How to be a conversational bank – Seven steps to smarter customer service

Share this post:

In my previous blog post I discussed the emergence of conversational banking – the use of highly-capable virtual agents, underpinned by customer data and cleverly-applied AI, to make digital banking personal. So how do you get started – and what are the key things to focus on?

Technology is already enabling conversational banking. Platforms such as IBM Watson have made it relatively easy to run virtual assistant proof-of-concepts and build conversational flows.

What presents more of a challenge is the process of designing, building and implementing a computer program capable of having human-like conversations and resolving customer questions within a large, heavily regulated organisation. And make no mistake, this is where the real value lies: in creating a virtual assistant with a character unique to your bank and brand, with proprietary knowledge of customer journeys, backed up by analytics.

Firm foundations

Based on IBM’s experience working with tier 1 banks, below is a rundown of things you need to consider in order to get it right.

1. Nail down your strategy, vision and ambition up front. It’s important to differentiate your experience from that of other banks; you have to prioritise, design and own your conversations and rule out certain solutions. This shouldn’t come at the cost of agility, but it’s essential to know what “good” looks like from the outset.
2. Treat your virtual agent as a member of the workforce. Put as much effort into designing it as you would into hiring and training a human agent. Give it a persona and job specification, and the end result will be something worthy of talking to your customers.
3. Build for what you need. Don’t build for every “what if” scenario. Base your build and backlog on data and continual research, even if that means focusing on a single use-case or customer journey first. Don’t be afraid to fail fast: keep learning, keep adjusting and stay true to what your customers demand from you.
4. Design beautiful conversational experiences. This doesn’t happen by accident; it takes time and investment in diligent research. IBM invests heavily in conversational design informed by sociology, psychology and linguistics, as well as newer fields in computing. This research ensures the virtual assistant is clearly signposted, able, natural and pleasurable to interact with.
5. Bring your risk team on the journey. Don’t build and then try to “get past risk” at the end – it’s sure to backfire. What you see as an exciting new channel could be seen by your risk colleagues as a new compliance nightmare. Include them early and use IBM’s proven governance and risk controls for human-assisted learning and data stewardship.
6. Build in the flexibility to react to change. AI technology is changing every day, as is the content you need to be able to react to, so your instantiation will require carefully managed but rapidly deployable change. The ability to continually deploy new training and content must be engineered into the solution.
7. Keep your standards high. Accuracy and precision are absolutely key to ensuring the virtual agent can understand and process customer queries, and create the impression of a human interaction. Powerful classification models are essential for success. Equally, the virtual agent needs to be able to fail gracefully when it doesn’t understand something. IBM has developed a fully scalable proprietary “conversational architecture” that gives results of over 90% accuracy.

Expertise and experience

IBM is working with a host of tier 1 banks to develop artificially intelligent conversational banking experiences in the following areas:

• Helping customers find answers to their queries at the first point of contact
• Dealing with one-off events that cause a spike in calls to call centres
• HR and IT functions for routine task management
• Knowledge support capabilities, such as product and policy information

Working with a large UK bank, IBM deploys three times a week to constantly improve the virtual agent’s knowledge and react in near-real time to world news. For example, when the Bank of England changed interest rates in 2017, IBM worked with the bank to understand the impact on mortgage and savings customers and define treatment strategies. So when customers asked questions about the impact of the rate changes for them via the mobile channel, they got the answers they needed instantly. This is a great example of how building in the flexibility to react to change can reap rewards.

IBM offers enterprise-ready products such as Watson Assistant, proven methodologies and frameworks executed by experienced teams, all of which enable complex organisations to fully exploit the power of artificial intelligence.

Get started today

Conversational banking is a tangible example of the way AI can augment human intelligence to make day-to-day exchanges not only convenient but valuable.

Talk to us to find out how you can get started today.