A leading retail energy provider was experiencing increased customer service inquiries via call center and live chat. Typical consumer concerns had been exacerbated by the pandemic and unprecedented weather events, leading to increased volumes.
The retailer needed to find strategies to maintain customer satisfaction and quality of customer service, while managing the increasing costs of customer service operations.
The client decided to build an AI-powered Chatbot that would enable customers to self-serve their most frequent requests and minimize the need for customer service calls.
The team designed and developed a Chatbot powered by Salesforce Einstein natural language processing (NLP) to provide customers with conversational dialogs for self-service.
Through the Chatbot, customers would be able to:
- Verify identity against their account billing information
- Confirm current account balance and get details of last payment
- Make payments for bills and disconnection notices
- Set up deferred payment plans and payment arrangements, which are finalized with a live agent
- Set up follow-ups for specific customer requests that require handling by another business team
- Connect customers to live chat agent
The Chatbot was deployed for (3) retail brands with brand-specific styling but implemented as a single Chatbot deployment on the back-end. This streamlined maintenance and reporting significantly.
Based on retail energy provider’s identified KPI’s, we designed metadata and processes to track key activities and visualized them in a Salesforce Dashboard:
- All payments and transactions processed by the Chatbot
- Live agent transfers and follow-up requests
- Volume of chats by retail brand, referral website
- Volume of chat topics triggered within a chat conversation, with topics defined by the business utilizing natural language processing rules (e.g., make a payment, I want to move)
The Chatbot had an immediate positive impact in the first three months, and the usage by customers is continuing to grow.
In the first 3 months, the Chatbot:
- Served over 24K unique customers, averaging 8K monthly
- Deflected 30K calls or live chats, which included all chats that did not require an agent follow-up
- Provided support for common issues, including fielding 1300 billing inquiries
Based on an industry benchmark of $3 per customer service call, the client is saving an estimated $30K/month from call deflection.