ScotiaBot

A chatbot is integrated into Scotiabank's main page to resolve doubts and frequently asked questions that collect information later displayed internally on a dashboard where Scotiabank teams can measure the interactions and frequent topics consulted in the chatbot.

Web application dashboard and Integrated Chatbot

Solve doubts with a chatbot

Scotiabank wanted to improve the experience of customers who seek information about the problems they present for charges and clarifications in Scotiabank Mexico.

The people of Scotiabank who followed up on these topics provided us with the necessary information to create the dialogue of a virtual assistant or chatbot to resolve frequently asked questions and kick feedback on the information requested.

Example of a dialog node in Watson Assistant

My role in the project

As UX Designer, I was responsible for the conversation experience and dialog created in IBM Watson Assistant for the chatbot. I made a better user experience by rephrasing the sentences to have a conversation according to the assistant's tone, previously defined in a design thinking workshop.

I participated in designing the feedback dashboard interface that shows the results of the NPS of the bank and some questions about the experience with Scotiabot based on the Scotiabank design guidelines.

Constraints, processes, and activities

We were working hand in hand with the representatives of the accounts receivable and collection teams; we used the garage method to teach them how to develop MVPs in a project. We had a month to create the assistant and the web page where they could access a dashboard that would show the most frequent interactions and topics and customer feedback by year, month, subject, and feedback mood.

Together with them, we divided the information into topics and identified the intentions for the dialogue. I worked on the low-fidelity wireframes of the dashboard and the page where the internal team could consult Scotiabot feedback. The stakeholders reviewed this, and after feedback, I made changes to the dashboard wireframe.

InVision low-fidelity wireframes of the internal dashboard.

One of the main challenges we had was that in the first phase, the chatbot had been committed to responding only to information previously curated. During the refinement, we discovered that what would add value is the ability to connect to the database and provide information about user tickets.

Example of chatbot responding to an identified topic asking for additional details in order to give an accurate response.

To provide user ticket information, we would have to include more security and steps on the chatbot. This feature was out of scope; we couldn't develop this feature within the budget and time allocated for the first phase, so we added this feature to the backlog for future project phases.

We created the chatbot dialog; we would display the information collected on an internal site. The teams could see the topics most requested by customers, make comparisons, measure the number of cases from each team, review the feedback and take actions to improve the issues with the clients. This helps Scotiabank to identify problems faster and take action more quickly to improve its customer service, have a quicker response time, and enable customers to find information interactively.