Oxford City Council
Can chatbots help councils?
After winning (a highly contested) GDS Discovery project, we set out to understand the impact chatbots and AI technology could have for 13 councils across the UK.
Our goal was to find out if chatbots can help councils:
- Improve the user experience for citizens engaging with council services
- Increase councils’ efficiency in delivering these services, and
- Reduce the burden on call centres (and saving money!)
We looked at four key services at 13 different councils: Planning, Waste & Recycling, Revenues & Benefits and Highways. We worked with 20 stakeholders spread across the UK, hailing from a variety of professional backgrounds (senior managers, Customer Service, IT and Data Officers). Everyone had differing levels of experience with technology, user research, and collaborative project management.
Let's get started
We needed to implement the right tools and processes from the outset, in order to tap into the value of a large team within a short time frame for the project (we only had eight weeks to produce the final report).
We followed a repeatable process across four research sprints to uncover customer needs in each council service area. This approach also helped us work with such a geographically dispersed set of stakeholders. In each four-day sprint, we ran:
- User interviews to understand citizens’ needs of council services, and how they choose to make contact with their councils
- Stakeholder interviews to better understand councils’ internal needs and structure, and to dig in to how the provision of services is managed from within councils
- UX research training sessions to share a set of tools and techniques for current and future UX research
- Business data review to understand patterns in the usage of council websites and phone lines, with a view to establishing the service with the greatest potential return on investment
- Analysis and documentation to share our findings in an accessible way for the councils and the government sponsors
- Show and Tell meetings in order to share findings and foster a culture of openness and collaboration. We also contributed to the project blog.
What we learned
We learned that Waste and Recycling services are a good fit for a chatbot. Here’s why:
- Waste and Recycling makes up a high proportion of calls into councils.
- Nearly all these calls are requesting basic information or are task-based enquiries suitable for handling by a chatbot with comparatively few complex calls requiring second-line intervention
- There are relatively few distinct reasons for calling which means a chatbot can address a large proportion of all Waste and Recycling calls.
- Users have an expressed desire to self-serve without needing to make a phone call, but only turn to the phones as a last resort.
- Waste and Recycling exhibits spikes in call volumes that coincide with call centre resources being under strain (e.g. at Christmas or during adverse weather).
- Provision of Waste and Recycling services is largely consistent across a high number of different councils, meaning the investment in a chatbot could be shared across a number of collaborating councils.
We anticipated that a single council investing in a Waste and Recycling chatbot could save £38,000 in the first year of implementation, including the costs of building and training the conversational AI. But if every council built their own Waste and Recycling chatbot in isolation, then while potential savings exist, any return on investment will be limited. We would end up with a fragmented council ecosystem and enormous duplication of effort.
We were interested in the benefits and savings of council collaboration across a technical solution. We calculated that if 20 councils were to collaborate on development of a Waste and Recycling chatbot, it would be reasonable to estimate total savings across the participating councils at £2.2m annually.
We were delighted with how well Torchbox responded to the challenging timetable of our project, and how well they led us through its delivery. Their flexible approach, expert knowledge and high quality reports were significant elements in making the project a successNeil Lawrence, Digital Transformation Manager, Oxford City Council
Plenty of the observed user needs and frustrations were addressable through the existing council websites by, for example, making improvements to site structure, navigation, and search functionality. Additionally, a large number of calls to the council could reasonably be handled by a chatbot, fulfilling basic transactional tasks and info requests to free up call-centre agents to focus on the calls where human conversation is essential.
Our final report has been published and shared with the Ministry of Housing, Communities and Local Government. Hopefully, they will secure funding for an Alpha stage, during which we could look to begin developing a chatbot, carrying forward the culture of collaborating with multiple councils.