CMO

How Deakin, TPG and Newcastle Permanent Building Society balance chatbots with human interaction

Australian brands talk through their digital-first customer service deployments and how they're striving to balance the mix of tech and human empathy in modern interactions

Focusing closely on a defined use case, stringent alignment to design principles and a commitment to learning through customer feedback are a few ways Australian brands are balancing chatbots and human empathy in digital interactions.

Speaking at this week’s Genesys G-Summit conference in Sydney, customer service leaders from Deakin University, TPG Telecom and Newcastle Permanent Building Society shared how harnessing voice bot and chatbot capability in a digital-first environment where human connection still remains critical.

Deakin University manager of customer services, Hayley Grey, admitted the education institution experienced lots of “intent mismatches” during the proof-of-concept trial of its chatbot. Deakin’s approach to combat this and ensure empathy in the digital-first experience is to focus clearly on the key purpose of the bot.

“If you just splatter this technology across everything, you probably won’t get something that wins,” she told attendees. “So for us being a university, we focused on our student base of 60,000 students. We thought about the one thing we want to get out of this. For us, it was being able to support students 24/7. That led us to identify what five things customers needed to support their journey 24/7.

“We provided guided journeys for those things, backed up by knowledge and self-service capabilities to make sure our students could be self-sufficient. Then when we’re open, we can pass to an agent, or when we’re not open, they can book a scheduled appointment via Zoom.”  

TPG Telecom rolled out chatbots as part of its Genesys Multicloud CX deployment. As well as voice and chatbots, the telco launched its speech IVR on the platform and with Google in October 21. Head of workforce optimisation and care platform operations, Adrian Cunningham, said the chatbot is running at about a 70 per cent containment rate today.

Customer feedback is key in understanding how digital technologies can improve and be better utilised, and TPG uses NPS surveys, for instance, to both look for any mention of IVR in order to find customer interactions and review them.

“It’s the same with our chatbot,” Cunningham said, adding that the global Vodafone chatbot is known as ‘Toby’.

In starting the journey to develop the chatbot, TPG commenced work by devising clear design principles. These were set from the beginning and signed off cross-functionally.

“We call these the 12 apostles – things like we never use our own language, we only use customer language,” Cunningham said. “The main thing was the bot is not there to get in the way of a customer talking to someone. It’s there to get them to the service and answer they need as quickly as possible.

“The design principles were really important. We brainstormed that and shared that with marketing, business and product and got alignment. We then followed those rigidly. What we’re talking about the whole time is all about empathy and putting the customer first. We work hard to put ourselves in the customer’s shoes.

“Basically, the bot will either help you and it’ll be really helpful, or you’ll talk to agent. There’s no inbetween.”

Another principle TPG ascribes to is ‘two nos and a go’. This is used in the event a customer asks a chatbot for something it can’t understand. After two tries, the chatbot will apologise, ask for a short sentence to describe what the call is about, then put the customer through to an agent. In all, TPG has devised about 45 different phrases and ways customers ask to speak to an agent.

“You have to set your principles and be truthful, faithful and honest to that. Then it’s all about your feedback loops,” Cunningham advised. “That’s why we constantly dive into surveys and agent feedback to find the kinks and nuances when we have ended up in a loop for whatever reason. And sometimes call the customer back for further feedback and to apologise.

“Like any business, we will release new products, make changes, so you have to create new journeys. We test and test. But the best test is for customers to roadtest this, so we’re very keen to find out what they think.”

Over at Newcastle Permanent Building Society, bot technology is used in two ways. One is a voice bot, which utilises the foundation Genesys platform as well as Google AI and has now automated about 30 per cent of daily voice call interactions with customers via IVR. Many are account balance inquiries or basic transaction inquiries, such as card activation.

“After that, we use the bot and AI piece from Google to listen to what the customer wants to do and asking to do, and leverage that to find the right agent at right time and right place very quickly,” head of digital customer experience and innovation, Simon Burt, explained.

The building society’s business strategy centres around keeping waiting times as short as possible. “Any more than a few minutes wait time stresses people out including our CEO. Our strategy is finding people very quickly to help that customer, because they made the decision to call as they probably couldn’t self-serve.”

That’s not just servicing through the contact centre. Newcastle Permanent Building Society used its deployment of Genesys three years ago to stretch its customer support base out of the call centre and across its physical branch network.

“The beauty of the voicebot is understanding and learning, as customers use a certain phrase and it learns to guide them… we are surprised at how far and fast that starts to happen,” Burt added.  

Panellists also shared how they’d approached development of their bot technology. In Deakin’s case, the team trialled a round range of options, such as a bring-your-own bot.

“We went with doing guided journeys, developing those internally. Our strategy around this is understanding the things that will cause the most angst or volume outside of core hours. Then we focus on doing a process map for those things and building that into the bot,” Grey said. “That’s particularly things that are far reaching from an enterprise perspective.

“We recently supported 60,000 students doing online exams and had 200 phone calls at once. There was no way we had enough people to do that. Making sure the bot understands every point of instigation for a customer as well as point of resolution, so we can funnel our students into that environment and get up and running very quickly, was key.”

TPG built out its chatbot internally with help from Genesys. “We have the advantage of having speech analytics, so we knew what our top 100 customer intents were,” Cunningham said.

“We wrote journeys for those main intents. We started with the chatbot and quickly developed new ones. The reason is we did go fast and hard…. That’s driven by good discipline and listening constantly to customer feedback. Then we went live with the speech bot… we had speech analytics, chatbot and our experiences and we ported that across. Speech analytics means you get very accurate insights into what customers are asking for.”  

Newcastle Permanent Building Society also partnered with Genesys to implement the Google AI voice bot for its IVR as part of early deployment. “We really were learning on the fly with Google AI and Genesys,” Burt said, who claimed the rollout was one of the first instances globally of Genesys and Google AI technology harnessed in concert.

“We have the voice bot right now, but will leverage that to go into chat,” Burt said. “Where the smarts have started to come in is finding the right agent and the right time with the right skillset. But also being able to make changes on the fly to the voice IVR is useful. We might have predetermined journeys in banking, but if the government benefit payments don’t land at right time, we get flooded with calls.

“We can quickly react to that, type into the bot to voice a response and give the customer a message to say, for example, payments have been delayed until 5pm and resolve their question. That’s automation in action – knowing what the customer wants and we’re trying to give them an answer.”

Read more: What not to do when building chatbots and voice-based brand interactions

Don’t miss out on the wealth of insight and content provided by CMO A/NZ and sign up to our weekly CMO Digest newsletters and information services here.  

You can also follow CMO on Twitter: @CMOAustralia, take part in the CMO conversation on LinkedIn: CMO ANZ, follow our regular updates via CMO Australia's Linkedin company page