What Is Attribution? Assigning Credit You Cannot Fully See

In plain English

Someone sees a video, reads two articles, searches your name and buys. Attribution decides which of those gets the credit, and every model answers differently.

No model is correct. Last-click undervalues everything early; first-click undervalues everything late; multi-touch spreads credit by rules someone invented. Private sharing and offline conversation are invisible to all of them.

What to know

Model-dependent

Each distributes credit differently.

Last-click favours capture

And defunds demand creation.

Blind spots are structural

Dark social and offline cannot be seen.

Tested, not modelled

Holdouts give real answers.

Why it matters

Budget decisions made strictly on attributed data progressively defund whatever created the demand. Adding a self-reported source question recovers more truth than any model change.

Common mistakes

×Treating one model's output as fact.

×Comparing channels across different attribution windows.

×No self-reported attribution question anywhere.

×Adding up platform-claimed conversions.

FAQs

Which attribution model should I use?

Whichever you use consistently, understanding what it undervalues.

How do I measure what attribution cannot see?

Holdout tests, and asking customers how they found you.

WRITTEN BY TARIQ SALLAM

Marketing Consultant. Entrepreneur. Content Creator.

I'm a marketing consultant, entrepreneur and content creator. I help businesses grow through practical marketing, websites, SEO, content and AI.

More About Tariq →