Case Study

Local Government AI Assistants

The team at EBI.AI developed two conversational AI assistants for a local government authority who were looking to innovate their customer service to reduce the number of costly frequently asked questions that their contact centre received.

In a phone interview with the Head of Customer Services, we learned more about the project and his experience with EBI.AI. Find out why he took on this project, how the assistant managed to reduce contact volume during the PoC, and what it was like innovating with EBI.AI. 

Key results at a glance of local government AI assistants:

  • 35% of total queries were accurately resolved outside of the call centre hours by a 24/7 virtual assistant.
  • Within a month, 85% of queries to the voice assistant were successfully resolved.

Background

In 2019, EBI.AI were put in touch with a provider of outsourced government process services and a local government authority. Through intelligent customer engagement and improved service delivery they had managed to create millions of pounds worth of savings each year and they recognised the need to continue to innovate to build on this success.

EBI.AI tested whether an AI assistant could provide customers with faster self-serve answers to their FAQs, whilst reducing call handling costs.

Challenges & opportunities

Enquiries to the contact centres are handled over the phone between 8:30am and 5:30pm. The biggest volumes of calls are in two areas; council tax and missed bin collections.

There are certain Key Performance Indicators (KPIs) like abandonment rates and speed to answer that are vital to maintain to achieve the Service Level Agreements (SLAs) with local government clients. In addition, the higher the volume of calls, the greater the cost to the business because we have many agents waiting for those calls.

By its very nature of being a contact centre, staff attrition and sickness rates are potential problems that can endanger the SLAs with clients. Therefore, customer self-serve options like online forms and AI assistants are a great way to help achieve KPIs and importantly, also provide a great service for the customer who should get their query resolved very quickly.

Therefore, EBI.AI were asked to innovate the self-serve functions of clients’ sites to deflect calls away from their contact centre. Did you know: by implementing the latest self-service options such as chatbots and AI assistants, you can expect to see deflection rates of calls to busy contact centres between 30% and 70%? To get more statistics relating to AI and local government applications, we wrote a handy blog about it here.

The solution

Local Government AI Web Assistant

EBI.AI created two AI assistants to automate FAQs around council tax and missed collections; the two topic areas that resulted in the highest volume of calls for local governments. A web-based AI assistant was deployed to handle council tax queries and a telephone-based AI assistant to handle waste management queries. The assistants were trialled for a little over one month with residents of one local government partner.

The assistants were configured on EBI.AI’s Lobster communications platform (Lobster) using Natural Language Processing (NLP) Technology. Due to its technology agnostic design, Lobster allows clients to switch to other NLP technologies, or deploy to other chat and voice channels, including telephony and smart home devices.

This means the service is future-proofed for rapidly developing technology orchanging resident needs. By using Lobster, clients can get enterprise-grade solutions live in weeks, rather than months.

Listen to the recording below to hear an example of how the AI assistant can handle a missed bin collection. 

According to the client, “the initial engagement and my initial meeting was very easy, very straightforward. I came along to the EBI.AI offices in Warwick where the team had mocked up something for me to have a look at which was relevant to the questions I had previously provided. It was an impressive demonstration and I could see at that point that there was definitely going to be potential with what we wanted to do.”

Local Government AI Assistant Performance

What was your favourite feature?

“A favourite feature was the assistant’s ability to send a message or an SMS with a link if it was required in the context of the conversation. This was a very good feature.”

The results

There were two high level objectives for this Proof of Concept (PoC).

  1. To see how well an assistant could interpret and understand customer questions.
  2. To assess whether the assistant could reduce calls coming into the contact centre.
Local Government AI Assistant Success Metrics

The graph showing the waste management assistant learning over time and increasing the % of queries it can handle successfully

In the final week of the PoC, the web assistant successfully answered the questions of users 85% of the time. “Looking through the transcripts there is plenty of evidence to show that an assistant is capable of answering questions posed by our residents and therefore the assistants were deemed to be successful enough for us to want to go further with this technology.”

“The beauty of the assistants, of course, is that we can have them running 24 hours a day. The web assistant ran all the time and we found that around 35% of queries were handled when the contact centre was closed, which is good because we can infer, that the customer may well have just waited until the contact centre was open and called in had the AI assistant not been available.”

Handling out of office queries alone presents a significant ROI opportunity for local government authorities. The improvement of the assistant’s ability to understand and successfully respond to residents’ questions, even over the month that the PoC was active, shows the effectiveness of the EBI.AI experts to respond to resident’s needs and improve the performance over the assistant over time.

Working with EBI.AI

“This was a Proof of Concept and we have learnt a lot from it. A big thing for us is that a signposting assistant isn’t going to do enough for us. The assistants have got to be able to integrate with our back-office systems so that they can pull out and push in the necessary information as required.

The technology exists and this is within the scope of what EBI.AI can do for us going forward.” “But in general, the whole experience of working with EBI, whether it is the salesperson, or Matthew or the technical people like Olly. It’s been a very positive experience.”

In summary

Overall, the two AI assistants built by EBI.AI helped to provide faster customer self-service answers to their FAQs that generated huge cost savings and efficiencies. Voice assistants are just one example of our AI solutions – to discover other types of AI assistants and how they can transform your business, check out our solutions page or get in touch with us to discuss your business’s needs further.

Reference Architecture

Curious how Lobster works and integrates with your business? Take a look at the reference architecture.