Smartcaller
2024
Improving the call experience of bankers with their clients by empowering them with useful insights
Smartcaller, a call co-pilot for bankers, utilizes AI/ML on live call transcripts and CRM data to offer real-time insights. This enables advisors to deliver accurate information seamlessly, without juggling multiple apps.
Skills
Research
UX & UI
User testing
Industry
Banking
Team
I led the research and UI/UX design while working with a team of 8 engineers and 2 product owners.
Problem background
The tech team launched a pilot for 15 users, providing live call transcripts. Users found it distracting and wanted to opt out. The product team collaborated with design to enhance the transcript's value during calls. My role was to research and lead product design.
Scope of work and timeline
The project included conducting research to understand current user journeys in order to design a new product that caters to user needs.
Research
User Experience
User Interface
User Testing
Research & analysis
We conducted primary research through 1:1 interviews and gathered information from 'Tell Us' data to understand current calling experience and related pain points
Design canvas
Design canvas, inspired by the business model canvas, offers a comprehensive view of a problem, encompassing customer needs, project requirements, and success metrics. It's a collaborative tool for product managers and designers to understand and address challenges effectively.
Understanding users
Meet Mary
Mary, a banker in London's partner coverage team, manages over 40 clients who frequently call her for various needs like portfolio updates, new opportunities, or trade placements. She gathers information from diverse sources to assist clients and connects them with specialists when needed. With a focus on client acquisition, Mary juggles unplanned incoming calls, requiring her to quickly adapt and respond.
User journey
Key user needs
01
Increase accuracy of Caller ID
Advisors need to swiftly identify callers to access their profiles and accounts in CRM. However, due to complex profile structures and data issues, caller ID accuracy is currently low.
"The caller ID only works 50% of the time. Most of the time its rather inaccurate or fails to identify caller"
02
Summarise previous call memos
A summary of the last five call memos can aid advisors in quickly recalling recent client discussions, serving as a helpful refresher between meetings. Given the time gap between touch points, a concise TLDR version is beneficial, as full memos are too lengthy for real-time phone conversation review.
"The call memos are too long to read during an ongoing call"
03
Reduce cognitive load
Accessing key information from multiple applications through smart and bite-sized insights will help advisors build context without opening multiple apps. This will help lighten the cognitive load by reducing clicks and the number of windows advisors must go through during each call.
"I usually go through transactions, positions, accounts, Profile 360 and Call memos to build a context on the client while engaging with them on call"
04
Make it easy to do post-call actions
Transcription technologies alleviate the burden of manual note-taking during calls, allowing advisors to focus solely on delivering accurate information to clients. However, advisors often forget post-call actions such as scheduling meetings or sending documents, particularly on busy days.
"On a busy day, it is almost impossible to post call memos; thus, I keep that task for the end of the day."
User Experience
The aim was to understand the existing user journeys and create a future journey that solves the pain-points and caters to the need of users.
Principles
Future journey
User Interface
Translating the future journey to a high-fidelity user-interface that will help users, engineers and product partners to understand the future state of the product
Concept 1
Based on a dashboard-like view, this concept shows all the information an advisor requires during the call to build context. The cards provide bite-sized information from the apps that exist in CRM,
Pros
-
Upfront and in the face
-
Minimal clicks required
-
Less navigation needed
-
Build a holistic picture of caller
Cons
-
Takes up too much space
-
High cognitive load
-
Busy interface
-
Scalability can become an issue
-
Not a mobile-first approach
Concept 2
A sidebar type of window can be pinned to the right or left side of the desktop. The landing screen focuses on identifying the caller and giving active insights about the call to the advisor.
Pros
-
Takes up less space
-
Light cognitive load
-
Mobile-first approach
-
Uses progressive disclosure
-
Scalability is easier
Cons
-
Increased clicks
-
Information can be hidden
Active insights
LLM on live transcript provided real-time insights from the call, eliminating the need for advisors to search through multiple apps for caller-requested information.
Profile snapshot
A quick snapshot lets advisors refresh their memory about the caller's profile. The advisors will no longer need to open a separate app called 'Profile 360' to get this information.
Call memo summary
"Utilizing the LLM-based summarization engine on past call memos significantly cuts down the time for advisors to comprehend past conversations with callers, leading to more informed discussions."
Transactions
Callers often call in to quickly check on recent transactions they made on their account. Having this information handy without opening a separate app makes advisor look smart and savvy,
Call summary
Providing a call summary at the end of each call alleviates the burden of manual note-taking for advisors, addressing a significant pain point. This allows them to prioritize urgent tasks without the need to worry about jotting down notes.
Outcomes
In January 2024, the product was released to 120 users across the US, EMEA and Asia. After six weeks, the feedback was gathered, and metrics were captured. The key findings included:
-
The caller ID accuracy improved by 80% by using new logic suggested by design and tech
-
Connect canvas sessions dropped by 46% during active calls
-
Callbacks reduced by 32%
-
Publishing of call memos increased by 24%
-
No advisor has asked to opt out of the pilot so far. Many advisors have told their colleagues about the new product, leading to more enquiries about the larger roll-out.
The product is now set to launch to all the advisors in the firm in a phased manner across all markets.
Learnings
"The product's inclusion of machine learning and large language models motivated me to deepen my understanding of these technologies, enabling me to pitch innovative ideas to the team.
Through this project, I reaffirmed the importance of user-centric technology in delivering value to both users and businesses. Its success granted me equal collaboration opportunities with product partners, business stakeholders, and engineers."
© Shubham Khatkar 2024