Attribution Windows Explained
Attribution is a framework for assigning credit to marketing interactions associated with an outcome. The model is useful only when its assumptions are visible.
Tracking Comes First
Tracking concerns what interactions and conversions you can observe. Attribution concerns how credit is assigned among the interactions you observed. Improving collection can improve the inputs without deciding which model is appropriate.
Questions Every Attribution View Should Answer
- What outcome receives credit?
- Which interactions are eligible?
- How far back does the window look?
- How is credit divided?
- How are cross-device and offline events handled?
- What decision will this report inform?
Use More Than One Lens When Necessary
No attribution view eliminates every assumption. Reconcile reporting with business outcomes, document model changes, and use experiments or aggregate measurement when the decision requires causal evidence rather than journey credit.
Common Attribution Mistakes
- Adding Platform Totals Together: Multiple platforms can claim the same conversion under their own rules.
- Changing Models Mid-Comparison: A first-touch report and a last-touch report answer different questions.
- Ignoring The Window: A seven-day window and a ninety-day window can produce very different credit patterns.
- Confusing Credit With Causality: An attributed touchpoint is not automatically an incremental touchpoint.
- Skipping Revenue Reconciliation: Attribution reporting should be checked against the system where the business records actual outcomes.
How To Use This In Practice
Choose the decision first: creative optimization, channel budgeting, funnel diagnosis, forecasting, or incrementality. Then choose the measurement lens that can reasonably inform that decision. Preserve model definitions over time so changes in performance are not actually changes in methodology.
Worked Example
Imagine a buyer first discovers a brand through a paid social ad, returns through branded search several days later, joins an email list, and finally buys from an email link. First-touch attribution gives the discovery ad the credit; last-touch gives the email interaction the credit; a multi-touch model may divide credit across several interactions. None of those views alone proves which touchpoint caused the purchase.
The example shows why model choice should follow the business question. Acquisition teams may care about discovery, lifecycle teams may care about the final conversion assist, and finance may care more about incremental revenue at the channel level. One journey can therefore support several legitimate reports as long as their definitions are explicit.
Measurement Hygiene
Keep a short data dictionary for important metrics: conversion event, revenue field, attribution window, time zone, currency, refund treatment, and model. When teams disagree about a number, check the definition before debugging the technology. Many apparent tracking problems are actually differences in reporting rules.
Editorial Standard
Product capabilities, integrations, pricing, and platform policies can change. We distinguish hands-on observations from time-sensitive specifications and recheck changing facts against current primary sources before relying on them for a purchase or implementation decision.