When only 20-25% of customers complete satisfaction surveys — and even those are primarily negative experiences — how can you truly understand your entire customer base? In this episode of AI CX Innovators, Prasanna Chand, Head of Data & Digital Transformation at Wayfair, reveals how they’re using AI to predict customer satisfaction scores with 85% correlation to actual survey results, providing a complete picture beyond the inherently skewed feedback pool.

Prasanna takes Ashish through Wayfair’s journey implementing AI across their customer experience operations, from identifying critical issues within days of launching their loyalty program to helping agents self-coach through personalized insights rather than generic examples. With ChatGPT’s launch as the tipping point, he explains how Wayfair strategically separated which AI solutions to build versus buy, and why their partnership with Level AI has been transformative for users across the organization.

Topics Discussed:

• How Wayfair’s three-pronged approach to customer data analytics focuses on conversational insights, making business users more data-friendly without SQL knowledge, and creating an enterprise architecture that balances hyperscaler platforms with boutique vendor solutions.
• The tactical advantage of AI-powered analytics that discovered loyalty program issues within days of launch, bypassing the months-long traditional data warehouse reporting cycle and uncovering specific functional problems hindering customer adoption.
• Why AI-predicted customer satisfaction scores (achieving 85% correlation with actual surveys) solve the inherent bias problem when only 20-25% of customers complete surveys, but still don’t replace manual CSAT collection.
• Wayfair’s strategic bifurcation approach to AI implementation: building and extending homegrown systems for agent support while purchasing software for integration with third-party telephony, workforce management, and quality systems.
• How connecting journey analytics with conversation data enables FCR analysis to identify and reduce multi-contact scenarios, allowing teams to immediately see negative sentiment pathways and make targeted improvements.
• Three essential best practices for implementing AI transformation: educating stakeholders to manage resistance and expectations, selecting partners who can innovate at the market’s pace, and identifying use cases with quick ROI through plug-and-play implementations.
• The evolution from random sampling in quality assurance to holistic review capabilities, enabling personalized agent coaching with specific conversation examples rather than generic feedback, fundamentally changing how agents self-improve.
• Leveraging AI for language translation and virtual training to overcome language barriers in agent development, creating training in one language and delivering it through human-like virtual instructors in multiple languages.

Listen to more episodes: 

Apple 

Spotify 

YouTube