Marketing Attribution: Ending the ROAS Guesswork

---

The Monday morning meeting about ROAS

It’s the weekly marketing meeting. The Google Ads manager shows a deck with a 4.1x ROAS. Your Meta agency is next, reporting a 3.8x ROAS. Then TikTok shows a promising 3.2x on a new campaign. Everyone feels they've had a productive week.

Then the CFO asks the question you knew was coming: "If all the channels are doing this well, why did our overall revenue only go up by 2x?"

Things get a bit awkward. This isn't really a problem with your team, it's a problem with the system. You're working with a setup that's no longer fit for purpose, not leading an incompetent team.

In my experience, this is a very common situation for most high-growth direct-to-consumer brands. You've invested in a modern data stack, perhaps with Snowflake, dbt, and Looker, but you've just moved the same confusion into a new, faster system. Automating a confusing process just means you get confusing reports more quickly. The problem isn't the tools you're using, it's the underlying logic.

Why the ad platforms will never agree

To be blunt, Facebook, Google, and TikTok aren't neutral observers. They have a commercial interest in taking credit for any conversion they had a hand in. Each uses a different attribution window, a different method (view-through, click-through), and has very little reason to give credit to a competitor.

Expecting them to agree is a bit like asking three rival football managers to agree on who scored the winning goal. It's just not going to happen.

The reality is, the problem isn't your dashboard tool. It's that your business logic is spread across different ad platforms instead of being managed in one place. The real work isn't about hiring more analysts, it's about getting everyone in the business to agree on what a conversion is and how you assign value to it. Without a solid Data Governance strategy for your marketing data, you're just pulling that confusion into your own systems.

I see this all the time with scale-ups, particularly around the Series B to D stage. They've scaled their marketing channels much faster than their data setup. The result is a collection of conflicting reports that makes it hard for anyone to trust the numbers and makes proper Data-driven Marketing almost impossible.

How to build your own source of truth

To get a clearer picture, you need to stop relying on the platforms that take up your budget to tell you how they're performing. You need to build your own version of the truth. This isn't about buying another tool, it's about changing how the data is put together.

Here's the practical, three-step approach I've found works well:

Centralise the raw data: We stop using the summarised, self-reported numbers from the platform dashboards. Instead, we pull the raw, detailed data on spend, impressions, clicks, and conversions from every channel into your central data warehouse. This way, you can see exactly what's happening.

Build a unified model: This is the heart of the work. Using a tool like dbt, we bring these different sources together into a single user journey, from first touch to final purchase. This is where you define the rules. You decide on the attribution model (linear, time-decay, or u-shaped, for example) and the lookback window. These rules become the Semantic Layer for all your marketing performance.

Visualise the result: Only when that logic is in one place do we connect a BI tool. The dashboard then shows a single, blended ROAS and a clear view of Channel Performance that everyone, from marketing to finance, can trust. The weekly debate usually ends because the source of truth is your own, not someone else's.

Getting everyone to agree on the new numbers

I should be honest, this can be a bit uncomfortable in the short term. It means taking away the familiar, often inflated numbers your channel managers are used to. You might find they're resistant, because their performance has always been measured by those platform numbers. The figures will probably look a bit worse before they become more accurate.

Getting this done is more about getting people to agree than it is about the technology. You have to get everyone on the same page about definitions and methods. It might feel like you're slowing down for a few weeks, but it helps you move much faster for years to come. It's the kind of foundational work that often gets put off, which is why the problem is so common.

But what you get in the end is a clear picture and proper control. You can stop guessing and start making more informed decisions about your budget. You can finally have a sensible conversation with the CFO about scaling because you're both looking at the same, trusted numbers. You can move on from arguing about whose numbers are right, and start discussing the right thing to do next. That's where the real value is.