Why Your Google Ads ROAS Is a Lie, and What to Measure Instead

Google Ads ROAS is misleading because the platform reports its own performance, using a measurement system whose vendor controls the auction, the placement, the tracking and the attribution window. Three structural mechanisms inflate the number reported in the dashboard:

• Last-click attribution credits Google for conversions that upstream channels like Meta, YouTube or LinkedIn actually created.
• Branded search campaigns report 8:1 to 20:1 ROAS by billing for demand that already existed before the ad was served.
• Cross-platform double-counting makes the sum of platform-reported revenue exceed actual booked revenue by 30 to 60 per cent on accounts running a full paid mix.

The metric that replaces it is incremental ROAS, measured through geo holdouts, conversion lift studies or marketing mix modelling. The metric that drives budget decisions is marginal ROAS, the return on the next euro of spend, which is what reveals channel saturation. The operating model that holds it together is a three-layer framework that separates platform optimisation, incremental measurement and budget decision, so that Google Ads stops being judge, jury and witness on its own performance.

The ROAS You See in Google Ads Is Not the ROAS You’re Getting

A pattern keeps surfacing in CMO conversations through Q1 2026: the Google Ads dashboard shows ROAS climbing month over month, the team scales spend behind those numbers, and somewhere between month three and month six the revenue line goes flat. The honest explanation is that nothing changed except the team’s willingness to trust a single number reported by a platform that simultaneously runs the auction, sets the attribution window, and decides which conversions belong to which click.

No other media vendor operates with that vertical control. The ROAS shown in the platform is an optimisation signal for the bidding algorithm, not a decision signal for the budget. Treating it as both is the original sin of paid media measurement, and it is why most accounts overstate their own contribution by a factor that grows quietly every quarter.

Last-Click Attribution Hands Google Credit for Conversions It Didn’t Cause

Last-click attribution, and its lightly upgraded successor data-driven attribution inside Google Ads, assigns the bulk of conversion credit to whichever paid touchpoint sat closest to the sale. That is structurally favourable to Google in a way that becomes visible only when you trace a single customer journey end to end. Consider the prospect who sees a Meta ad on a Tuesday, watches a YouTube explainer two days later, reads a LinkedIn post from an account executive the following week, then searches your brand name on Google and clicks the sponsored result before buying. In Google Ads, that is one paid search win. In reality, the paid search click was the cheapest, easiest part of the work: it intercepted demand three earlier touchpoints had created. Eighty per cent of the value sits upstream, but zero per cent of it gets reported there.

Scale that asymmetry across an enterprise account and the consequence becomes mechanical. As long as upstream channels keep generating demand, Google Ads keeps converting it and posting a strong ROAS. The dashboard rewards harvesting and punishes seeding. Teams respond rationally: they cut the channels that look weak, push more budget into Google, watch the ROAS climb for one or two quarters. Then the upstream demand they stopped funding runs out, and the Google ROAS collapses. Not because Google got worse, but because there is nothing left to harvest.

Branded Search Inflates Your ROAS Without Creating Any Value

The second inflation mechanism sits inside almost every enterprise paid search account. Branded campaigns, bidding on your own company name, consistently report ROAS in the 8:1 to 20:1 range and pull the account-wide average upward by sheer weight of efficiency. The problem is that the overwhelming majority of those conversions would have happened without the ad. The user typed your brand into the search bar; they were going to find you with or without paid placement. What you are actually buying on branded keywords is:

• Defensive coverage against competitors who bid on your name
• A marginal lift from users who would otherwise have clicked the organic listing one position below
• Nothing more

Measured through brand holdout tests, the incremental ROAS of branded search typically sits an order of magnitude below the reported figure.

The downstream effect is what matters. An account where 40% of spend sits in branded campaigns may report a blended ROAS of 6:1 while the incremental ROAS of the non-brand campaigns, the ones actually meant to acquire new customers, is closer to 1.5:1. The CFO sees 6:1 and approves a budget increase. The team scales non-brand. The acquisition engine runs at break-even or below, and the gap between dashboard reality and P&L reality widens for another quarter.

Cross-Platform Double-Counting Makes Your Total ROAS Mathematically Impossible

The third mechanism only becomes visible when paid search reporting is read alongside paid social, paid video and affiliate dashboards. A prospect who clicks a Meta ad on Monday and a Google ad on Wednesday before purchasing on Thursday produces, in the platform dashboards, two attributed conversions. In the CRM, there is one sale. Multiply that overlap across thousands of conversions per month and the sum of platform-reported revenue routinely exceeds actual booked revenue by thirty to sixty per cent on mid-market accounts running a full paid mix.

This is why blended dashboards aggregating platform-reported ROAS are structurally broken. They are not approximately right with some noise: they are systematically wrong, and the error grows with the number of platforms aggregated. The only ROAS that can be reconciled with the finance function is the one computed from a single source of truth: total recognised revenue from the CRM or ERP, divided by total spend across every channel. Everything else is a vendor’s marketing material.

