intending to buy. View-through crediting counts a weak signal. Once the room understands why gaps appear where they do, the specific numbers land as confirmation of a mechanism rather than as an accusation.
The mechanism is set out in platform attribution overcounting and the credit-versus-causation distinction in the correlation versus causation problem.
Show the intervals, including the wide ones
Present the confidence intervals alongside the estimates and point out the widest one yourself. This does two things: it shows the method is not claiming more than it has, and it gives the channel owner a legitimate route to push back, which they will need in order to accept the rest.
Give the channel owner the next move
Ask which channel they would most want tested properly, and then test it. That converts the meeting from a verdict into a plan, and the person closest to the channel usually has the best instinct about where the model is likely wrong.
They are often right. Aggregate methods cannot see creative fatigue, auction dynamics or what was tried last quarter, and the channel owner can. Settling attribution debates with your agency covers the same dynamic with an external party.
What not to do
Do not present the gap and a budget decision in the same meeting. The two together guarantee the discussion becomes about the decision, and the measurement never gets examined on its own terms. Separate them by at least a week.
The underlying point
Nobody is being caught out. The platform is answering the question it was built to answer, the channel owner is optimising against the number they were given, and the gap is a property of the instrument. Say that first and the rest of the conversation is possible.
Related answers
Frequently asked questions
How do I tell a channel owner their reported ROAS is inflated?
Explain the mechanism before showing their number, so the finding lands as a property of the measurement rather than a judgement of their work. Branded search captures demand created elsewhere; retargeting reaches people already intending to buy.Should I show a ROAS gap and a budget decision together?
No. Presented together, the discussion becomes about the decision and the measurement never gets examined on its own terms. Separate them by at least a week so the method can be questioned before anything depends on it.What if the channel owner disagrees with the causal estimate?
Take it seriously and ask what they would test. Aggregate methods cannot see creative fatigue, auction dynamics or what was tried last quarter, and the person running the channel can. That is often where a model is genuinely wrong.
Go deeper: Incrementality testing, explained.
Sixty-second versions of these ideas: Causality Engine on YouTube Shorts.
Have an audience of ecommerce operators? Earn 20% recurring commission by recommending the €99 read.
Keep reading
Terms in this article
- AttributionAttribution identifies user actions that contribute to a desired outcome and assigns value to each. It reveals which marketing touchpoints drive conversions.
- CausationCausation is the relationship where a change in one variable directly causes a change in another.
- Confidence IntervalConfidence Interval is a statistical range of values that likely contains the true value of a metric. In marketing analytics, it quantifies uncertainty around estimates, indicating the precision of an outcome or causal effect.
- CorrelationCorrelation is a statistical measure showing a relationship between variables; it does not imply causation.
- RetargetingRetargeting is online advertising that targets users who have previously interacted with your website or content. Attribution analysis shows the causal role of retargeting in driving conversions and improving ad spend.