What is marketing attribution?

A plain answer to what marketing attribution is, the main models, and why most attribution reports disagree with each other.

By T.J. Muro

Published

Updated

Short answer: Marketing attribution is the practice of assigning credit for a conversion to the marketing touchpoints that came before it. It answers one question — which spend produced the outcome — and every attribution model is just a different rule for splitting that credit.

Why attribution exists

Buyers rarely convert on one touch. They see a paid social ad, ignore it, search the brand two weeks later, click an ad, come back through email, and buy. Attribution is the accounting layer that decides which of those moments gets paid for the sale.

Without it, every channel claims the same conversion, and reported return on spend adds up to more revenue than the business actually booked.

The main attribution models

  • Last click — all credit to the final touch. Simple, and it systematically overpays search and retargeting.
  • First click — all credit to the first touch. Overpays awareness, undervalues the closing work.
  • Linear — credit split evenly across every touch. Fair-looking, rarely true.
  • Time decay — more credit the closer a touch sits to the conversion.
  • Position based (U-shaped) — heavy weight on first and last, remainder spread across the middle.
  • Data driven — weights derived from observed conversion paths rather than a fixed rule. Only as good as the data feeding it.
  • Incrementality testing — holdouts and geo tests that measure lift instead of assigning credit. The most honest read, the slowest to produce.

Why two attribution reports never agree

Each ad platform attributes conversions to itself using its own window, its own view-through rules, and its own identity graph. Your analytics tool uses a third set. Nobody is lying — they are answering different questions with different data.

The common causes of disagreement:

  • Different lookback windows (7-day click versus 28-day click plus 1-day view)
  • View-through conversions counted by the platform and invisible to analytics
  • Cross-device journeys stitched by one system and broken by another
  • Consent choices and tracking prevention removing touches from the path
  • Missing or overwritten campaign parameters on inbound traffic
  • Bot and automated traffic entering the path as if it were a person

The challenges of marketing attribution

The models are the easy part. The hard part is the input. Attribution assumes every touch in the path was a real person taking a real action. When part of your click volume is automated, the model still runs, still produces a number, and still routes budget toward whichever channel produced the most recorded touches.

That is the failure mode worth planning for: attribution does not tell you the traffic was real. It tells you how to divide credit among whatever was recorded.

What to do about it

  1. Pick one model as the internal standard and hold every channel to it. Consistency beats accuracy for week-to-week decisions.
  2. Keep the platform numbers as directional, not as truth.
  3. Validate the traffic feeding the model before you tune the model.
  4. Use incrementality tests for budget-level decisions and attribution for optimization-level decisions.
  5. Reconcile against booked revenue monthly, not quarterly.

Related reading