As of August 2020, our Chase Digital Assistant rolled out to 106,000 customers with a 13% decrease in call volume. Since our digital assistant is still very new, we’re constantly refining and revising.
Transaction search is an essential task for our bank users. Currently, we show pending and posted transactions for demand deposit accounts, card, and investments, with drill-down disambiguation if there are multiple accounts.
PARTNERS:
Product
Design
Research
Tech
BSA + scrum team
Legal
RESEARCH (MAY 12-13 AND JUNE 9-10 2020):
Due to Covid-19, we held virtual research sessions via UserZoom. We sought participants with multiple accounts (checkings, savings, and credit cards) and validated use cases around five specific transaction types: refunds, deposits, merchants, categories, and time-based spend summaries. We also looked for feedback and reactions to content and design elements with four prototypes.
Most participants view spending through weekly and monthly brackets. The widgets were well-received, because they want visuals of their spending progression. Participants also responded positively to the CDA repeating all entities mentioned in their initial utterance; explicit confirmation is best practice. Participants also chose prototypes with as less taps as possible.
See my research presentation deck from July 1, 2020 for more insights. I presented this to the overall content team. Without these important research developments, it’s nearly impossible to build a true digital assistant.
CONTENT SPREADSHEET FOR CHATBOT REPLIES (0-26+ TRANSACTIONS WITHIN THE PAST 90 DAYS)
I captured any and all possible scenarios and replies in this spreadsheet. For example, a user with a checking and investment account and would like to see pending transactions for both via a merchant.
USER REFINES BY MERCHANT WITH A CHECKING AND INVESTMENT ACCOUNT:
