What budget decisions on reported ROAS alone cost

Funding channels on platform-reported numbers produces three predictable misallocations, and all three are estimable from data you already hold. The total is usually larger than any tooling decision under discussion.

The three

Sizing the first

Take the channels whose reported ROAS is high and whose audience is largely people who already know you: branded search, retargeting, and often email inside platform-reported views. Take their combined spend. A causal read typically returns lower estimates for these, and the difference between funding at the reported level and funding at the causal level is the first number.

Do not treat that difference as pure waste. Some of it is defensible for reasons other than incremental revenue, such as defending brand terms. But it should be a decision rather than an artefact of the reporting.

Sizing the second

Harder, because you are estimating something that did not happen. The proxy is: which channels have you cut in the last two years on efficiency grounds, and were any of them upstream channels whose contribution would show up as other channels' credit? The wasted ad spend calculator is a starting point for the arithmetic.

The third is the quiet one

Channels below the level of spend at which any method can separate an effect show up with no number. In a review, no number reads as no value, and they get cut. That is a decision taken on the absence of evidence rather than on evidence of absence, and it is the most common way small experimental channels die.

The honest handling is a three-way report: measured and above threshold, measured and below threshold, and not measurable at this spend. The floor arithmetic is in the measurability floor and the report structure in which channels to cut, for the CFO.

The comparison worth putting on one line

Your monthly ad spend, times the share going to channels credited primarily on proximity, times a conservative correction factor from your own gap analysis. Set that against the cost of measuring properly, which starts at a €99 one-time read on a Google Analytics export, refundable if it does not move a budget decision.

For most brands spending meaningfully on ads the arithmetic is not close, and the reason the decision gets deferred is not cost, it is that nobody has run the line.

What the read gives you

Per channel: an estimate, its confidence interval, the coverage share of your orders, and a design label. Plus an explicit list of what could not be measured, so absence of evidence stays visibly distinct from evidence of absence. The interactive demo shows it with no signup.

The one thing not to conclude

None of this means your ads do not work. It means the number you are funding on was built to answer a different question, and the misallocation is a consequence of the mismatch rather than of the channels themselves.

Related answers

• Why platform ROAS and causal ROAS disagree
• How to test the gap between reported and causal ROAS
• Showing the reported vs causal ROAS gap to a team
• The cost of an attribution report nobody opens
• The cost of waiting for a data hire