Narrative

Cookieless Multi-Touch Marketing Attribution Platform

Third-party cookie deprecation and reliance on naive last-click models starved top-of-funnel discovery channels and misallocated advertising spend.

ClickHouseAttribution ModelingPythonTypeScriptNext.js

← Narrative Hub

// the problem

What Was Broken

// solution

How It Was Built

Built a 4.2 KB first-party tracking beacon using daily salted rotating hashes, streaming events into ClickHouse OLAP, and formulated Markov chain removal effect algorithms in Python.

// results

What Changed

Accurately distributed conversion credit across multi-channel paths, eliminating blind spots in ad spend allocation with sub-200ms query times.

// deep dive q&a

Common Questions

Heuristic models like Last-Touch or Linear rely on arbitrary rules. Markov removal effects mathematically measure what percentage of conversions disappear if a specific channel is removed from customer journeys.