Opportunity

The value proposition Engage provided of pairing customers with their own personal banker was a compelling story of excellent customer service. However, the product scaled in a way we did not expect, leading to a decline in customer service quality.

Receiving both agents’ and customers’ feedback, we focused on helping bankers manage their customers' service response time expectations.

Impact

Even though I wasn’t around for all the results of these feature releases (since I had left the company), I was able to observe some benefits:

• Improved app reviews: shortly after the initial feature releases, the financial institutions we worked with saw an uptick in positive app reviews around timely agent service and responses.

• More research opportunities: working with financial institutions can be tricky because we had to balance providing agent and customer value with maintaining a positive relationship with their leadership. Though it was an uphill battle to conduct research and receive feedback, we were able to open the gates to more research opportunities by bringing their leadership into our design process.

• Increased agent engagement: since the majority of our solution revolved around improving the agent platform experience, we received positive feedback on how their daily customer engagements improved.

Going to the source

When we began this project, we were only able to frame the problem based on feedback from one bank: they were receiving customer complaints about their service times.

Ethnographic studies

To better understand the problem, we took this as an opportunity to really dive into research. We visited two financial institutions to observe for ourselves how bankers typically work. Additionally, we were able to interview bankers and view anonymized conversations to synthesize qualitative data around bank agent and customer issues.

Key findings

• Even the simplest customer requests take effort: Many bankers have additional tasks and roles in addition to servicing customer chats, like answering phone calls, updating customer data, etc.

• Agents were struggling to juggle customer requests: agents would read through conversations and then handle more urgent requests. This led to conversations being forgotten since those conversations would be marked as read.

• Customers just wanted to be seen: it wasn't so much about speed, but customers just didn't know when they would be helped since agents weren’t responding.

Defining success

Based on these findings, we saw that in addition to being able to better manage customer expectations, we needed to find solutions that would enable bankers to better meet customer service expectations. We measured our success based on our ability to:

1. Decrease customer service complaints

2. Decrease average follow-up time

Managing (+ meeting) customer expectations

I focused on the problems that our product could solve based on the most urgent customer needs, scope, and product fit.

I knew it would be important to closely collaborate with our developers to understand the technical feasibility of certain solutions. Since we wanted to release solutions quickly, keeping the scope to a minimum was essential.

Finding the sweet spot

Working within the design system

Many of the iterations I proposed were focused on component updates. Having an understanding of how to work within the design system paradoxically gave me the freedom to iterate quickly and effectively.

small updates, big impact

Our solution comprised of introducing a series of new features and improvements:

• Conversation list sorting: agents weren’t always responding to conversations in chronological order, so we wanted to give agents the ability to order their customer conversations based on contextual service prioritization.

• Tags: surfacing key customer identifiers enabled agents to prioritize their customers without having to open conversations.

• Mark unread: something we introduced directly based on our ethnographic studies, agents could mark when to revisit a conversation by marking it unread.

• Status updates: to alleviate the stress of being unable to answer customer requests in a timely manner, agents now had an additional way to globally communicate their situational context to their customers.

• Typing indicators: because agents would carefully craft detailed responses to their customers, introducing typing indicators provided customers some comfort that they were being taken care of.

A phased release

Defining our solution as a feature set gave our team the benefit of strategically releasing these features to minimize risk and maximize our learnings. I worked closely with our product team to define a feature rollout that would provide the most benefit to agents and their customers while managing technical impact.

Takeaways

There’s never a clear path to designing excellent solutions, especially with such an ambiguous problem of achieving a 'good’ response time. Working through this design process has taught me:

• Design with the system in mind: what I originally perceived as a constraint became an enabler. Being able to consume and contribute to our design system gave me the opportunity to more deeply collaborate with our developers.

• Gather feedback from ALL involved parties: in addition to designing for our agents and customers, we also needed approval from financial institution leadership, which actually strengthened our solutions since we had to take their considerations and constraints into account.

• Define success clearly: reflecting on this design process, I think we ultimately achieved our goal. However, if we were able to define our success metrics in a quantitative way, I believe that would have resulted in better, innovative ideas.