Breaking Free from Misleading Ad Results: Using First-Party Data for Smarter Measurement

As marketers, we’re all familiar with the sinking feeling when a client asks, “So… how do we know this is working?”

You’re not alone — according to recent research, 39% of marketing decision-makers struggle to measure the impact of each marketing channel. Personally, I think that number is conservative. Measurement has always been difficult, but in the last few years, it’s become exponentially more complex.

Welcome to what Gen Z might call “a new era” — one with reduced tracking, weakened targeting, and the slow, painful death of the cookie.

The Privacy-First Shift

Not long ago, digital advertising was the Wild West. Data was free-flowing, user tracking was easy, and regulation was minimal. Then came GDPR, cookie banners, and new legislation across the globe — reshaping how marketers collect and use personal data.

Meanwhile, ad blockers and Apple’s App Tracking Transparency (iOS 14) have empowered users to opt out of tracking entirely. As a result, marketers face an increasingly fragmented and opaque measurement landscape, with major platforms like Google, Meta, and TikTok operating in silos.

And while GA4 was meant to help, it’s brought its own challenges:

• Session-based insights are harder to access.
• Attribution models have shifted from last-click to data-driven (but in less transparent ways).
• Cross-platform visibility remains extremely limited.

Put simply: our ecosystem has lost clarity.

Why Paid Social Measurement Is So Hard

Paid social has never been a last-touch channel. It influences rather than closes. That makes it inherently tricky to measure — especially when most clients live inside Google Analytics dashboards that favor last-click attribution.

Imagine a customer who sees a brand’s ads across Spotify, Instagram, and connected TV, before finally Googling the product and buying it. GA will give all the credit to Google Search — even though 84% of paid search conversions are driven by other factors, not paid search.* (Admittedly, this is according to Meta’s research, who have some skin in the game.)

That’s what I call the streetlight effect: searching for your keys only where the light is shining, even if they’re probably in the dark.

The Problem with Platform Attribution

1. Meta’s View

Meta offers powerful in-platform data: granular audience insights, better cross-device tracking, and quick cost-efficiency feedback. It’s great for optimizing campaigns within Meta.

But — and it’s a big “but” — Meta also loves to overclaim credit. It’s blind to other channels, assumes full ownership of any conversion it can see, and can’t be easily validated.

In one of our client examples, Meta reported a massive spike in conversions. The reality? The client had just launched a sale and sent out a series of email newsletters. Meta simply took credit for purchases that were already happening.

2. Google’s View

Google Analytics, on the other hand, under-credits social activity — ignoring the impact of impressions, view-throughs, and ad engagement.

So we’re stuck between two extremes: Meta’s over-attribution and Google’s under-attribution. Neither gives the full picture, and both fail to capture offline or multi-channel activity.

Attribution vs. Measurement

It’s important to distinguish the two:

• Attribution matches clicks to purchases — it’s user-level, immediate, and cookie-dependent.
• Measurement takes a statistical approach — it’s broader, long-term, and privacy-resilient.

Attribution helps you optimize short-term spend.
Measurement helps you understand true effectiveness.

Both have their place, but if you want to make strategic decisions, you need to move beyond attribution and toward incrementality.

Understanding the “Messy Middle”

We’ve all seen the classic marketing funnel: Awareness → Interest → Desire → Action.

Reality isn’t that neat anymore. Google calls it the Messy Middle — a complex web where consumers jump in and out of research, evaluation, and purchase phases across multiple channels.

The good news? Consumers are more open to discovering new brands than ever before. The bad news? Proving which of your activities actually moved them is harder than ever.

From Attribution to Contribution

To truly understand impact, marketers are increasingly using three complementary methods:

1. Multi-Touch Attribution (MTA) – Still valuable for digital optimization.
2. Incrementality Testing – Isolating a channel’s true impact through controlled experiments.
3. Marketing Mix Modeling (MMM) – Combining everything into a holistic, forecast-driven view.

Let’s focus on incrementality testing, because that’s where first-party data and smarter experimentation shine.

Case Study: Norse Atlantic Airways

When we began working with Norse Atlantic Airways, they didn’t even have a website. Fast forward four years, and they’re flying globally with over a million passengers booked.

But as a challenger brand competing with legacy airlines like British Airways and Virgin Atlantic, Norse couldn’t rely on natural demand or brand familiarity. Paid social — especially Meta — was a key driver of awareness and intent.

However, GA was showing Meta contributing less than 5% of revenue, leading to the question: “Should we just turn off social and see what happens?”

We knew that wasn’t the right approach. Instead, we proposed a geo holdout test — turning off Meta ads in 50% of Norway for four weeks, while keeping everything else constant.

What Happened?

Within a week, revenue in the “no-ads” regions fell 40% below forecast. The following week, it was still 20% down. We ended the test early — confident that Meta ads were driving real incremental value.

When we compared our findings to GA data, the difference was staggering:

• GA4 said: Meta = <5% of revenue
  
• Incrementality test showed: Meta = 32% of revenue
  

It was a powerful demonstration that digital attribution alone had been grossly underestimating paid social’s contribution.

What’s Next for Norse?

Following the success of the experiment, Norse has fully embraced a test-and-learn mindset:

• Running new incrementality tests for TikTok and PPC.
• Exploring positive tests (scaling spend rather than cutting it).
• Building their first marketing mix model to integrate all channel data and forecast future investment.
  

Steps to Improve Your Measurement Maturity

No matter your client size or budget, you can start evolving your approach:

1. Get your data foundations right.
   
   • Ensure clean analytics, deduplication, server-side tagging, and first-party tracking (like CAPI).
     
2. Start with small experiments.
   
   • Launch geo holdout or audience split tests to gather real-world impact data.
     
3. Build and refine your model.
   
   • Use the results of your experiments to feed into a marketing mix model that evolves with your brand and environment.
     

In summary

Don’t put blind faith in bad numbers. Avoid the streetlight effect — the temptation to look only where the data is easiest to find.

Instead, start small, start testing, and start measuring what truly matters. Because the future of performance marketing isn’t just about attribution — it’s about understanding contribution.

Amy Stamper is Head of Paid Social at Impression Digital. This article is based on the talk she gave at Hero Conf UK in April 2025, which you can watch in full below.