In September 2020, Believe Housing launched its first AI-powered chatbot. Six months later, Simon Bartlett, Believe Housing’s customer experience business leader, explains the realities of automated customer contact and how it’s helping the housing provider to achieve ‘a life without barriers’.
Sometimes it’s better to start with a clean slate and a fresh outlook rather than carry the burdens or expectations of the past, and I certainly think that’s true in our case. Since we rebranded in 2019, we still feel an adolescent energy and optimism that has fed into our approach to technology and digital transformation. This has resulted in some very generous plaudits over recent months, praising the business for the various ways we’ve adapted to the challenges presented by the pandemic.
But in my opinion, the seeds of our adaptability and resilience were sown in our foundations in 2015 where, under our previous guise as County Durham Housing Group, agile working and digital transformation became part of our DNA.
A life without barriers
Our chief executive Bill Fullen was clear from the outset, instilling a belief in “a life without barriers”, both for the way we work and how we serve our customers. This gave us a licence to be brave, try new things and take calculated risks in order to quickly prove what works and, just as importantly, what doesn’t work.
It was this approach that led us to the potential benefits of automated customer messaging, and March 2021 marked the six-month point of launching our new chatbot service.
We took the decision to work with Futr, who at the time had limited experience in housing but did have considerable experience of working with civic organisations with a similar audience profile, such as local authorities and the police.
We decided to soft launch the service initially, deploying it only on our website so that we could measure its performance. We asked ourselves, “Could an automated messaging service be seen as denying our customers the opportunity to talk to a ‘real person’ who can answer all their questions?”
To understand this, you first have to understand the nature of housing customer contact and the needs of our tenants.
What customers want
More often than not, the customer isn’t seeking a conversation, but the right information at a time of their choosing. We had a good understanding of this based on feedback from our customer contact team, who would regularly be signposting people to the right information via our app, rather than untangling more complex situations.
This is supported by our six-month numbers for the chatbot service; the 5,000 questions received by the chatbot generated only two hours of automated conversation time. This equates to 0.67 seconds of conversation per message. That sounds like an awful lot of very quick chats, right?
In reality, most messages are answered by the chatbot supplying a link to more information rather than an ongoing automated conversation.
The content and accuracy of the chatbot’s response is determined by ‘confidence scoring’; this uses intuitive AI to measure content against an existing and ever-expanding library of data. A high level of confidence or accuracy means the answer to the question is given instantly, accounting for zero seconds of conversion.
Around-the-clock service
Another telling piece of data is when customers choose to use the service, with 31 per cent of enquiries being made outside normal business hours, which is increasingly becoming the norm in the private sector.
Automated contact also removes many of the language barriers, and so far our chatbot has responded to customers using seven languages as well as English.
We have had to make changes to optimise performance and learn some lessons along the way, but that’s only to be expected from any new solution. For example, we initially didn’t set a limit to the length of an inbound question, which meant some enquiries became too lengthy for the system to decipher.
We’ve also had to make significant changes to our coronavirus customer safety policies, particularly around repairs and maintenance visits, but the system is very easy to update and intelligently link to new content.
The chatbot also helps us to proactively shape the information and content we produce and publish by highlighting trends in customers’ questions.
On reflection, automated messaging has been a positive step for us, meshing well with our values and helping us to promote digital transfer and customer self-service during a difficult time for us all. And most importantly, our customers are finding it an increasingly useful source of information as and when they need it.
Simon Bartlett is the customer experience business leader at Believe Housing.
