Your ROAS Is Lying to You (And What to Track Instead)

The Number That Feels Good (But Is Usually Wrong)

Your Google Ads dashboard is showing a 4.2× ROAS. Your agency sent you a report with a green arrow and the word “strong.” You feel like the ads are working. You keep spending. Here’s the problem: that 4.2× is almost certainly wrong — not because Google’s tracking is broken, but because the number is built on a set of attribution assumptions that don’t match how your customers actually buy. According to Forrester’s 2025 Marketing Measurement Report, 68% of digital advertisers overestimate paid media impact by 30–50% due to attribution model choice alone — a gap that compounds as media spend scales.

This isn’t an edge case. In our experience at Varnan auditing ad accounts across e-commerce, hospitality, and B2B, the reported ROAS is higher than the real ROAS in the vast majority of accounts — often by 30–60%. That gap is money being mis-allocated, scaled into campaigns that aren’t driving the growth they appear to.

“Your ROAS doesn’t tell you what’s working. It tells you what Google is taking credit for.”

— Amal Jandheer, Founder, Varnan

Why Does Attribution Break Your ROAS Number?

Google Ads, by default, uses a data-driven attribution model — which sounds sophisticated but has a fundamental limit. It can only attribute credit across Google’s own ecosystem. The customer who saw your Instagram ad, Googled your brand name three days later, and clicked a branded search ad before purchasing? Google gives 100% credit to the branded search ad. Your brand awareness spend gets zero.

The Three Attribution Distortions That Inflate ROAS

• Cross-channel blindness. Any touchpoint outside Google (Meta, YouTube non-TrueView, organic social, word-of-mouth) is invisible to Google’s attribution. If your funnel uses multiple channels — and almost every funnel does — your Google ROAS number is overcounting Google’s contribution.
• Brand vs. non-brand confusion. Branded search campaigns capture intent that already existed. If a customer would have found you organically anyway, your branded ROAS is inflated by credit for conversions that didn’t require the ad. Many accounts blend brand and non-brand ROAS without realising the brand campaigns are doing most of the heavy lifting.
• View-through attribution. In some setups, a customer who merely saw your display ad (never clicked) can trigger a conversion credit if they purchase within a 30-day window. This is one of the least understood — and most common — sources of ROAS inflation.

How Different ROAS Numbers Compare: A Real Example

The Real Test: Does Turning It Off Change Revenue?

The most uncomfortable question in performance marketing is: if I paused this campaign tomorrow, would my revenue drop?

For many branded search campaigns, the honest answer is: probably not much. Customers with strong purchase intent will find you organically, through direct visits, or through other channels. The branded campaign is often claiming credit for a journey it didn’t drive — it’s the last door before a destination the customer was already headed to.

This is why incrementality testing exists. Run a holdout group — 10–20% of your audience who don’t see your ads — and compare their conversion rate to the exposed group. The difference is your real lift. The number is almost always lower than ROAS suggests. We ran this test for a B2B SaaS client at Varnan and found that their branded search “ROAS” of 12× was actually delivering an incremental lift of approximately 2.8× — meaning 77% of the reported credit was for conversions that would have happened organically.

What Should You Track Instead of ROAS?

Abandoning ROAS entirely would be wrong — it’s still a useful relative metric for comparing campaigns within the same channel. But running a business on ROAS alone is how you end up scaling waste.

These three metrics give you a more honest picture:

The Practical Fix: Build a Truth Dashboard

You don’t need to abandon Google Ads reporting. You need a second layer — a “truth dashboard” that sits outside the platforms and reconciles what you’re actually seeing in the business against what the dashboards claim.

Here’s what we build for every client account at Varnan:

• A single source of truth for revenue. Pull actual revenue from your payment processor or CRM — not from platform-reported conversions. This is your denominator.
• Total spend across all channels in one row. This gives you MER without platform attribution affecting it.
• Brand vs. non-brand ROAS split. Report these separately. If your non-brand ROAS is below target while brand ROAS looks great, you have a discovery problem — not a performance success.
• New customer ROAS separate from returning customer ROAS. Most e-commerce businesses should accept lower ROAS on new customers (you’re buying long-term LTV) and higher ROAS on returning (they already know you, you’re just reminding).

The Uncomfortable Truth Worth Knowing

Most agencies don’t show you this because a lower — but accurate — ROAS makes their work look worse. A 2.8× real ROAS on non-brand campaigns with incremental testing is more valuable than a 4.2× reported ROAS that includes credited-but-not-caused conversions. But 2.8 is a smaller number, and smaller numbers are harder to sell.

Our rule at Varnan: we’d rather show you a number you can trust and build on than a number that feels good and leads you to bad decisions. If your current reporting setup doesn’t pass the “would revenue change if I paused this?” test, it’s time to rebuild the dashboard.

We audit ad accounts as the first step in every new client engagement — free, no pitch. If you want to know what your ROAS is actually telling you, book a call here.

Frequently Asked Questions

Why is my Google Ads ROAS higher than my actual return?

Because ROAS measures credit, not causation. Google’s attribution model gives full credit to whichever touchpoint your settings assign it to — even if that touchpoint didn’t cause the sale. Across the accounts we audit in e-commerce, hospitality, and B2B, reported ROAS sits above real ROAS by 30–60% in most cases.

What is view-through attribution and why does it inflate ROAS?

View-through attribution credits a conversion to an ad the customer merely saw — never clicked — if they purchase within a set window (often 30 days). It’s one of the least understood and most common sources of inflated ROAS, because it counts purchases that may have happened anyway as if the ad caused them.

How do I know if my branded search campaigns are actually driving sales?

Run the “would revenue drop if I paused this?” test. For most branded search campaigns, the honest answer is probably not much — customers with strong purchase intent would find you organically or through direct visits. An incrementality test with a 10–20% holdout group that doesn’t see your ads gives you the real lift number, which is almost always lower than ROAS suggests.

What metrics should I track instead of ROAS?

Three give a more honest picture: MER (Marketing Efficiency Ratio — total revenue ÷ total ad spend across all channels), nCAC (ad spend ÷ new customers only, stripping out repeat buyers), and Contribution Margin ROAS (gross profit ÷ ad spend, which accounts for COGS). ROAS still has a role for comparing campaigns within the same channel, but running a business on ROAS alone leads to scaling waste.

What is a ‘truth dashboard’ and how do I build one?

A truth dashboard is a reporting layer that sits outside the ad platforms and reconciles what the dashboards claim against what’s actually happening in the business. It pulls revenue directly from your payment processor or CRM, totals spend across all channels in one row for true MER, and splits ROAS by brand vs. non-brand and new vs. returning customers so you can see where performance is real.

Is ROAS still useful at all in 2026?

Yes — for relative comparisons within the same channel and the same attribution model. ROAS is useful for comparing Campaign A vs. Campaign B on Google Ads, or Ad Set 1 vs. Ad Set 2 on Meta. It breaks down as a cross-channel or business-level metric because it doesn’t account for channel interactions, brand vs. non-brand dynamics, or incremental causation. Use it for in-platform optimization; use MER and nCAC for business decisions.