ANZ Bank Dispute Transaction 'Lotti' Bot

Project: ANZ Bank

ANZ Bank is one of the largest financial institutions in Australia, providing a wide range of banking and financial services to millions of customers across the country.

Problem:

'Dispute a transaction' is the Bank's top contact driver. This is because customers often don't initially recognise a transaction and want to report it as fraud or a scam.

Solution:

In reality, over 50% of enquiries are resolved as recognised transactions once users utilise the bank's 'Look Who's Charging' feature, Google the merchant name, or wait until the transaction is cleared and no longer pending.

What I Did:

• Designed the Chatbot 'Lotti': Developed the chatbot for both web and in-app channels for existing ANZ consumer customers. The bot consists of over 60 intents and handles 45% of all user interactions.

• Analysed Existing Conversations: Conducted an in-depth analysis of existing dispute-related conversations to identify friction points and deflection opportunities.

• Agent Observations: Conducted side-by-side agent observations in ANZ's Manila contact centre to understand the agent process and customer interactions.

• Designed 'Gather Details' Component: Based on the insights from agent observations, I designed a 'gather details' component in the chatbot flow. This component collects all the necessary information the human agent needs before escalation, significantly reducing the average handle time for agents.

• Developed Customer Education Flow: Created a new chatbot flow for 'dispute a transaction' that included multiple levels of customer education. This flow advised customers on the due diligence they should perform before lodging a dispute.

• Expectation Management: Managed customer expectations by informing them that submitting a dispute through the bank is not the best or fastest way to resolution, and that it's often better to contact the merchant first.
  

Results:

• 60% of users wishing to submit a dispute were deflected by the new flow

• The chatbot is now doing the equivalent work of 80 human agents, saving ANZ Bank millions of dollars a year

• Reduced Average Handle Time: The 'gather details' component has significantly reduced the average handle time for agents.

• This project significantly reduced the volume of dispute enquiries, improved user satisfaction by guiding them to quicker resolutions, and provided substantial cost savings for ANZ Bank.

Note: While the chatbot 'Lotti' covers over 60 intents and handles 45% of all user interactions, this case study specifically highlights the 'transaction dispute' use case.