Marketing
The Truth About Attribution and Why It Is Never Perfect
Every attribution model is wrong. Some are useful. Here is how to pick one without deceiving yourself.
Every model encodes a bias
First-touch attribution flatters awareness channels. Last-touch flatters closing channels. Linear pretends every touch mattered equally, which is convenient and false. There is no neutral model, because attribution is not measurement; it is a story you choose to tell about a purchase decision made inside a human head you cannot see into.
Once you accept that every model is a lens rather than a truth, the question changes from which model is correct to which lens helps us make better decisions. That is a question you can actually answer.
Use attribution to allocate, not to judge
Attribution works best as a directional guide for where to spend the next dollar, not as a scoreboard for rewarding or punishing teams. The moment a channel owner's bonus depends on the model, they will optimize for the model rather than for real revenue, and the numbers stop meaning anything.
Treat the output as a hypothesis. If the model says a channel is underperforming, test the claim by pulling spend and watching what happens to pipeline. The market is the only attribution model that cannot be gamed.
Talk to customers to fill the gaps
The most underrated attribution tool is a conversation. Ask new customers how they first heard of you and what finally convinced them. Their answers rarely match your model, and the mismatch is exactly the information you were missing.
Self-reported attribution has its own biases, but combined with your quantitative model it triangulates toward the truth better than either source alone. The goal is not certainty; it is being less wrong than you were last quarter.