Table of Contents

Removal effect with a Markov chain model

Nolan Garrido Updated by Nolan Garrido

Removal Effect with Markov Chain Model

This article assumes you are familiar with the Markov chain model as it relates to attribution. If you'd like a brief overview, check out our knowledge base article on the topic.

In order to determine the contribution of a channel in a customer’s journey from first touch to last touch, you can leverage the principle of removal effect. Removal effect principle demonstrates that the contribution of each channel in the customer journey is determined by removing each channel and seeing how many conversions occur without that channel being in place. In other words, it answers these question:

  • How much does a given channel affect the probability of conversion?
  • Which touch points are the most important in the customer journey?

Removal affect can take into account either conversion or revenue, depending on the outcome you want to track.

Here's an example removal effect report:

Sample CaliberMind Removal Effect Report

In the above graph, the removal effect of the "Visit Monthly Newsletter Email" event is 7% - meaning, if the company were to stop sending this email, they can expect to have a 7% reduction in conversions. This is powerful, actionable marketing intelligence that other attribution models cannot provide.

There are limitations, of course. For example, the model groups inaccurate or generically labeled events as "Event" (the first bar on the left). These events have a 58% removal effect, but cannot provide actionable insight because of the way the events are labeled. We would recommend that this company correctly classify these events a bit more granularly. As there are enough events in each classification that they are statistically significant, this report will provide even more insight.

To see how you can use the removal effect report in CaliberMind, contact your Customer Success Manager.

How did we do?

Attribution Reports Summary

Contact