Evaluating advertising channels comes with its own set of challenges, and mistakes in these evaluations can come at a high price — literally.

This post will highlight two major pitfalls and how the intricacies of attribution models can affect your ROI calculations:

1. Overestimating a Channel’s Impact: The Incrementality Problem

It’s not uncommon to see ROI figures that overstate a channel’s effectiveness. This happens when you mistakenly attribute all results to a channel, even though some purchases or sign-ups would have occurred regardless. This issue is known as the incrementality problem.

Example: Imagine you were already planning to buy a product and happened to see an ad for it on your way to the store. Did that ad actually influence your decision? In the world of attribution, the channel often gets credit, inflating the ROI beyond its real impact.

2. Underestimating a Channel’s Value

On the flip side, in an effort to avoid overestimating, marketers sometimes undervalue a channel, which can lead to missed opportunities. In this scenario, ROI calculations fall short of reality, and promising channels might be abandoned due to miscalculations.

Real-world scenario: During growth phases, your ROI might be based on a Last Paid Click model with UTM tags. This can lead to under-crediting channels like YouTube influencer campaigns, where users may not be correctly attributed, making campaigns appear less effective than they actually are.

Channels commonly misjudged due to attribution errors:

1. Facebook, Instagram, and YouTube pre-roll ads — often suffer when using post-view attribution, as actual influence may be underestimated.

2. Remarketing — tends to have low incrementality since it targets users already familiar with your product, who might return organically.

3. Email marketing — while almost cost-free, it can sometimes overclaim credit when not analyzed properly, skewing the channel’s true contribution.

4. Branded search—often absorbs some organic traffic, taking credit for conversions that might have happened without it.

How to address these attribution challenges:

1. Use multiple attribution models for a broader perspective.

2. Conduct incrementality A/B tests—including straightforward "on/off" testing for channels.

3. Experiment with algorithmic and probabilistic attribution models to refine your analyses.

If you work with GA4 to BigQuery exports, be sure to check out my SQL cheat sheet.