Incrementality Is the Only Question That Actually Matters

Once the three inflation mechanisms are visible, the right question changes shape. It is never “how much revenue did this campaign generate”, which is unanswerable from inside the platform. It is “how much revenue would have been generated without this campaign”. The gap between those two numbers is the incrementality, and it is the only ROAS-adjacent metric that maps cleanly onto a budget decision.

Incrementality is measured, not reported. Three practical methods exist, ordered by rigour:

• Geo holdouts. Pause paid activity in a comparable region for a defined window and measure the revenue delta against control regions. The gold standard for any account with sufficient geographic spread.
• Conversion lift studies offered by the platforms themselves. Useful directional reads, with the caveat that the vendor is auditing its own contribution.
• Marketing mix modelling, viable above a certain budget threshold. Isolates each channel’s incremental impact across the full media plan and produces both a coefficient of correction for platform ROAS and a marginal ROAS curve for forward planning.

The shift is operational, not technical. Incrementality is a quarterly protocol that recalibrates every other number on the dashboard, not a monthly metric.

Marginal ROAS Tells You Where the Next Dollar Should Go

Average ROAS is a backward-looking summary. Marginal ROAS is a forward-looking decision. A channel can report a healthy average ROAS of 4:1 across the quarter while its marginal ROAS, the return on the next euro of incremental spend, has already dropped to 0.8:1. That is saturation: the dashboard does not flag it, because the dashboard reports the average.

The reverse pattern matters as much. An upper-funnel prospecting channel may show an unimpressive average ROAS of 2:1 while its marginal ROAS sits above 2.5:1, a clear signal that the next increment of budget will outperform the current portfolio. The implication for any paid media agency planning the next quarter is straightforward: budget arbitrage belongs on marginal ROAS, not on average ROAS. Google Ads in bottom-funnel converts demand that already exists, which means it reaches marginal saturation faster than channels that create demand. The number that justifies scaling Google is rarely the same one that justifies sustaining it.

The Three-Layer Measurement Framework That Replaces ROAS Reporting

The framework that holds all of this together has three concurrent layers, not three sequential steps. The separation between them is what prevents Google Ads from being judge, jury and witness on its own performance.

• Layer 1, platform optimisation. ROAS and target ROAS bidding inside Google Ads remain useful for what they were built to do: steer the bidding algorithm in real time. The team running the google ads agency relationship, or the in-house equivalent, operates this layer day to day. It is not used for budget decisions.
• Layer 2, incremental measurement. Quarterly holdout tests, broken out by channel and by campaign type, produce a coefficient of correction that translates platform ROAS into incremental ROAS. Branded search might be multiplied by 0.15, non-brand by 0.5 to 0.7, prospecting by a coefficient calibrated against the holdout. This layer is operated by analytics or marketing science, deliberately separated from the team buying the media. For mid-market accounts, this is often where the case for an external ppc agency becomes structural rather than tactical: not for execution capacity, but for the independence of measurement.
• Layer 3, budget decision. Marginal ROAS, derived from the first two layers and cross-referenced with customer lifetime value and product margin, drives the arbitrage. This is the layer the CMO and the finance function consume. The only one of the three where the question “should we increase the Google Ads budget next quarter” has a defensible answer.

Three layers, three stakeholders, three different questions. The dashboard that reports all three side by side is the one that ends the quarterly argument between marketing and finance about what the ROAS actually means.

Recap and Next Step

Three mechanisms (last-click bias, branded search inflation, cross-platform double-counting) explain why the Google Ads ROAS misleading pattern is structural rather than accidental. Incrementality is the only metric that produces a decision-grade number, and a three-layer framework separates the optimisation signal that runs the algorithm from the decision signal that runs the budget. The teams that get this right are not the ones with the most sophisticated dashboard. They are the ones who stopped letting a single platform grade itself.

If the gap between your Google Ads dashboard and your CRM is starting to feel structural, a 30-minute incrementality readiness review with Primelis maps where your reported ROAS is most likely overstating reality, and which holdout tests will give you a defensible number for the next budget cycle.

FAQ: Common Questions About Google Ads ROAS and Incrementality

What is a good ROAS for Google Ads?

There is no universal benchmark. A good ROAS depends on product margin, customer lifetime value, and the share of branded versus non-brand spend in the account. A reported 4:1 on a branded-heavy account can be worse, on an incremental basis, than a reported 2:1 on a pure prospecting account.

Why does Google Ads show a higher ROAS than my CRM?

Three reasons: last-click attribution credits Google for conversions other channels created, branded search counts demand that pre-existed, and the platform’s attribution window is typically longer than the one used by finance. The CRM number is the one the business actually banks.

What is the difference between ROAS and incremental ROAS?

ROAS measures revenue attributed to a campaign by the platform’s own model. Incremental ROAS measures the revenue that would not have occurred without the campaign, established through a holdout test or marketing mix model. The two can differ by a factor of five or more on branded search.

How do you measure the incrementality of Google Ads?

The most reliable method is a geo holdout: pause Google Ads in comparable regions for several weeks and measure the revenue delta against control regions. Conversion lift studies and marketing mix modelling are complementary, with the caveat that platform-run studies should be cross-checked independently.