Credit where the journey says it belongs.

Connect your event tables, say which rows are conversions and which are marketing touches, and EkamFlow works out the journey before each conversion and how much credit each channel earns under eight attribution models.

Compare channel credit before changing spend

Use differences between attribution methods to plan a budget experiment.

Channel credit by attribution methodIllustrative data
Last touchSearch: 60%, Paid social: 25%, Email: 15%
LinearSearch: 40%, Paid social: 35%, Email: 25%
  • Search
  • Paid social
  • Email

What you can do

Set up from your own data

Four steps — conversions, touches, channels and journeys — each proposed from a profile of your tables, with every column and filter shown before anything runs.

Messy channel values, resolved

A channel resolver maps the values your tables actually hold onto the channels you want to report on, and shows you the raw values it found before you agree to the mapping.

Eight models, side by side

First touch, last touch, last non-direct, linear, position-based, time decay, Markov and Shapley. Rather than one model you have to defend, you get the spread, and the Markov and Shapley shares carry a stated range rather than a single confident number. Where models agree the finding is robust; where they disagree, the credit depends on the rule you picked.

Per market

Results break down by market, with a market that has too few conversions withheld rather than reported from too little evidence.

Tables, not a slide

Every run writes its results into the workflow's database. Query them, join them to your own tables, or ask AI Studio about them. No table holds each individual touch or how identities were linked.

It runs as a pipeline

Attribution is a lane in your pipelines with its own run log, results history and job record — rerun it on a schedule like anything else.

Setup and requirements

Steps

  1. Point at the tables holding conversions and marketing touches.
  2. Agree the channel mapping from the raw values found in your data.
  3. Finalize the setup and run the attribution pipeline.
  4. Read the credit per channel and per market, and query the tables.

What you need

  • Event tables with marketing touches, each with a customer key and a time
  • A conversion event with a customer key and a time
  • Source-data storage on

Attribution from an observed log is a reading of your history, not a controlled experiment. The eight models exist because no single rule for splitting credit is correct.

Build an Agentic Worker for your team.

Tell us which workflow you want to hand off.

Book a demo