Leveraging IBM Watsonx for RAG-based Conversational Search in CPQ for a National Wireless Provider

Client Overview
A leading national wireless provider with millions of subscribers faced challenges in managing the complexity of its Configure, Price, Quote (CPQ) system for enterprise customers. With a vast catalog of plans, devices, and add-ons, sales representatives struggled to quickly retrieve relevant information and generate accurate quotes.

Challenges

1. Complex Product Configurations: Thousands of plan combinations, device options, and contract terms made it difficult for sales teams to find the right offering.
2. Inefficient Search & Knowledge Retrieval: Sales teams relied on static documents and outdated knowledge bases, leading to delays in responding to customer inquiries.
3. Inconsistent Quoting Process: Manual processes increased the risk of quoting errors and non-compliant configurations.
4. Scalability Issues: As the product catalog expanded, the existing search and recommendation system became less effective.

Solution: Implementing a RAG-based Conversational Search with IBM Watsonx
The wireless provider deployed a Retrieval-Augmented Generation (RAG) solution using IBM Watsonx to enhance the CPQ system. The solution included:

• Data Ingestion & Indexing: Structured and unstructured data from product catalogs, pricing databases, and regulatory documents were indexed using IBM Watson Discovery.
• Conversational AI Integration: Watsonx Assistant was integrated to provide natural language interactions for sales teams.
• RAG-based Retrieval Mechanism: Instead of relying solely on a pre-trained model, real-time retrieval of the most relevant information was incorporated into responses.
• Dynamic Quote Generation: Sales reps could query the system in natural language (e.g., “What’s the best enterprise 5G plan with unlimited data?”) and receive personalized recommendations with pricing details.

Implementation Details

• Platform: IBM Watsonx AI and Watson Discovery for retrieval, Watsonx Assistant for conversational AI.
• Technology Stack: Watsonx Knowledge Studio for domain-specific NLP training, IBM Cloud for hosting, and API integrations with the CPQ system.
• Integration: Embedded directly into the sales portal and CRM system for seamless user experience.

Results

1. Faster Sales Cycles: Reduced average quote generation time from hours to minutes.
2. Improved Search Accuracy: Sales reps found relevant product and pricing information 90% faster.
3. Enhanced User Experience: AI-driven conversational search improved usability and adoption.
4. Increased Sales Efficiency: Enabled sales teams to handle more inquiries with higher accuracy, leading to improved conversion rates.