The uncomfortable truth about attribution and why chasing perfect measurement is holding you back.
We’ve worked with hundreds of scaling ecommerce brands, and there’s one conversation that happens in virtually every initial strategy session. It goes something like this: “Meta says our ROAS is 4.2x, Google claims 3.8x, but when we add it all up, the maths doesn’t match our actual revenue growth.”
The dirty secret of digital marketing? Perfect attribution is a myth, but you don’t need perfect measurement to make informed decisions about your media spend.
The Attribution Paradox: Why Platform Metrics Are Optimistic
Let’s start with some uncomfortable truths about how platforms report performance.
Every advertising platform has one primary job: to look as valuable as possible. But it’s important to understand that they’re not trying to deceive you; they’re optimising for what helps their campaigns work.
- Meta and Google’s algorithms need conversion data to target your ads effectively
- The more conversions they can claim, the better their systems perform
But this creates a fundamental problem. When Meta attributes a conversion because someone clicked your ad seven days ago, then Google claims the same conversion because they clicked a search ad 3 days ago, and your email platform takes credit because they opened a newsletter, suddenly your total attributed ROAS looks fantastic, but it’s not a true reflection of your marketing performance.
The brands that scale successfully move beyond just using platform metrics to measure performance. Instead, they use what we call an Evidence-Based Attribution Framework: a systematic approach to gathering information from multiple sources to help them make informed decisions about how to optimise their marketing, whilst acknowledging that the grey areas still exist.
The Four-Step System for Better Paid Media ROI Measurement
After years of testing, iterating, and occasionally pulling our hair out over attribution discrepancies, we’ve developed a framework that helps bring clarity to the chaos. Think of it as building a case in court: you need multiple pieces of evidence that point to the same conclusion.
Tier 1: Platform Metrics (The Starting Point)
Platform metrics aren’t useless; they’re just incomplete. We use them as relative guides within their own ecosystem.
Key insight: If your Meta CPA drops from £45 to £32 whilst maintaining volume through prospecting campaigns, that’s a sign that things are likely working better. If your Google ROAS jumps from 3.2x to 4.7x after restructuring campaigns to focus on non-branded terms, then that indicates efficiency has improved.
The main thing is to use platform metrics to optimise within that platform, but never as your single source of truth for business decisions (which is something that the platforms themselves don’t recommend).
Where most brands go wrong: They spend 70% of their Meta budget on retargeting and existing customers because the platform metrics look incredible. The problem? You’re paying premium prices to show ads to people who were already going to buy.
What we do instead: We focus paid social spend on true new customer acquisition. New customer CPAs might look higher in-platform than retargeting campaigns, but they’re driving actual business growth. This approach differs from search advertising, where targeting existing demand and branded terms often delivers efficient results.
Real-world example: When hims faced challenges breaking into the UK market, we restructured their Meta and Google ad accounts, shifting from granular targeting to targeting broad audiences, enabling them to scale far more efficiently. We also optimised creative assets and messaging to resonate with UK consumers. The result? hims rapidly scaled whilst maintaining their target CPA.
Our work with hims demonstrates how the right strategic approach can unlock growth in new markets. You can read more about how we drove robust growth for hims & hers in the UK market and our Advanced Lifetime Value Analysis & Forecasting approach.
Tier 2: Top-Line Revenue Analysis (Your North Star)
This is where the magic happens. Instead of getting lost in the rabbit hole of attribution, we start with the simplest question: When you change your media spend, what happens to your revenue?
Running simple experiments to understand the effect of different marketing channels is a great place to start:
- Increase Meta spend by 40% for two weeks; did revenue grow proportionally or was it flat?
- Cut Google brand search investment by 50%; did sales decline or stay the same?
- Launch Pinterest advertising with a 10-20% of total budget; does revenue stay the same or decline?
The beauty of this approach is that it cuts through the noise and focuses on what actually matters, which is how your new customer revenue changes in response to marketing spend.
Tier 3: Third-Party Attribution (Your Reality Check)
Platforms like Triple Whale, Northbeam, and Google Analytics (and many others) provide a more conservative view of channel performance. They’re “harsher” by design, focusing on direct conversion paths which are directly attributable to your ads through click data.
