Getting underwriters from inbox to a decision
Designed an email-ingestion-powered triage product so underwriters work their stack by priority and appetite, not date received.
- Role
- Product Designer
- Company
- Two Sigma Insurance Quantified
- Year
- 2020–2021
- Focus
- Enterprise · B2B · Data viz · Workflow
Overview
Two Sigma Insurance Quantified builds software that helps commercial insurance underwriters decide which risks to take. After building a bespoke product for a single client, the company moved to selling focused modules instead of one large platform, and I designed the first one.
Underwriters were getting more applications than they could work, in no useful order. Applications arrived incomplete, later forms contradicted earlier ones, and nothing indicated which ones deserved attention first, so the stack got worked by date received. The module pulls applications in from email, normalizes the data, and puts a priority score and an appetite check on every one.
When you get hundreds of applications, the hard part isn't assessing risk, it's knowing which ones deserve your time at all.
Team: 2 product designers, 1 product manager, 5 software engineers, 3 data scientists.
The solution
Put every application in one place, ordered by priority
Explorer sorts the whole stack by priority score and filters it by appetite, age, and brokerage, so the most urgent applications come up first and the ones outside the book never take up time.
Made the priority score something an underwriter could question
Every score opens into the factors behind it: how close the effective date is, which broker sent it, and where the risk sits. Underwriters can see why an application ranked where it did instead of taking the number on faith.
Filtered out the business they would never write
The appetite check rates each application against the lines of business and geographies the carrier actually insures, so submissions outside the book get flagged before anyone spends time reading them.
Made ingested data navigable instead of raw
The application view lays out everything pulled in from email and attachments, with filters that jump straight to missing fields or conflicting values. Asking the broker for what is missing takes one click.
Showed where every value came from
Each field opens a screenshot of the source document next to a confidence score, so an underwriter can check a value against the paperwork without leaving the application.
Let underwriters settle conflicting data in place
When a later form contradicts an earlier one, every competing value shows up together with its source, and picking the right one happens right there in the field.
Kept the team working the same application
Field-level comments let the underwriting team sort out questions on the application itself instead of moving the conversation to email.
What came of it
- Shifted the MVP from a bespoke build for one client to a modular product strategy, after market feedback showed customers wanted focused tools more than one large platform.
- Built a working demo in three months, from first explorations to something the sales team could put in front of prospects.
- The module secured the company's first $1M in commitments. The document ingestion and normalization underneath it drew more interest than anything else we offered.
Next case study
RelayRides to Turo rebrand →