Multi-touch attribution

Attribution answers the question every channel report dodges: which channels actually caused your conversions, when a customer touched several before converting.

The models

Prooflytics computes seven attribution models over your journeys, side by side:

• First touch — all credit to the channel that started the journey.
• Last touch — all credit to the last channel before conversion.
• Linear — equal split across every touchpoint.
• Position-based (U-shape) — 40% first, 40% last, 20% across the middle.
• Time decay — recent touches weigh more.
• Shapley — each channel's average marginal contribution across journey orderings.
• Markov chain — how much conversion probability drops if a channel is removed.

The two data-driven models (Shapley, Markov) need roughly 100+ conversions in the window to stabilise — with less data, read them as directional.

Why the models disagree — and why that is the point

Each model encodes a different belief about how credit should flow. When they agree on a channel, that finding is robust; when they diverge, the divergence itself tells you where your reporting assumptions were doing the heavy lifting. The AI narrative on the page reads the spread for you and says what is safe to conclude.

What it needs to work

Journeys are built from your connected data — the Pixel and your CRM are what tie ad clicks to eventual deals. The more of the journey your sources cover, the more the models can see. See Connecting analytics and CRM and The Prooflytics Pixel.

Availability

Multi-touch attribution is included from the Scale plan — see the pricing page for the current tier map.