Important caveat: Unless someone is browsing on multiple devices on the same IP address, these platforms can’t measure people who take a cross-device conversion path. This is what makes it such a harsh lens for measuring upper funnel channels (such as Meta, TikTok and Youtube).
Why this matters: These platforms help identify where your attribution in platform might be inflated by view through conversions, or the sources which are driving the highest quality traffic.
Pro tip: Don’t get discouraged when third-party attribution shows lower numbers than paid social platforms, particularly Meta and TikTok, as we often see click based models massively undervalue them. This is where your experiments (see step 2) really come into play, can you validate what Triple Whale/Northbeam is telling you through experimentation?
Tier 4: Advanced Measurement (The Gold Standard)
For brands spending £100k+ monthly on multiple paid media channels, advanced measurement becomes not just valuable but essential. We’re talking about media mix modelling (statistical analysis to measure each channel’s true impact) and conversion lift studies: the techniques that Google, Facebook, and Amazon use to measure their own marketing effectiveness.
Media Mix Modelling uses statistical analysis to isolate the true contribution of each marketing channel, accounting for external factors like seasonality, competitors, and economic conditions. It’s particularly powerful for measuring upper funnel channels like paid social, podcasts, out-of-home advertising, TV, or brand awareness campaigns.
Conversion Lift Studies create controlled experiments where a portion of your target audience never sees your ads. By comparing purchasing behaviour between exposed and unexposed groups, you can measure true incremental impact.
Real-world example: Pom Pom London sought a clearer understanding of their true business impact from paid social campaigns, particularly when expanding into colder audience segments.
We deployed Meta’s Incremental Attribution, a first-party measurement tool built directly into Ads Manager that isolates the incremental impact of ads beyond standard last-click metrics. Unlike traditional attribution, this uses machine learning trained on Meta’s extensive Lift study database to predict which conversions wouldn’t have occurred without ad exposure.
This allowed Pom Pom London to see a 94% improvement in incremental ROAS, providing a more accurate reflection of campaign effectiveness beyond what third-party attribution tools could measure.
With this data, we refined targeting and creative strategies to maximise efficiency and drive higher-quality conversions, especially in less saturated audience pools.
This case study demonstrates the power of incremental measurement in understanding true campaign impact. You can explore the full details of Meta’s Incremental Attribution delivering 94% iROAS improvement, plus see our additional experiments with Meta Advantage+ uplift testing and Google Broad Match optimisation.
Integration: Making It All Work Together
The art of understanding what works is collecting the evidence from these four different sources and then forming an hypothesis of how we can improve the efficiency of our paid media.
A great way to do this is using the following process to understand the problem and validate it through experimentation.
- Start with business impact: What do top-line revenue patterns tell us?
- Validate with experiments: Can we reproduce results through controlled tests?
- Cross-reference with attribution: Do third-party platforms support our hypothesis?
- Optimise using platform data: How can we improve performance within each channel?
Once we’ve run the experiment, proved that performance improved and implemented the optimisation, we can then assess it’s impact on the business as a whole, before moving onto the next one.
Example in action: A fashion brand noticed their overall revenue growth had stalled despite increasing platform-reported ROAS across all channels.
Top-line analysis revealed the problem: they were spending increasingly on retargeting and lookalike audiences, driving up attributed revenue but not acquiring genuinely new customers.
We restructured their approach to paid social:
- Reduced retargeting spend by 50%
- Shifted budget to prospecting campaigns with higher CPAs in the ad account, but lower customer overlap
- Implemented proper audience exclusions
Platform metrics initially looked worse, but within six weeks, new customer acquisition increased 73% and overall revenue growth resumed.
Next Steps: Building Your Attribution Framework
The companies that scale aren’t the ones with perfect attribution, they’re the ones who have a process to learn of what drives profitable growth for their business.
Attribution will always be imperfect. Customer journeys are messy. Platforms will continue to make themselves look as good as possible. But armed with the right framework and proper metric hierarchy, these challenges can be overcome.
Want to dive deeper into building an attribution framework for your brand?
We’ve helped over 100 ecommerce brands implement these frameworks and consistently see 25-40% improvements in media efficiency within 90 days. Get in touch to discuss your specific attribution challenges and opportunities.
