South Cambridgeshire - Contact centre chatbot to improve resident satisfaction

South Cambridgeshire District Council's IT team, driven by leadership's vision for enhanced resident service, has initiated the phased implementation of an AI-powered solution within their contact centre. This strategic move aims to streamline responses to routine information queries, representing an estimated 27 per cent of all incoming calls. By deploying a comprehensive suite of tools, including chatbots, voicebots and interactive voice responses (IVR), the council anticipates a significant shift towards more efficient query resolution, freeing staff to address complex resident needs.

South Cambridgeshire Council rigorously assessed potential suppliers for functionality, safety, security, adaptability, and value for money. They were particularly meticulous in their procurement process with EBI, having explored a contract with another supplier. A detailed review of the contract agreement revealed a raft of hidden costs tucked away in the clauses, making the product far more expensive than initially thought. Consequently, the IT and procurement teams worked in close collaboration, to review the EBI contract clause by clause. Satisfied that they'd done their due diligence, the council felt confident in proceeding with the purchase. By trusting their instincts and maintaining a critical eye throughout, they ensured they secured a supplier that offered genuine value.

As part of the EBI agreement, the council opted for a professional services charge. This allows them to request incremental improvements at a fixed cost, such as integrations with other services, ensuring the technology works optimally for their needs.

The technology's deployment has been remarkably swift. Within two months of contract signing, the chatbot element was operational. This was followed by the launch of live chat in early 2025, AI interactive voice responses in February, and the scheduled rollout of AI voice in March.

Importantly, this rapid implementation hasn't come at the expense of accuracy. Aware that industry standard chatbot accuracy typically ranges from 70-90 per cent, the council set an ambitious target of 90 per cent. Through close collaboration with the vendor, they've surpassed this, achieving an impressive 93 per cent accuracy rate. Continuous refinement has been key to this success, with the council training the AI model using their website and policy documents and establishing predefined flows to prevent overly generic responses to sensitive queries. Staff user testing has also played a crucial role, helping identify and rectify areas where AI was not accurate, such as distinguishing between the chief executive and the council leader.

Ongoing development is focused on integrating the AI with various council services. A notable example is the integration with waste services, enabling the system to respond to queries regarding refuse collections. EBI's team have proven highly responsive to the council's requirements, demonstrating agility in delivering solutions – the waste services integration, for instance, was completed in just a few hours.

Residents have been actively involved and engaged in this transformative project. To foster a sense of community ownership, the council ran a competition to name the chatbot, with 'Sam' emerging as the winning entry. This, coupled with a magazine article outlining the chatbot and AI tools as a helpful and knowledgeable friend, aimed to introduce and raise resident awareness of AI in an accessible and welcoming manner. This approach has been important in building trust and transparency around the council's use of AI in public-facing services.

Post-chatbot deployment, the council has maintained a proactive approach to resident engagement, continuously monitoring user experience. They're actively collating resident feedback through a survey to gain a deeper understanding of how well the chatbot meets their needs. The tool has garnered a positive reception, evidenced by the thumbs-up/thumbs-down feedback feature, which consistently yields favourable results averaging between 86-90 per cent approval.

While the full impact of the technology's rollout is still unfolding, early indicators are encouraging. In the chatbot's first three months of operation, the council observed a noticeable reduction in call volumes compared to the previous year's figures. They anticipate that the introduction of voice AI will further improve this trend. As the technology becomes fully integrated into their operations, the council aims to see a decrease in routine information requests handled by contact centre staff, accompanied by an increase in the volume of complex enquiries handled in timely manner. This shift is expected to increase the average call duration, a metric that is being closely monitored.

The IVR technology is also yielding positive outcomes for call centre staff, particularly in its enhanced call wrapping capabilities. By accurately categorising calls based on subject matter – a process liable to human error by callers – the IVR ensures officers receive more calls on the correct topics, thereby boosting the centre's overall efficiency.

To ensure robust governance throughout the project, South Cambridgeshire Council implemented a combination of existing and newly developed measures.

• Data Protection Impact Assessments (DPIAs) and Equality Impact Assessments (EqIAs), standard practice for the IT team, were conducted with enhanced rigour, reflecting the specific considerations associated with generative AI.
• Comprehensive AI guidance and policy documents were developed concurrently with the project's progression.
• A dedicated AI Board was established to oversee the project, with regular progress updates also provided to the Transformation Board.
• A senior leadership sponsor was appointed, providing a direct input and oversight from the senior leadership team.
• Regular demonstrations were presented to the Cabinet, facilitating timely project approvals. For example, following a presentation explaining how the IVR technology would streamline caller routing without replacing human interaction, approval was granted, and the technology was launched the subsequent day.
• The council is in the process of establishing a Digital Working Group. This group will convene to discuss a range of digital initiatives, including AI, and will comprise technology specialists such as systems administrators, to bolster the council's contingency planning.

This project has provided valuable insights into areas where the council's communication strategies can be refined. Specifically, challenges related to outdated website information have been highlighted, as the AI tool relies on these pages and policy documents for its responses, underscoring the necessity for a robust system to ensure website content remains consistently up to date. Furthermore, the AI chatbot has revealed instances where web content lacks clarity or fails to fulfil its intended purpose. For example, despite the chatbot being located on the 'Contact Us' webpage, users are frequently asking how to contact the council, indicating difficulties in locating the desired information. Additionally, the AI tool's confusion when distinguishing between council programmes with similar names has raised concerns. The council acknowledges that if the tool experiences such difficulties, residents are likely facing similar challenges, necessitating a review and amendment of the relevant information architecture.

South Cambridgeshire Council is committed to the ongoing development of AI integration to enhance service delivery. The council intends to expand the number of services connected to the AI, however, they are encountering obstacles due to legacy systems that lack the necessary Application Programming Interfaces (APIs). Furthermore, they are exploring the possibility of integrating the voicebot with Gov.uk Notify, enabling the council to send text message updates to residents who have requested information.

Having been satisfied with the performance and collaboration with EBI, the council is now actively investigating the feasibility of deploying the same chatbot technology for internal use.