Marketing

Attribution and Its Limits

Every marketer wants to know which channel gets credit for a sale. The uncomfortable truth: attribution is genuinely hard, often wrong, and easy to fool yourself with. Understanding its limits protects you from confidently cutting the channels that are actually working.

  • Advanced
  • 13 min total
  • 13 chapters

What decision this helps you make: How much to trust your attribution data, and which robust, hard-to-fool measures (blended metrics, holdout tests, "how did you hear about us") to lean on instead of over-precise per-channel numbers.

What this topic is

Attribution is assigning credit for a conversion to the marketing touchpoints that led to it. This lesson is about how attribution models work, why they systematically mislead, and what to trust instead.

Why it matters

Buyers touch many channels before converting, so crediting the sale to any single one is inherently ambiguous, and different models give opposite answers. Trusting attribution blindly leads to cutting channels that are actually driving demand and over-crediting the ones that merely close it.

Who should learn it

Anyone running more than one marketing channel and trying to decide where the budget really works, especially anyone tempted to cut a channel because "it doesn't show conversions."

What you will understand

  • Why attributing a sale to one channel is inherently ambiguous
  • How different attribution models give opposite answers from the same data
  • The systematic biases (last-click over-credits closers; upper-funnel is under-credited)
  • What to trust instead: blended metrics, holdout tests, and just asking customers

Prerequisites

Common misconception

"My analytics tell me exactly which channel drove each sale, so I know where my marketing works." Attribution is far less certain than the dashboards imply. A buyer might see your video ad, read a blog post, get a friend's recommendation, then finally click a Google search for your name and buy. Last-click attribution credits 100% to Google search, making it look like the hero and everything upstream look worthless. The precise-looking numbers hide enormous ambiguity and systematic bias.