Customer State vs. Decision Trees in marketing automation

Marketing automation solutions allow the marketer to build scenarios that can be simple or complex. For the marketer, the true test of a marketing automation solution is the ease of use when building complex scenarios: how sophisticated can the scenario be before it gets too complicated to understand, to manage and to modify on an on-going basis.

A principle factor contributing to the complexity of marketing automation solutions is the decision tree. Marketing automation solutions offer various graphical and script based workflow tools that are just dressed up decision trees that automate complex multistage decision problems. Below I will be comparing decision trees vs. a customer state approach.

Decision trees

Decision trees such as figure 1 are a useful to for predicting outcome and are often used to map workflows.

Decision trees are also used to do dynamic segmentation. For example all subscribers who are female and click on the Fashion1206 link in the last email sent to them. Decisions are used to determine the proper response according to a particular customer’s micro-segment in respond to an event (e.g. a contact that signed up then click on the confirmation link and then did not open the next three emails that were sent.)

Decisions trees quickly become complex. The human brain is best when tracking 7 or fewer items. Figure 1 shows a multi-level decision tree for a simple process. Imagine trying to build a sophisticated email programs, let alone a cross channel marketing program. Building a complex marketing automation programs often requires dedicated resources in order to ensure consistent rules setting at every level of the decision tree. Unfortunately dedicated resources are something that is in short-supply in most marketing departments.

Customer State

Customer state is a means to map workflow processes according to a customer’s state at the time of a customer event (e.g. signs up for a newsletter). A customer state diagram (figure 2) maps an event / action scenario similar to the decision tree above yet is much simpler to understand. Reading from left to right the scenario is as follows:

1)      Contact

1. When a contact subscribes then send them a confirmation email and move then to an Unconfirmed state

2)      Unconfirmed

1. When the contact clicks on an email ling then if they are a female then send them the Welcome email and refer and friend email else send the Welcome email only. Move the contact to a Subscriber state
2. If the contact does nothing for 3 days then if they are a female then send the Welcome email and refer and friend reminder
3. If they bounce then flag them in the CRM

3)      Subscriber

1. When the contact clicks on the link then send them a targeted promotion.
2. If they refer a friend then send them the Confirmation email promotion
3. If they bounce then flag them in the C

Figure 2

Only events that are relevant (i.e. possible or interesting) to the customer’s state need be considered when describing a scenario.  For example, the only events that need to be considered when the subject is in an Unconfirmed state in Figure 2 are whether they click on a link, do nothing for three days or bounce. Customer state uses a simple event / action paradigm such as “When a customer signs up (event) then Send a confirmation email (action)”.

Since each action/reaction set is according to the contact’s state there are a finite number of events to consider for any given customer state. Rather than building large decision trees that must describe all possible outcomes and are hard to understand, customer state permits the design simple marketing scenario, such as acquisition, conversion, retention. Contacts can be in different states within each scenario. The various simple scenarios can be combined to create a rich marketing ecosystem.

The benefit of this approach is that scenario design is simple and manageable whereas decision tree modeling is constrained by its complexity. Customer state marketing significantly streamlines cross channel marketing program design and implementation. It enables an agile marketing and removes the need for dedicated marketing resources to build the marketing automation workflow.

Conclusion

As I said in the opening of this article, the true test of a marketing automation solution is the ease of use when building complex scenarios. Customer state has the distinct advantage of being easy to understand, is simple to implement, and is significantly faster to market that decision tree-based solutions.