The social housing sector is second to none in its aspirations, but competing priorities require creative solutions when it comes to delivering great customer service. There’s always a need for human-to-human contact but a well-trained chatbot can go a long way towards providing the 24/7, immediate service desired by both tenants and housing providers.
After creating and implementing chatbots for housing providers over the past five years, I’ve learned the three most important things to get right are:
1. Ensure the accuracy of the chatbot’s responses
Customers get annoyed when a chatbot doesn’t recognise what they’re asking. It’s worse when the chatbot recognises their intent but provides a response that’s not very helpful. When this happens, customers return to traditional contact methods but having now wasted time, are more frustrated than they would otherwise have been.
- Train your chatbot to understand the customers’ intent: Training the AI so it recognises the range of questions and language your chatbot’s likely to encounter is critical to success. I’ve learned customers will ask for things in ways I could never imagine and that it’s sometimes necessary to clarify the customer’s query. A chatbot should be familiar with different types of housing stock, repair problems and ASB complaints to name but a few things needed to be helpful to tenants.
- Provide a breadth of responses: A housing chatbot must have a broad set of responses. For example, it’s no good answering the myriad possible rent questions with a few generic responses. Your chatbot needs a specific response for everything from rent holidays to rent increases and must be able to provide tenants with rent statements if you want to reduce call volumes.
- Provide well-crafted responses: To give customers the information they need, a chatbot’s response must be detailed. A word of caution here though … chat windows are small, so responses need to be concise and well edited. If you want your chatbot available on voice devices such as Alexa, it’s even more important to make sure the responses are well crafted and straight to the point.
2. Create a self-service channel
From a customer’s perspective, it’s never ideal to be left on hold, told “I’m sorry, I can’t help with that”, or sent elsewhere to get a problem addressed. This is equally true for chatbots; customers prefer to have their issue resolved without being redirected to a portal, website or telephone.
- Train your chatbot to handle more than FAQs: Using a chatbot to display FAQs is a sure way to frustrate your customers. While it’s true that there’s an 80/20 distribution of problems, that 20 per cent is a long list when you’re talking about social housing. It’s very different to banking or retail which have a relatively narrow field and consequently small number of questions customers might ask.
- Automate business processes: If a tenant asks, “how do I book a repair?”, the likelihood is they want to book a repair, not merely understand the process. A chatbot that directs them to the contact centre or portal to book a repair doesn’t add much value. On the other hand, a chatbot that can help identify the fault, raise a work order in your HMS and book the repair with your scheduling system would be helpful and impressive.
- Replace application forms: It’s commonly accepted that whenever we ask for anything these days, a form must be filled in. This is an odd way for an advanced civilisation to behave. A chatbot can provide a more natural experience for customers and is a way to reach those who might be digitally excluded by more complex user interfaces. It can remove the confusion caused by complicated forms by using its intelligence to skip irrelevant questions and options.
A good chatbot should offer the conversational experience of gathering information while keeping the efficiency and cost benefits of forms. Things such as housing applications, home swaps and adaptation requests can all be managed one question at a time through a conversational interface.
3. Manage escalations well
It’s unlikely any chatbot will be capable of answering every question it’s asked and the ability to escalate to a real person is therefore important for good customer service. Some customers need to talk and where appropriate, it’s good practice for a chatbot to offer them the path of least resistance to an advisor.
- Handover to live chat: If your chatbot can’t resolve a customer’s problem, passing them to a live agent is the best possible outcome. A transcript of the chatbot’s conversation should be available to the agent so that customers don’t have to repeat themselves. If they do, they’ll (rightly) perceive their engagement with the chatbot as a waste of time.
- Offer call-back requests: If handover to an advisor is necessary but the contact centre is closed, it’s best for your chatbot to collect the customer’s details and arrange a call-back.
- Monitor how your chatbot performs: The last point I’d like to make is that it’s important to monitor those questions which require handover. There might be a reason why these questions aren’t being handled by the chatbot – maybe the response isn’t detailed enough or it’s missing the point. If you can identify these and continuously improve the product, your chatbot will naturally get better over time.
A chatbot represents your organisation and is often your customers’ first point of contact. It’s an extremely important job. A good chatbot that’s designed specifically for social housing will not only reduce call volumes, it’ll also create happy customers – and that’s what everyone wants.
Scott Summers is the co-founder of Fuzzlab.