A Local Government Story - Through CX Practitioners' Eyes

We asked one of our customers working in a leading county council to share their experience of adopting the use of AI Chatbots to manage their contact centre:

In my county council role, I’ve been working on telephony for many years, providing technical support for a contact centre with thousands of unified communications users. I also helped re-platform our self-service IVR, so I know about automating call flows. However, up to that point, I had no direct experience of building bots using AI and speech recognition technology.

For us, the real driver was the need to noticeably improve public services for residents. To do that we needed to escape the role confusion that comes with over-broad IVR menus. Although our system presented customers with only four choices – option one for adult services, option two for child services, option three for highways, and option four for everything else – that inflexible approach was too hit and miss.

Not least because it relied on the customer understanding how the council functioned as an organisation to get themselves to the right place. In fact, wherever you are, not a lot of people have that kind of insight. In addition, the IVR couldn’t capture valuable data such as why citizens were calling or the nature of their requests. So, our general enquiries option had become an impenetrable black hole.

We had already been using Genesys Cloud for around a year. Supported by Kerv Experience specialists, I was actively involved with our contact centre migration.

Setting up an effective call-triage bot
Just like most AI guides recommend, we started small and experimented on calls coming into our 24-hour IT helpdesk, a comparatively safe environment. This was also a known gap as we lacked visibility of out-of-hours calls coming into our IT teams during the night. Within four weeks we’d built our first voice bot with an AI platform using neuro-linguistic programming, which we chose because it was easy to learn, code and maintain.

The time-consuming bit was gathering data around our processes and actions. We kept things simple. The bot asked three questions: the customer’s name, their asset number and why they were making contact. Using speech recognition, the call got routed to the best-placed agent to deal with the specific issue. Very quickly we saw abandoned calls reduce by up to 50 percent.

Since then, we’ve been learning as we go. The trick is to fail fast. For example, we recently launched a self-serve password reset tool for IT users. However, we continued to receive a vast number of calls, so knew something still wasn’t quite right. So, we briefly suspended service, looked at the data, and tweaked the speech-recognition software. Job done.

As well as self-service, we quickly discovered bots are great at doing the legwork and gathering valuable data. That’s also informed our plan. We now have several live bots dealing with specific customer journeys and are about to launch another to assist with general customer enquiries.

Yet, in many ways, the exciting part is still to come. In 2023, along with other local councils we’ll merge into a single unitary authority. But, on day one, it’s likely each council will still retain its own contact centre. So, we’re figuring out how to best streamline service and minimise disruption as we consolidate onto to a single inbound number.

That’s another many-to-many AI-powered communications opportunity where bots with speech recognition technology will be able to do the heavy lifting. For instance, helping smooth volume spikes and deflecting calls through self-service. Or identifying customer intent and routing calls to the right agent, right first time, thereby increasing efficiency by reducing transfers, emails, and call backs.