Case study · TalkTalk

TalkTalk’s customers were being passed between teams, and too many contacts were landing on the phone line when they could be resolved in messaging. The brand needed customers to reach the right help first time, in the channel they preferred, with agents supported to resolve issues faster.

Designed the Sales AI Agent, redesigned the Billing and Accounts journeys and ID&V (identity and verification), and designed the top-of-bot logic that understands what customers need and sends them to the right place. I worked with a strategist, a solution architect and a solution engineer at LivePerson, and with TalkTalk’s Customer Support and Engineering teams.

Discovery with TalkTalk’s support and engineering teams → mapping transfer patterns and failure points → intent-based top-of-bot logic for billing, SIM and technical faults → redesigned Billing, Accounts and ID&V journeys → Sales AI Agent → testing and optimisation

A customer calling about a payment can choose to message instead. Their intent passes from the IVR to WhatsApp, so the conversation opens already knowing why they’re there. Automated journeys resolve what they can, and generative AI supports agents with conversation summaries, on-demand recommendations and smart rewrite for everything else.

• +52 points NPS
• +37% IVR deflection to messaging
• +49% digital assist
• Agents using generative AI: conversations 7 minutes shorter, +5 points NPS and faster ramp time for new agents

“On top of our successes to date with LivePerson’s Conversational AI and platform, LLMs and generative AI hold enormous promise to help us scale seamless and personalized conversations even further.” — Bhavesh Panchal, Head of Channels and Digital Adoption, TalkTalk