Nedbank Live Chat & Chat Bot Integration
Nedbank clients want quick answers or solutions to resolve their banking problems in their time and in a way that suits them.
Enable Nedbank Clients to engage with the bank through Live Chat and Chat Bot services, via web and app.
• CX Design Lead (me)
• UX & UI Designers
• Design Lead
• Product Owner
• Stakeholders: Contact Centre Managers, Product Design Head, Digital Executives
Nedbank has an award-winning contact centre and the bank wanted to provide their clients more access to the contact centre agents other than just via a phone call
JOURNEY MAPPING AND USER FLOWS:
I needed to understand what journey a typical Nedbank client followed when wanting to contact the bank with any query, specifically within the live chat space..
I decided to conduct interviews with Nedbank Contact Centre (NCC) staff and requested statistics about the top ten queries that came through everyday. I compiled more customer journeys and user flows to determine where the pain points were.
I organised usability testing sessions and user interviews to discover how people might interact with a banker over a chat system and also how that journey might be affected if a chat bot were introduced.
Guerrilla Field Tests:
We performed quick ad hoc user interviews to test the UI for one of the proposed solutions.
The feedback was recorded and noted in order to amend the UI and UX to prepare for the next set of testing.
SENTIMENT TOWARDS THE CHATBOT:
In the interviews with some participants to determine how they would interact with the chatbot and the changes to the Live Chat system, we collected what they said in order to present this to our stakeholders.
From the interviews we found out a few things:
- Human interaction is preferred over the chatbot due to the personable nature of the interaction – one person mentioned they would rather go to the branch because they knew them personally.
- Most would not trust a chatbot with sensitive information
- They wouldn’t mind interacting with the chatbot as long as it would be able to solve their query.
Qualitative data about whether people preferred talking to a bot or a real person, were inconclusive. I determined from the results from my other tests and interviews, that putting the chatbot before a human agent wouldn’t be a problem so as long as the client’s query was answered straight away in coherent, human speech.
After more testing and research, we tried to find ways to design what the bot would say and how to present options for a client’s query.
We chatted with a human contact centre agent an had an idea based on their response for a query that fell in the top ten category:
The Bot should be able to resolve top queries just as a contact centre agent would. Instead of trying to be too conversational, instead, get straight to the point in as friendly a way as possible.
To see how people would interact with the bot, we gave a few people a scenario and then asked them to use the chat system to try and resolve it. We observed their interaction with the bot and discovered that most of them would type out full sentences instead of just key words – which we assumed would be the case.
When the bot did not reply in the way that they expected, they became frustrated and typed a full sentence again! They also acknowledged that it is a bot that answered but continued regardless, hoping for a better answer.
We concluded that it would be better to give options upfront that allowed clients to choose from a list of the most common queries as these were the queries coming into the NCC anyway, via call and chat channels.
I was moved to another team before the conclusion of the project and launch.