Open Card Sorting Exercise for BMO's Chatbot
Background
BMO's Digital Messaging & Collaboration team struggled to sort 211 intents in our internal chatbot master spreadsheet.
Intents needed proper categorization, and there needed to be naming conventions for topics.
Challenges included not having access to Optimal Sort (provided by the UX research team), identifying related intents, creating new intents that don’t overlap, and breaking topics into slots (for combining intents).
Solution
I was tasked with organizing the intents manually into topics, but intents that could fit under multiple categories.
After some discovery work with the UX director and product owner, I led an open card sorting workshop with 9 participants from the BMO UX retail team.
​
​
​
Participants categorized them into 15 groups for consistency.
Exported topic names created by participants and exported to excel. Identified duplicates (21).
Identified 18 unique topic names (3 removed because they are synonymous).
Transferred topic names into FigJam.
Sorted FigJam intent stickies (the ones with 8 matches) into visual representation categories based on these topics.
Outcome
18
topics identified
6
similar intents discovered
200+
intents sorted into topics
40+
intents grouped (for slots)
-
e-Transfers
-
Mortgages
-
Security
-
Alerts
-
Reactions
-
Bill payments
-
Cards
-
Account management
-
Other
-
Account balance
-
Taxes
-
Loans
-
Deposits
-
Disputes
-
Account transfers
-
Fees
-
Investments
-
Support
Areas for improvement
-
More participants and time for broader and accurate insights.
-
Asking participants in advance to name categories.
-
Removing "HowdoI" from intent names that are not proper categories.
-
More time and follow-up research to explore the “Other” category since it’s very broad and doesn’t scale well.