Architect: Next Best Channel

There are countless user personas and behaviors, and when it comes to channels, it is not easy to know which user is interacting with which channel. Creating interactions with users through the right channels helps you to reach your goal in a highly-efficient way. Next Best Channel automates your decisions about users by leveraging strong data analytics functions.

Fundamentally, Next Best Channel decides on the best channel option for each user and leads them to the specified channel. ‘Best’ relates to the highest possibility of engagement among the specified channels.

You can select the following channels for Next Best Channel:

• Web Push
• App Push
• Email
• SMS
• WhatsApp

How does it work?

Next Best Channel uses machine learning (ML) that calculates the probability of user engagement for each messaging channel, which is the same algorithm that is used for Likelihood to Engage predictive segments of Insider. 

Each user’s engagement probability for each channel is predicted using ML models that consider hundreds of different signals from users on your website, mobile app, and messaging channels over the last 6 months including interactions with messages sent by both single-channel campaigns and Architect Journeys. These models are updated regularly by Insider’s AutoML platform Delphi. The engagement probability model adjusts itself based on the changing behaviors of customers and seasonal conditions. The number and importance of the input signals vary among channels and change with time to fit recent user behavior better. Below you can see some examples:

• The first time that a user receives/opens a message,
• The average time it takes a user to receive/open message,
• The number of times that a user receives/opens a message,
• User’s click-through rate for a channel,
• Total number of the user’s sessions on website and mobile app,
• Total number of the user’s page views on website and mobile app,
• Total number of the user’s purchases on website, mobile app and CRM,
• The user’s average period/frequency of purchases,
• The user’s average session duration,
• The user’s reachability status for the channel,
• The Engagement Probability algorithm decides which channel is best for that user by selecting the channel with the highest probability of engagement. 

If a user’s scores are equally good for multiple channels, Next Best Channel will randomly decide which channel to use, or use the Fallback Channel if specified by the customer.

Advantages of the Engagement Probability Algorithm

• Uses all messaging data including all message interactions from single-channel campaigns and Architect journeys.
• Considers Reachability as an additional parameter in its decision-making.
• Specializes in messaging channels such as Web Push, App Push, Email, SMS and WhatsApp. Each messaging channel has its algorithm based on the nature of its interaction types.
• Adjusts itself to changing conditions such as customers' campaign usage, and changing user-base to normalize the data. 

Channel scores calculated by Engagement Probability Algorithm are updated on a daily basis.

Engagement Probability algorithm is trained for 6 months of past user data.

Use Cases

With Next Best Channel in your journeys, you can:

• Engage users through the most appropriate channel to lead them towards your goal.
• Evaluate which channel is generally the best for your user base.