Customer Decision Hub predictions

This video provides an overview of the Customer Decision Hub predictions in the Prediction Studio portal. Prediction Studio is the workspace for data scientists, offering tools to manage predictions used by Customer Decision Hub for optimizing customer interactions.

Decision strategies utilize predictions to determine the optimal action and treatment for customers across all interaction channels. Customer Decision Hub includes predefined predictions for each relevant communication channel. The predefined Predict Inbound CallCenter Propensity prediction calculates the likelihood of a customer accepting a proposition offered by a Customer Service Representative when contacting the call center. The Outbound Email Propensity prediction calculates the likelihood that a customer clicks a link in an email.

The prediction that calculates the propensity of a customer clicking on a web banner is named Predict Web Propensity. For instance, in a cross-sell scenario, U+Bank leverages its website as a marketing tool. Upon customer login, the website presents a personalized credit card offer in a banner.

If a customer is eligible for multiple credit cards, the Predict Web Propensity prediction calculates the propensity of receiving a positive response from the customer for each of these cards. Customer Decision Hub determines which credit card to offer based on business rules, interaction context, and propensity.

The context of the Predict Web Propensity prediction includes the treatment that a customer receives, in this case, a personalized banner in the web channel. This allows the capture of variations in customer preferences, such as the background color of a credit card offer in a web banner.

The prediction's response labels are "Clicked" for the target behavior and "NoResponse" for the alternative behavior. The NoResponse outcome is recorded after the response timeout expires.

Predictive models drive predictions.

The propensity predictions integrated into Customer Decision Hub utilize adaptive models that learn from customer responses and receive automatic updates every hour. This ensures that U+Bank automatically responds to changes in customer behavior while the adaptive models self-optimize.

Customer Decision Hub makes customer-level decisions, while account-level decisions are made for a specific account associated with the customer. For example, a customer-level decision might be to send a promotional email to a customer based on the overall engagement with the company, while an account-level decision might be to send a reminder email to a customer based on their account balance.

The Predict Web Propensity prediction propagates the outcomes of a model based on the Web Click Through Rate adaptive model configuration. The outcomes of the adaptive model configurations align with the response labels of the prediction.

Business requirements might necessitate the use of a predictive model trained on historical data or a highly transparent scorecard model. These models can be incorporated as supporting models within the prediction, with their outcomes serving as predictors for the adaptive models.

You have reached the end of this video. What did it demonstrate?

• The predictions that come with Customer Decision Hub in the Prediction Studio portal.
• The Predict Web Propensity prediction utilizes the Web Click Through Rate adaptive model configuration to determine the outcomes.
• The outcomes of supporting predictive models can function as predictors for the adaptive models.