Vectors of improvement for conversational user interfaces

Vectors of Improvement.jpg

The iterative nature of digital product development allows product managers to explore many different vectors of improvement. In one sprint we can dedicate resources to optimizing the user experience, while the next is dedicated to implementing new product features. From simple bug fixes to launching digital products through new distribution channels, product managers must constantly weigh business and user needs to allocate resources towards the optimal vectors of improvement.

Thanks in part to this optimized development process, user interfaces have experienced massive evolution (think of the transition from typing commands, to mouse clicks, to touch screens), conversational user interfaces (UI) like virtual chatbots and voice assistants are beginning to rise up as the technology of the future.

We’ve previously shared details from the product team behind ATB’s virtual assistant in Facebook Messenger, including their fondness for ‘decay of delight’ that drives product improvement.

Now we’re giving a sneak peak at how they are taking their passion, learnings, and experimentation to be ahead of the curve with natural language processing and the user interfaces of the future. Check out some of the vectors of improvement where the team is creating business and customer value with conversational UI, starting with the ATB virtual assistant.

Vector 1 - UX improvements: mapping natural language to intent

It is one thing to train humans to input commands that a computer understands. It is a completely other process to teach a computer to map natural human language inputs to intent. That is the first layer for our approach to conversational UI.

Improving the user experience provides customers the solution to their problems with as little friction as possible. Improving confidence levels and increasing the scope of utterances is the key way to reduce friction for conversational UI.

From learning all the different ways that someone may ask for their account balance, to increasing the number of tasks that can be performed autonomously, every challenging question that a customer asks is an opportunity for us to learn and improve.

Try asking any voice assistant or chatbot what the exchange rate is right now, you’ll likely get a response, but it might not be the exact form you were looking for. From early users, we took this learning and improved the way the ATB virtual assistant responds. Now you’ll get options for CAD to USD, USD to CAD, or links to other exchange rates. Making sure that users get the answers they are looking for, not frustration.

Vector 2 - Integrations for task completion

Following mapping an utterance to intent, the next layer in conversational UI is to perform an action based on the identified intent.

Here, it is crucial to integrate back end systems so that tasks can be performed autonomously. If Alexa is not paired to your phone, you won’t get the song you requested. If Siri can’t activate your messenger app, it will be hard to use voice to text your mom.

We’ve got strong collaboration between our internal data teams to provide the needed back end integration. That has also allowed us to provide features like push notifications with proactive and personalized notifications to customers.

Can you imagine if your cell phone provider proactively called you to tell you that there was a plan that would meet all of your needs and save you money? Probably doesn’t happen now for most of us...what if ATB could do that?

We are now testing a feature in the ATB virtual assistant that will track customers’ recent account fees and transactions, and will notify them if they could have saved money on fees using a different type of account.

With conversational UI optimized for natural language processing and tied into back-end systems, the future is bright for creating frictionless banking experiences.

Vector 3 - Optichannel experiences to support conversational UI adoption

We’ve talked before on alphaBeta about our commitment to providing exceptional optichannel experiences, where customers can select the appropriate experience based on the channel they are in and the situational context.

As users become more accustomed to the capabilities and limitations of voice and text interactions, optichannel remains a crucially important vector of improvement for conversational UI.

We are striving for seamless optichannel experiences through a feature that allows customers to move from the AI based chat to a customer care agent within Facebook Messenger.  

As we work on expanding the tasks that can be autonomously completed, there are still instances where customers need to connect with an ATB team member. So if someone is chatting and they are faced with a problem that the bot can’t complete - the bot can now connect them directly to a customer care agent who can authenticate the customer within the chat and provide support.

We recognize that dedicating development resources to maintain optichannel integrations will remain an important factor in adoption rate of conversational UI and we are optimizing our development time accordingly.

Vector 4 - New channel implementation: the magic of channel-agnostic conversational UI

Conversational UI is fairly channel-agnostic. You’ll talk the same way with a voice assistance through your smart-watch, phone, or desktop computer.

For the ATB virtual assistant, that means that the learnings we draw from our integration with Facebook Messenger can be easily rolled out across other channels.

One of the chief benefits of the ATB virtual assistant when we launched was having Facebook Messenger as the channel. ATB customers were able to connect with their bank in the same channel where they connect with family and friends. Our aim was to help people integrate positive financial habit building into a channel that they were already using.

That channel was just a start for our conversational UI. Internally, we are working with a virtual-assistant-as-a-service model to deploy the user experience into different ATB properties. We have the capabilities to integrate the bot into channels such as the ATB website or our own mobile application.

Our development resources are dedicated to the vectors of improvement that will provide value to ATB customers. With the ATB virtual assistant, we’re improving mappings of natural language to increase the scope of tasks that can be undertaken by the bot. We’re making sure there is seamless optichannel integration and rolling out conversational UI into multiple channels to help integrate the banking experience.

There is much work to be done in fintech and across sectors in the capabilities and adoption of conversational UI, but we see its potential to greatly reduce friction in banking and an almost unlimited number of uses cases. We’ll continue to work across vectors of improvement to create happiness through banking.

Interested in joining the Transformation team and working on reducing friction in banking through conversational UI? Check out our open positions.