Joining the program, I was surprised by the absence of customer research. However, owing to an elevated risk of insider trading within the Treasury division, the bank’s compliance team isolated the staff from customers. Only specific teams were permitted to interact with the outside world.

One of those teams was the help desk, but interviews proved impossible with their workload. Instead, I studied their monthly reports and found that most calls were instructional requests and not technical issues. This insight was curious because every function within the app contained a link to a tutorial microsite.

Traffic analytics of the tutorials showed little searching or browsing. Users entered a specific topic page from a help link, dwelled a few seconds, and then dropped out of the application (presumably to call the help desk). Apart from being a content issue, the question remained as to why the app was so difficult for users.

I interviewed the Client Champion (or CC) team, an advocacy group that trains a customer’s staff on the app when they sign up with the bank. The CCs showed how they segmented customers into three groups (high, medium, and low) based on the volume of transactions. They focused their training program on the medium group because the high group favored the API over the app, and the low group was only casual users.

Without data to show otherwise, the UX team presumed the archetypical user was working in the application extensively and expected a sophisticated UI — a superuser. But my research with the CCs disabused anyone of this notion.

The high group only interacted with APIs through the bank’s treasury management systems. They never saw the web interface and were not the source of the problem, nor was the medium group, who were highly compartmentalized teams whose duties relied on a single app function and required little instruction.

The problem rested with the low group. These were companies where only one or two finance people were responsible for all of the tasks in the app. The complexity of the interface overwhelmed them, and because their use was intermittent, retention of instructions was low. Given that the CCs paid the low group little attention, they dominated the help desk.

The imminent transition of SMB customers to the app set the stage for a help desk disaster. Like the low group, the mom-and-pops that comprised the SMB group didn’t need the level of intricacy the app offered. However, finding a solution for the low group would mean a solution for these newcomers. It was two birds, one stone.

Using what I gathered from the CCs, I developed models to represent users with a low understanding of the app. Without access to actual customers, I simulated cognitive walkthroughs with each model, which wasn’t a stretch, considering my own newness to the app.

It was apparent the app would need an overhaul, but my effort was limited to what could be done within the application as it was. As such, I developed a set of experience strategies based on the needs of each model as they encountered tasks requiring help.