38. Attribution vs. Causation
Estimated reading time: 1 minute
Is the marketing concept of Attribution (giving credit to a channel or activity for conversion) the same or different from the statistical concept of Causation (a direct cause-and-effect relationship)?
Joe Domaleski’s article on the role of attribution, correlation, and causation in everyday life inspired me to think about it this week.
An action can get ‘credit’ for an outcome without directly causing it.
For example, say a company runs a social media ad campaign for a new product. A user sees the ad on Facebook but doesn’t click on it. A few days later, they visit the company’s website organically (without any ad) and make a purchase. Depending on the company's attribution model, Facebook will get some credit for this outcome.
The Facebook ad didn’t directly cause the user to purchase. The purchase could have been influenced by many other factors— such as previous product usage, a friend’s recommendation, etc.
Attribution indicates association and contribution. It is a measure of correlation and doesn't imply Causation. So, you can say a particular ad is associated with conversions, but this doesn’t mean the ad caused the conversion.
How does one establish causation?
One way is randomized trials. Find two groups of people—one that saw your ad and one that did not. Compare the average conversions between groups. If the difference is statistically significant, then the ad likely caused the conversions!
Both ideas are important for different marketing analyses. Attribution helps with resource allocation across channels, while causation is necessary to determine what actions truly drive results.