Underwriting — Feature design

Final user interface for the underwriting feature

About

  • Karat Financial is a bank for creators. Karat provides financial services, one of which is a credit card.
  • Underwriting is the process of determining a person’s credit limit. Usually banks look at financials, employment, and credit history to determine that limit. At Karat, we also take into account the applicant’s social media presence.

Problems

We have an internal tool, where our risks analysts can view and record applicants’ underwriting information. Unfortunately, it wasn’t very user friendly, so we decided to redesign it.

After talking with our analysts, Rithika and Cynthia, we identified 3 main problems:

No big picture

If the applicant linked multiple banks, it was impossible to see the overall financial picture (you had to toggle between the banks).

Finances section of the undewriting, which shows two dropdown for filtering data: one based on income type and one for bank accounts
Why can't I see these at the same time?

Lack of clarity

It was difficult to understand the reasons behind drastic changes (which are common for creators) in their financial history.

Two examples of finances section, one displays history graph for the past 3 months and one for the past 12 months
What are these spikes? Brand deals? YouTube revenue? Are these typical?

Manual work

All work for examining socials had to be done manually. It was repetitive and took a significant amount of time.

Socials sections of the underwriting that lists all social media accounts, basic info (like followers), and an automatically calculated engagement score
Is this level of engagement common for this type of creator?

Solutions

We added a way to display information from multiple banks at the same time. Also, an option to filter different accounts and transaction types.

We added a way to filter history by the past 3, 6, and 12 months. This was the most feasible solution technically and gave clarity into any spikes or dips.

We surfaced more social data on the underwriting tab and a faster way to get to each account. We also started exploring automated ways to calculate engagement. While it wouldn’t be replacing human evaluation, it would be a better starting point for the analysts.

This Figma file includes full flow, components, and explorations we did for this project:

Results

  • Together with other process improvements, we decreased the time it takes to underwrite each applicant.
  • More clear insights led us to lower delinquency rates.
  • And of course, it became much more pleasant to use the internal tool.