Vodafone reinvents customer service using gen AI

Making customer service cool

VOXI, the UK’s mobile network for Gen Z, knows that keeping pace with the ever-evolving language of this generation is vital to staying relevant. It’s a challenge facing brands across the world, and many are turning to technology to seamlessly combine the relevant language with a professional tone for customer interactions. At Accenture, we believe that generative AI (gen AI) and large language models (LLMs) can help by delivering more human-like conversations for a new frontier of customer service.

VOXI mobile —a sub-brand of multinational technology company Vodafone targeting people typically aged 16-29 - has built its brand reputation and customer base on a strong differentiated platform, designed for Gen Z. As standard, social media data comes unlimited on all VOXI packages, as these represent the platforms that their customers spend the most time on. Now, it’s using gen AI and LLMs to power a first of its kind large language model (LLM) AI chatbot in the UK, aimed at making customer service more engaging, useful and fun. It’s the first customer facing gen AI chatbot in the UK telecommunications industry and sets a benchmark for customer experience.

Accenture helped VOXI mobile create the chatbot, using our expertise in AI and customer service innovation to guide development. In the first 6 months since its launch, the chatbot has:

• Boosted customer experience with simpler and faster interactions.
• Increased containment rates (percentage of queries resolved by the chatbot without a human agent) meaning fewer customers need to be transferred to agents.
• Reduced average handling time for customers who need agent assistance.

Better brand representation

VOXI mobile’s previous chatbot sometimes struggled to understand questions and handle multiple queries. This is a common issue with traditional chatbots that require questions to include predefined words or phrases they’ve been programmed to understand.

Initially programmed using existing brand guidelines, the tone of voice came across as that of a teenager rather than a mobile network operator. The chatbot's tone was refined to balance being both youthful and professional, and informed new brand guidelines to better represent the desired tone for future communications.

Using Microsoft’s Azure AI Studio, the chatbot is built on the Azure OpenAI LLM framework which enhances its ability to understand context and handle multiple queries simultaneously. Development took approximately three months, involving multiple iterations to refine functionality and user interface. Accenture implemented an intent classifier that filters incoming questions to determine if they are suitable for response, focusing solely on topics related to VOXI mobile services.

The chatbot was trained using varied customer scenarios, including different tones and moods (e.g., angry or happy), to test and refine responses and ensure they’re appropriate to the tone of the query.

Regular checks ensure language remains appropriate. A kill switch was implemented, allowing the system to be shut down quickly if it were ever to deviate from expected operations. Alert systems were set up to notify the relevant teams immediately if the chatbot produced unexpected or inappropriate content, ensuring rapid response and resolution.

Setting the standard for telecom customer service

The chatbot has significantly reduced the volume of chats to VOXI’s customer service agents. Additionally, VOXI mobile employees benefit from reduced routine tasks and improved process efficiency in customer service. More important is the improvement to customer experience that helps drive Vodafone’s goal of becoming the number one mobile network for customer experience.

As the VOXI mobile chatbot continues to evolve, it not only enhances customer interactions through its advanced gen AI capabilities but also sets a benchmark by being the first of its kind in the UK telecoms industry.