Traditional digital marketing measurement heavily relies on attribution, a methodology used by ad channels like Meta, Google, and TikTok, as well as tools like Google Analytics, to observe marketing touchpoints and assign credit for purchases.

In the Incrementality Academy, we've been making the case that attribution isn't enough—brands need measurement.

In this article, we dive into what brands can do to wean themselves off their reliance on attribution, and on to methodologies that better reflect the true relationship between their ad dollars and sales.

Understanding Causality Vs. Correlation

As we've shown in the previous chapter of the Incrementality Academy, the fundamental limitation of attribution is that it relies on correlation, not causation. Just because a user clicked on an ad and then purchased doesn't definitively mean the ad caused the purchase. The purchase might have occurred organically or due to other factors.

Worse still, the purchase might've happened anyway (don't tell finance about this waste in ad spend!) and the ad was simply the easiest path to purchase. Understanding this key difference is Step 1 in weaning your brand off click-based attribution.

Understanding click-based attribution was the best we could do (back then)

Click-based attribution models like Multi-Touch Attribution (MTA) emerged as a more sophisticated way to understand the customer journey compared to single-touch models (like last-click or first-click), or simply ... guessing.

Click-based attribution was valuable in a time where marketers had limited data or control over where ads were being shown, seen, and measured, but times have changed.

More recently, platforms like Google Analytics have turned to Data-Driven Attribution (DDA), a more advanced, signal-enhanced MTA model that utilizes machine learning algorithms to evaluate all previous user paths and assign credit for conversions based on how people engage across various touchpoints.

But the challenge remains: trying to measure every touchpoint is not good enough in terms of methodology—you're bound to miss out on some. Still, for a time where marketers were data-starved, click-based attribution did a pretty good job.

Run simple experiments—you don't have to throw everything out at once

This doesn't mean you need to toss everything you know out the window. Think of it like steps on a ladder:

In the absence of any form of measurement, last-click attribution is better than nothing.

Brands that can, should layer on more signal in the form of data-driven attribution to get themselves better clarity. And at the top of the ladder, brands that regularly use incrementality tests don't have to toss out DDA — companies like WorkMagic leverage the results of incrementality tests to calibrate data-driven attribution models for brands.

This model can then be applied across all dashboards and levels of granularity, from business-level to ad-level, making optimization for incrementality simpler.

The goal? To move up the ladder, layering on granularity, accuracy, and predictivity through DDA, incrementality testing, and incrementality-calibrated Media Mix Modeling (MMM).

Here's why you should start with DDA

Platform-reported attribution can significantly overstate the impact of marketing channels. Here are just a few examples we've seen while working with dozens of eCommerce brands:

A WorkMagic incrementality test revealed that Meta Ads' true incremental impact was 12% of sales, while Meta's reporting claimed 25% for the same period—an overstatement of 108%.

In another test with Moonbrew, we learned that YouTube's impact was being underreported by 4x—typical of upper-funnel channels that drive awareness but struggle to track conversions.

Enhancing the click-based attribution with the signal and machine learning algorithms used in DDA allows brands to make better, data-driven decisions, without significant investment or changes to their measurement approach.

Next, layer in incrementality testing

By calibrating DDA with incrementality test results, brands gain a more accurate understanding of their Return on Ad Spend (ROAS) and Profit on Ad Spend (POAS) for each channel.

Brands using incrementality-based attribution have seen significant improvements in measurement accuracy. Switching from rule-based attribution (like last-click) to incrementality-based attribution has been shown to reduce the attribution margin of error from as high as 40% to as low as 1–5%.Incrementality testing can also uncover the "halo effect", where marketing on one channel drives conversions on other channels (e.g., TikTok ads leading to purchases on Amazon).

Traditional click-based attribution often fails to capture this cross-channel impact.

WorkMagic's incrementality testing can measure this halo effect, providing a more complete picture of marketing effectiveness.

And a sprinkle of iMMM

Finally, brands should factor incrementality-calibrated MMM into the mix, to get the ability to predict outcomes like the right amount of ad spend to achieve maximum incremental ROAS, or even maximum sales.

The ability to model and predict these outcomes with a high degree of accuracy is unlocked only through the combination of incrementality testing and MMM, two highly-regarded methodologies that gain added utility and effectiveness when used in tandem.

Attribution got us far, but it’s no longer enough. Today’s marketers need more than just correlation—they need causality, calibration, and confidence.

Data-Driven Attribution is a great first step, offering better signal and smarter credit assignment than traditional models. But the real breakthroughs happen when you layer in incrementality testing and iMMM, giving you both a clear picture of what’s working and the foresight to optimize future spend.

Ready to take that next step towards moving off click-based attribution? Book a demo with us.