Over the last 10 years, automated email workflow tools have provided a big increase in efficiency over bulk manual sends. Most of these tools look something like this.


An event triggers the inclusion of contacts in a list and then they receive some communication. Also included are some combination of time delays and conditional logic based on behavior. This can save a lot of time over the bulk, manual method. The shortcoming of this method is that the processes are often built arbitrarily and are not data-driven. There is still a lot more room for optimization in this process. Additional efficiencies can be gained by getting away from the low-code journey builders, and using data science to determine your email process.

Data Overview

The are two primary metrics needed to make data-driven decisions about email. These values are:

  • Contact Value - What is the projected dollar value of this contact?
  • Topic Preference - What topics or content is this contact interested in?

The contact value determines who you email, and how you market to them. The topic preference determines what message you communicate to them.

Calculating Contact Value

Contact value is a critical metric for making fast, scalable decisions about contacts. It's also the core metric behind our email automation strategy. This article provides a detailed explanation of how to calculate contact value.

Actions Based on Contact Value

Once you have accurate contact values for all of the contacts in your CRM or data warehouse, you can create rules based on those values. Our contact value rules are focused on three different actions we can take. These actions are content emails, offer emails, and triggering an opportunity for a sales rep. We then set contact value ranges for when these different actions are taken. This is illustrated in the example below.

Low Contact Value

  • $0.01 to $2.00 in contact value 
  • Send informational content emails to increase engagement

Medium Contact Value

  • $2.01 to $5.00 in contact value
  • Send offer emails with the goal of generating a lead

High Contact Value

  • Contact value of $5.01 or higher
  • Trigger creation of an opportunity for a sales rep, and schedule a sales call

Topic Preference and Global Taxonomy

Your users are interacting with hundreds of different web pages and emails. The data about those engagements is used to calculate the topic preference. The first step in measuring topic preference is to categorize your content in a standardized way. We accomplish this using Drupal's taxonomy module. We create a list of topics and categorize every page on the website. Don't use the categories for your blog posts. That taxonomy has a specific, user-facing purpose. Intertwining your blog taxonomy with a global one creates problems. 

This same taxonomy also needs to be applied to email, SMS, and app content. Most email and SMS software have fields for categories or tags where you can apply this.

Once you have your content properly categorized, you can apply a score to each pageview and email click. Let's say this is one point. You then need to sum up the points by user and by topic. You'll end up with something that looks like this:

Category Score
Line of Credit 4
Asset Based Finance 2
Insurance 0
Commercial Loan 0
Business Credit 8
Merchant Services 0
Technology 3

Since you've already categorized your email based on your global taxonomy, determining which email to send to the contact above should be easy.

Email Marketing Segments

Once you've calculated your contact value and scored your contacts based on topic preference, it's time to send some emails. We build a segment for each different piece of content or offer we plan to send. For example, if you have a contact with a preference for commercial loan content and a contact value of $0.50, they should receive emails with informational commercial loan content.

Once a contact gets added to that segment they receive the email associated with that segment.

Operational Workflow

For us, the operational workflow looks something like this. 

email data workflow