Go Back

Designing an analytics dashboard powering loan disbursals of ₹10L+ daily

Overview

Video KYC is a highly operational workflow where teams of 20–50 agents process thousands of verifications daily. Operations managers rely heavily on analytics dashboards to monitor throughput, identify bottlenecks, and ensure compliance.

The existing analytics experience at HyperVerge was fragmented - I redesigned the analytics system to provide a clearer view of the VKYC funnel, helping operations teams quickly identify drop-offs and understand failure patterns.

The existing analytics experience at HyperVerge was fragmented - I redesigned the analytics system to provide a clearer view of the VKYC funnel.

Team

1 product manager
1 product designer
2 engineers

Duration

2 month

My role

End to end design and handoff of the feature

Impact

Improved internal team efficiency by 10%

Lower TAT for custom metrics

Contributed to NPS of 8.2 for FY25

About the product

Video KYC (Know Your Customer) enables users to get access to loans by simply verifying the documents over a live video call with a VKYC agent. As part of enabling this operation, HyperVerge provides a suite of collection, reviewing and monitoring products to it's clients (SBI, CRED, Fampay,etc)

Users Involved

Operations Managers:

They are analytical in nature
Typically come from customer support backgrounds
Want as much control and oversight to their operations as possible

Operations Managers:

They are analytical in nature
Typically come from customer support backgrounds
Want as much control and oversight to their operations as possible

The problem

Our current analytics was a dump of numbers without enabling clear actionable insights across different steps of VKYC.

Operation managers would navigate around without direction to look for the numbers that will provide them the insight they are looking for.

Our current analytics was a dump of numbers without enabling clear actionable insights across different steps of VKYC.

Operation managers would navigate around without direction to look for the numbers that will provide them the insight they are looking for.

OLD EXPERIENCE

# Problem 1

Time to action was very high - with users having to make more than 6-7 interactions aftering finding a gap in their numbers

CLIENT REQUESTS FOR CUSTOM ANALYTICS

# Problem 2

Each client would want to see numbers in a different way - causing multiple custom requests

Why solve this?

#1

Opportunity to create differentiation from our competitors - in analytical capabilities.

Aligned with our targets to scale to 80K MRR

Opportunity to create differentiation from our competitors - in analytical capabilities.

Aligned with our targets to scale to 80K MRR

#2

Each custom client request meant dependancy on internal dev team.

High TAT (3-5 days) for clients

Each custom client request meant dependancy on internal dev team.

High TAT (3-5 days) for clients

The goal

Build an analytics dashboard that would enable operations managers to analyse metrics and understand points of improvement progressively

Key considerations:

Simpler IA to enable actionable inights
Enable configuration of custom metrics easily
Ensure respsonsiveness and compatiblity with chartjs

Build an analytics dashboard that would enable operations managers to analyse metrics and understand points of improvement progressively

Key considerations:

Simpler IA to enable actionable inights
Enable configuration of custom metrics easily
Ensure respsonsiveness and compatiblity with chartjs

The messy middle i.e the process

Insights from client visits, wireframes, going through other analytics platforms, documenting user scenarios, user flows and many dicsussions

🔴 Approach 1 - Overview Redesign

I tried redesigning the existing dashboards by combining related metrics together

This approach would prove to not very scalable.

Introduction of new metrics would mean new designs.

Back to the drawing board —>

🔴 Approach 2 - Getting closer

Breaking metrics into funnels across various steps.

On discussion with my developers - there were certain constraints with the graphs I had designed.

But we were getting closer to the final solution

Iterating with funnels and firming up the layout …

The solution

Intent based metrics

Now users can progressively disclose step specific data.

Intent based metrics

Now users can progressively disclose step specific data.

Ability to analyse user ID's for each metric

A common behaviour for the operation managers was to analyse individual cases to figure out what went wrong - we replicated that behaviour with filtered user ID's at each step

BEFORE

AFTER

The new solution reduced the required interactions for customer ID flow from 5-6 clicks and users having to recall steps to 2 interactions. Reducing time to action by 3-4 seconds

Enabling configurable graphs

We set up "building blocks" these were pre-made graphs components where our support teams could configure custom graphs for clients via events

DOCUMENTING COMPONENTS AND BEHAVIOUR OF GRAPHS

Impact

On a daily basis, clients like SBIC process loans of upto 10lakhs a day - with more insightful analytics, Operations managers were able to spot issues faster and increase loan disbursals

Improving internal effeciency by freeing up bandwidth of dev teams

Reduction in interaction cost and time to action by reducing 4-5 clicks to 2 clicks and making and time to action from 8s to 3s.

My learnings & Reflections

I would spend a lot more time prototyping the graphs to get a feel for interactions.

Learnt to consider responsiveness and scalability in design

Using interaction design to distill complex data points to insightful metrics