Why honest ROAS accounting exposes the budget decisions most CMOs avoid

Return on ad spendReturn on ad spendReturn on Ad Spend (ROAS) measures the revenue generated for every unit of currency spent on advertising, calculated as revenue divided by ad cost.View full definition → sits at the center of nearly every performance marketing review in 2026. AttributionAttributionA framework for assigning credit to the touchpoints that contributed to a conversion, so you can measure which channels and interactions actually drive results.View full definition → platforms have grown more sophisticated, privacy regulations have pushed marketers toward modeled data, and CFOs have grown sharper at demanding proof. The result is a pervasive confidence in ROASROASReturn on Ad Spend (ROAS) measures the revenue generated for every unit of currency spent on advertising, calculated as revenue divided by ad cost.View full definition → as a reliable compass. Boards cite it. Agencies defend their contracts with it. CMOs use it to justify budget requests.

The problem is not the metric. The problem is the selective arithmetic that surrounds it.

The consensus view: ROAS is a mature, actionable metric

The standard position is reasonable. ROAS (revenue divided by ad spend) gives marketers a fast read on channel efficiency, allows comparison across campaigns, and connects directly to financial reporting. Platforms like Google and Meta (both vendors, with obvious commercial interests in high-reported ROAS figures) have invested heavily in measurement infrastructure: conversion APIs, enhanced conversions, data-drivendata-drivenAn approach where decisions are systematically informed by data analysis rather than intuition alone.View full definition → attribution. The consensus holds that better data inputs produce better ROAS figures, and that optimizing toward those figures improves business outcomes.

There is genuine logic here. A brand running five channels with wildly different conversion rates does need a shared unit of measurement. ROAS, for all its flaws, is understood by finance teams and can survive a board presentation. Alternatives like marketing-mix modeling (MMM) or incrementality testing are slower, more expensive, and harder to explain. The consensus is not wrong to prefer operational simplicity when speed matters.

Where the consensus quietly fails

The trouble starts with what gets counted as "ad spend" and what gets counted as "revenue."

Most ROAS calculations include only paid mediapaid mediaVisitors arriving via paid ads or sponsored placements, where you pay a platform to display your message rather than earning visits organically.View full definition → spend: the line item on the agency invoice or the platform billing statement. They exclude agency fees, creative production costs, the salaries of the internal team managing the campaigns, technology fees for the attribution platform itself, and the cost of the offers or discounts embedded in the ads. When Meta reports that a campaign generated a 6x ROAS, that figure typically reflects platform spend against attributed revenue. It does not reflect the 15% discount code in the ad, the creative agency retainer, or the attribution tool that cost $8,000 a month.

A more complete accounting, sometimes called "true ROAS" or "blended ROAS," divides total revenue by total marketing investment. When companies run that calculation honestly, the resulting figures are routinely 30 to 50 percent lower than platform-reported ROAS. That gap matters because budget decisions get made on the inflated number.

The attribution layer compounds the problem. Last-click attribution overweights the final touchpoint and systematically rewards retargetingretargetingShowing ads to users who have previously visited your site or interacted with your brand, to bring them back and drive conversion.View full definition → campaigns, which reachreachThe number of unique people exposed to your message in a given period. Unlike impressions, reach counts each person once, no matter how often they see it.View full definition → people who were likely to convert anyway. Data-driven attribution from Google (a vendor whose advertising revenue depends on continued spend) is an improvement, but it is still built on proprietary models that cannot be independently audited. Research from the Data & Marketing Association has consistently shown that cross-channel attribution in complex purchase journeys carries error margins that make precise channel comparison unreliable. Marketers treating a 4.2x versus a 3.8x ROAS difference as a meaningful signal are often reading noise.

Then there is the incrementality problem. A campaign showing a strong ROAS may simply be capitalizing on existing demand rather than generating new revenue. Byron Sharp's work at the Ehrenberg-Bass Institute, which is academically independent, demonstrates that brand search campaigns often capture intent that would have converted organically. The ROAS looks excellent. The incremental revenue contribution is close to zero. Netflix ran incrementality tests on its direct response campaigns in the early 2020s and reportedly found that a significant portion of attributed conversions would have occurred without the ad. That kind of finding is what honest ROAS accounting eventually forces you to confront.

The second-order effect of bad ROAS accounting is a predictable and damaging pattern: upper-funnelfunnelThe customer journey from awareness to purchase, typically Awareness, Interest, Consideration, Decision, Action, with prospects narrowing at each stage.View full definition → and brand-building spend gets cut because it produces poor reported ROAS, while lower-funnel retargeting and brand-search campaigns get scaled because their reported ROAS looks strong. Over two or three years, the brand weakens, demand generationdemand generationMarketing activities designed to attract and capture contact information from prospects interested in your offer, creating a pipeline of potential customers.View full definition → slows, and the lower-funnel campaigns start to underperform because the pipelinepipelineAll active sales opportunities across the stages of the sales process, together with their combined potential value and probability of closing.View full definition → feeding them has thinned. Finance then asks why revenue growth is stalling despite strong ROAS. The CMO does not have a clean answer.

What a sharp operator should actually do

The first move is definitional. Establish a single, agreed definition of ROAS that includes all costs, not just platform spend. This means sitting down with finance and agreeing on what goes into the denominator: media spend, agency fees, creative costs, martech licensing, discounts embedded in paid offers. It is not a comfortable conversation, because the resulting number is lower and harder to defend. Do it anyway. The CMO who controls the definition controls the narrative.

The second move is to separate measurement methods by decision type. ROAS from platform dashboards is useful for daily campaign optimization, where speed matters and directional accuracy is sufficient. It should not be used for budget allocation across channels or for annual planning. Those decisions need MMM or incrementality testing, which take weeks rather than hours but produce findings that are not distorted by platform attribution logic. Several large consumer goods companies, including Unilever and P&G, have built in-house MMM capabilities specifically to reduce dependence on vendor-reported figures. The investment pays back quickly when it prevents a $5 million misallocation.

The third move is to institutionalize an incrementality testing calendar. Pick four to six campaigns per year, run geo-based holdout tests or intent-matched control groups, and measure actual lift. The results will sometimes be uncomfortable. A brand-search campaign that looks like a ROAS star may test as nearly worthless. A video awareness campaign with a reported ROAS of 1.2x may show incremental revenue that justifies doubling the spend. Let the tests change the plan.

The CMO who does this work will at some point have to tell a leadership team that a campaign everyone celebrated was largely capturing demand that existed regardless. That conversation is difficult. It is also the one that separates marketing leaders who manage budgets from those who actually improve business outcomes.

Honest ROAS accounting does not require perfect data. It requires the willingness to use a definition of the metric that finance would recognize as rigorous and to accept what the numbers say when they contradict comfortable assumptions. Most CMOs have the analytical tools to do this in 2026. The limiting factor is not capability.