Warning: Use of undefined constant ICL_LANGUAGE_CODE - assumed 'ICL_LANGUAGE_CODE' (this will throw an Error in a future version of PHP) in /home/customer/www/unemyr.com/public_html/wp-content/themes/genesisChild/functions.php on line 738

Attribution Modeling With AI

When talking about lead generation and customer acquisition, it may be of interest to consider the average customer acquisition cost (CAC) and customer lifetime value (CLTV).

Knowing how much it will cost to acquire a new customer, and how much they are worth, are essential in being on top of your marketing strategy. It is also good to know how long time it takes for the average customer to pay back the acquisition cost.

In addition to monitoring your CAC and CLTV, you probably want to know what marketing activities contributed to the sale, as this helps measure the ROI of different marketing assets and initiatives. This is called attribution modeling and helps optimize the customer journey from a cost and results point of view.

The problem is, real life is seldom as easy as we might want.

The customer journey is a multi-touch one, sometimes across many different channels and mediums. One particular ad, mailshot, or blog post may not be the sole reason someone bought a product. In fact, that would be highly unlikely.

This is where attribution modeling comes in. It is an attempt to give credit to several marketing activities that all helped to acquire the customer. There are many different types of attribution models that give credit in different ways. For example: 

  • First touch attribution
  • Lead conversion touch attribution
  • Last touch attribution
  • Last non-direct touch attribution
  • Linear attribution
  • Time decay attribution
  • U-shaped attribution
  • W-shaped attribution
  • Full path attribution

One of the companies performing attribution modeling with machine learning is Cubed. I had a discussion with Russel McAthy, their CEO, to talk about his thoughts on this and to learn more about his company.

He refers to his product as “a marketing analytics platform that helps businesses understand the performance of the full consumer journey—multi-visit and multi-device. It can be used as a single platform of truth, a place for all marketing channels and external data to be pulled into the same place to help inform the true story of how they impact revenue and value.” 

McAthy says his company’s algorithm “means a shift from a last interaction/click world and into an attributed space. This allows smarter decision making with the ability to look at how activity truly impacts consumers as they are acquired, research and hopefully ultimately purchase.”

Cubed uses a linear regression model to look at the conversions. The algorithm goes through every single touch point that has happened prior to a successful conversion and is trained on the most impactful elements.

A number of key factors on website pages are taken into consideration such as events that are triggered by the user as they engage marketing activities, including keywords, ads, and ultimately impression-led activities when users view creative on external websites.

Tools like this enable brands to invest in areas that truly add value to their consumers and drive incremental business growth.

Other companies in this space are C3Metrics, ConversionLogic, VisualIQ, and Windsor (to my understanding, they aren’t related to Windsor Circle, which also does AI-based marketing products). These companies also offer attribution modeling with machine learning for optimizing the cost efficiency of the customer journey.

Magnus Unemyr

Author, speaker and consultant in the aras of marketing automation, artificial intelligence, and the Internet-Of-Things. Contact me if you need help! Learn more.