Support organizations sit on a gold mine of customer information. The Predictive Customer Engagement initiative aims to improve proactive value delivery to customers based on what we already know.

The Predictive Customer Engagement model is a double loop model consisting of the Event Loop and the Improve Loop.

The Event Loop includes:

• Accessing data sources
• Listening at scale
• Mining for actionable information
• Creating that action
• Communicating that action
• Accessing data sources to assess the impact

---

The Improve Loop includes assessing the effectiveness and quality of each part of the Event Loop.

Ongoing Exploration

The Consortium’s work on Predictive Customer Engagement is continuing to explore the following topics:  

• Our ability to leverage big data analytics, emerging digital automation
• Our relationship with the customer – do they trust us?
• How do we get customers to take action on recommendations? How do we to provide compelling evidence of relevance, consequences of not taking action and information about effort involved?

Data Requirements

1. We have to know a lot about our products/services and how they are used
2. We have to know a lot about customers as companies and as individuals

Machine Learning: A Path to Contextual Knowledge

Customer Service Landscape

• 70%-80% Known: Most inquiries handled by people have already been solved
• Self-Service: Designed using inside-out thinking
• People are solvers: People are creative thinkers and should work on new
• Knowledge-Centered Service: Dynamic content created & updated in the workflow

Artificial Intelligence is essential for digital transformations, freeing up employees from mundane and repetitive tasks so they can pursue creative ones.

Machine Learning Techniques: Solving Problems

• Visualizations: Consumable Outputs (reports, dashboards, augmented reality, chat, machine to machine)
• Capabilities: Analysis abilities (patterns, predictions, recommendations, optimization)
• Machine Learning Technology: Classification, Regression, Clustering, Anomaly detection, etc., and the associated algorithms resulting in a trained model
• Data Repository: Storage entity (Data Lake, Data Warehouse, or other model for storing data to be mined)