1.3 Three Characteristics of Marketing Data
The hard part of Marketing Science is not the complexity of the methods. It is the nature of the data itself.
Characteristic 1: What’s measurable isn’t all that’s causal
Before a customer makes a purchase, they are influenced by countless touchpoints — your ads and website, yes, but also friends, competitors, the economy, the weather, even their mood. Most of this is invisible to your data.
What gets recorded is usually the easy-to-measure stuff: clicks, impressions, page views. The real drivers of a purchase, what statisticians call confounders, often are not captured as data at all. This creates a gap: the variables that are easiest to measure are not always the ones that matter most.
Reality is messier than any dataset suggests. You need domain knowledge to fill in the gaps, and causal logic to untangle what actually drove what. That is where human judgment matters most.
Characteristic 2: Analysis runs on business time
Marketing data is not built for research. It is a byproduct of daily business operations that have to keep moving. Campaigns launch, budgets shift, tags break, attribution windows change, and platform definitions get updated while the business keeps running.
In academic research, you can wait until the data is enough and clean. In business, decisions cannot wait. An answer that arrives after the budget meeting is not useful.
Characteristic 3: Drivers rarely move independently
In a controlled experiment, you change one variable at a time and hold everything else constant. In real-world marketing, that almost never happens. The variables that drive sales tend to move together because something else is moving them all.
Consider a new product launch. TV ads, paid social, retail media, shelf-space expansion, and a price cut on the old product all happen at once because management planned them for the same launch window.
It is multicollinearity. The first step toward honest analysis is to acknowledge this constraint and be clear about what you can and cannot say.