Multi-touch attribution: when it works, when it doesn't
Every agency's favorite buzzword. In most cases it's an illusion you're paying for. When does it actually drive a decision?
By Burak Özkan · Paid Media Lead
Multi-touch attribution has been the most-loved marketing term of the last three years. It sounds great in meetings; it looks great on invoices. But more than half the implementations in our portfolio never turn an MTA 'insight' into action.
Let's be clear: MTA produces decisions when three conditions hold at once. (1) You have volume — 100K+ monthly conversions. (2) At least four channels each take a meaningful share. (3) Your pre-cookieless attribution stack already works.
Without all three, an MTA model doesn't change what last-click or position-based already tells you. No matter how mathematical Markov or Shapley get, without volume they don't produce statistical significance.
Practical advice: for early-stage brands, GA4's data-driven attribution is enough. Once you cross 100K monthly conversions, we set up custom MTA with a data scientist on the team. Before that, your budget has somewhere better to go.