The "Triangulation" Framework: Why Your ROAS is a Lie
By: Martin Grozev | Performance Marketing Specialist 8 Years Experience | $3M Managed Ad Spend |
There is a massive, embarrassing gap in modern marketing. It’s the gap between what your Facebook/Google Ads Manager says you made, and what is actually sitting in your bank account.
You spend €50,000. The agency reports a 4.0 ROAS (€200k revenue). You high-five the team. But when you look at your P&L, total revenue only went up by €80k.
Where is the missing €120k?
It didn't disappear. It never existed. Platform dashboards are designed to take credit for every sale they touch, even the ones that would have happened anyway.
If you want to scale past €1m ARR, you need to stop looking at Attribution (who touched the user) and start looking at Incrementality (did the ad cause the sale?).
1. The "Blind Spot" (Why MTA is Dead)
Most startups rely solely on Multi-Touch Attribution (MTA)—the data inside Google Analytics or Facebook. This is dangerous. Relying on MTA today is like driving a car with a painted-over windshield.
The Hard Data: Since the release of iOS14, global opt-in rates for app tracking have flatlined at approximately 15%. That means ~85% of your iPhone users are effectively invisible to your pixel.
When a user sees your Instagram ad, clicks, and buys 2 days later, Facebook often cannot "see" that chain of events. So, it reports 0 ROAS. In reality, the ad worked. The dashboard just lost the signal. You cut the budget because the dashboard said "fail," and suddenly your total revenue drops. You were optimizing for a broken metric.
2. Incrementality: The eBay "Zero Lift" Lesson
The easiest way for an agency to show you a high ROAS is to spend your money on Retargeting and Branded Search. Why? Because those people were probably going to buy anyway.
The Science: In a famous large-scale study, eBay turned off all paid search advertising for their brand keywords across 30% of the US market.
Prediction: Sales would collapse.
Result: Sales remained identical.
The study concluded that branded search ads had "no measurable benefit" because users simply clicked the organic link when the ad wasn't there. If your agency is reporting a 10x ROAS on "Brand," they aren't generating revenue. They are just taxing your organic traffic.
The Fix: The "Holdout" Test To find the truth, we run Conversion Lift Studies. We take 10% of your audience and put them in a "Control Group." We deliberately hide your ads from them.
Group A (Saw Ads): Converts at 5.0%
Group B (Holdout - No Ads): Converts at 4.5%
The Incrementality is only that 0.5% difference. If your dashboard shows a 5.0 ROAS, but your incremental lift is near zero, you are setting money on fire.
3. Media Mix Modeling (MMM): The "Macro" Truth
If Incrementality is the "Truth Serum," MMM is the "Weather Report."
Since we can't track every user (thanks to that 85% signal loss), we stop trying to track people and start tracking variables. Media Mix Modeling (MMM) uses statistical regression (econometrics) to analyze the relationship between spend and revenue.
We feed the model:
Daily Spend (Meta, Google, TikTok)
Price changes
Seasonality
Total Verified Revenue
The model calculates the correlation coefficient. It tells us: "When you spend €1 extra on YouTube, total revenue rises by €1.50, even though the YouTube dashboard says it generated nothing." It captures the invisible impact of top-of-funnel channels that do the heavy lifting but don't get the click credit.
4. Allocating the Next €10k (The Operator's Move)
This is where the rubber meets the road. Most founders look at their dashboard and see:
Meta Retargeting: €5 CPA (ROAS 10.0)
YouTube Prospecting: €80 CPA (ROAS 0.8)
A junior media buyer says: "YouTube is failing. Cut the budget and put it all into Retargeting." This is a death spiral. They are cutting the channel that brings in new people to double down on the channel that harvests people you already have.
The Operator’s Allocation Strategy: We allocate budget based on Incremental CPA (iCPA).
Meta Retargeting: Dashboard CPA €5. But since 90% would have bought anyway (eBay Study logic), the iCPA is actually €50.
YouTube Prospecting: Dashboard CPA €80. But since these are 100% net new users, the iCPA is €80.
Suddenly, the gap closes. YouTube is the engine of new growth. Retargeting is just a tax. We move the next €10k into the channels driving net new customers, even if the dashboard looks uglier.
The Bottom Line
Stop optimizing for a pretty dashboard. Optimize for the bank account.
Acknowledge the Blind Spot: 85% of iOS data is missing.
Test for Lift: Run holdout tests to expose the "Retargeting Lie."
Triangulate: Use MMM to see the big picture.
We don't just report ROAS. We measure Lift. Let's audit what's actually working.