Improving all metrics by creating a connected customer service journey with conversational AI
Company ABN AMRO | 2022 Type Product development
Chatbot + mega menu component combined
The 'Service and Contact' page had always been significant touchpoint in the customer service journeys, balancing between customer value and costs efficiency. Our team created the right momentum to experiment with conversational AI on this page and validate if we could assist more customers. The result was a significant increase in adoption, the number of customers helped, and a reduction in contact costs.
Situation
Over several years the contact page was highly optimised and the embedded mega-menu helped many customers find answers to their questions.
On to a good start
The first victory was to get support from two Tribe leads to collaborate and approach the page holistically.
A few months before the project we had done a broad examination of a wide group of retail customers, to understand their needs and behaviour regarding self-service information and contacting support. This sparked a discussion between two tribe leads as they both saw an opportunity to work more efficient in terms of staff and costs. As a result, we got a clear assignment to collaborate and look for optimisations from various perspectives.
Key insights from the research
Chatbot Anna
The chatbot is hardly considered a self-service solution and past experiences of other chatbots prevent customers from trying the one from ABN AMRO.
Two competing solutions
Both the chatbot content and the one in the mega-menu provide almost the some content, but in a different interaction.
Navigation components
Navigation and mega-menus remain a popular way find information, for most situations this requires little cognitive load from the customers.
High drop-out in the Mega-menu
A substantial amount of customers (30%) drop out of the mega-menu, not reaching any content.
Opposite of seamless
Moving from one self-service solution to the other lacks a good transition. This is making customers feel they are starting from scratch.
Design inspiration
Most other companies position their chatbot with ‘answer suggestions’ and a separate component to navigate through all information.
Ideation
Quickly sketching out various possibilities to understand the opportunities, riskiest assumptions and design decisions.
Reposition chatbot to the start the customer journey
Do we need to keep the chat UI in order for customers to intuitively use the chatbot
How desired is a navigation component in the interaction with the bot
Bringing it down to two main concept directions.
Concept 1
In this concept, the chatbot and a mega-menu are integrated. All the important information is aligned on the left, as this follows a universal pattern for how people scan a page. Additionally, there is little visual feedback on the presence of the chatbot.
Concept 2
In this concept the chatbot is positioned prominently on the page, with a few suggestions what customers can start a conversation about. This design was inspired on the layout of other Dutch companies.
Customer research
With the goal choose a winner for an AB-test, we wanted to build confidence by doing a few rounds of user testing.
Round 1: Choosing a concept direction
Round 2: Understand customers’ experience with mobile browsers
UX researcher leading the remote interviews.
Figjam where various colleagues collaboratively took notes of the sessions.
❓Mixed feelings of customers how to show the chatbot. From unnecessary to more transparant/honest.
Main insights
✅ Customers were the most positive about concept 1, as it directly provides various options to find information at the of the page (above the fold).
✅ The transition from the mega-menu to the chatbot was new but didn’t disturb customers to continue.
❌ The MVP for mobile browsers caused some usability issues, so we decided to exclude these from the AB-test.
Experiment setup
For the experiment we preferred speed and gather qualitative insights as quickly as possible.
Although there was a preference in the team for the design in experiment 2, we chose a different design for the MVP. The design in experiment 1 could be built more quickly, and customer interviews suggested that both approaches could yield positive outcomes.
Experiment 1: Quicker to build
Experiment 2: Preferred design
Results
The new design improved all of the key metrics and was considered a great success.
Engagement ↑
20% increase in engagement with the self-service section.
Chatbot conversations ↑
5x more chatbot conversations, leading to more customers being helped.
Customer satisfaction ↑
Positive increase in customer feedback, and fewer negative comments
Fewer chats and calls ↓
24% reduction in handovers to live chat, and 6% fewer phone calls after visiting the page.
Cost reduction ↓
Reduces contact costs by €367k on a yearly basis
Key outcome
More important than the results was the insight into the new strategic position of the chatbot.
Would like to know more?
For ABN AMRO
Year 2022 - 2023
Objective Improve customer experience of self service and contact journeys and look for opportunities to work more efficiently and reduce costs.
Role Broad range of product development responsibilities; understanding customer needs and expectations, designing solutions and refining them with devs, facilitating the design process in the team and setting up experiments.
Collaboration UX Designer, Product owner, Business developers, Optimalisation specialists, UX researcher.