UX RESEARCH / CLAIMS

UX RESEARCH / CLAIMS

UX RESEARCH / CLAIMS

Claims Recovery Research

Claims Recovery
Research

Claims Recovery
Research

The brief pointed toward automation. I led the research to learn how recovery work actually happened before the team chose what to build.

The brief pointed toward automation. I led the research to learn how recovery work actually happened before the team chose what to build.

The brief pointed toward automation. I led the research to learn how recovery work actually happened before the team chose what to build.

RESEARCH ROUTE / RECONSTRUCTED

The work before
the build.

The work before
the build.

The work before
the build.

STARTING POINT

Automation proposal

RESEARCH

Interviews and surveys

FOLLOW-UP

Disputed accounts checked

ROLE

Lead UX Designer & Researcher

FOCUS

Recovery and subrogation in Motor Insurance

METHOD

Interviews, surveys, process map

STATUS

Completed · shared with leadership

1

THE CALL

Make room for discovery

Make room for discovery

Make room for discovery

The partner had explored automation and agentic AI, but the proposal said little about day-to-day recovery work. I worked with Product Management to make the case for discovery before choosing a solution, then led the study with a four-person team.

The partner had explored automation and agentic AI, but the proposal said little about day-to-day recovery work. I worked with Product Management to make the case for discovery before choosing a solution, then led the study with a four-person team.

The first window was tight and handlers were hard to reach. We began with subject-matter experts while recruiting handlers and leads. Auto/Motor claims took priority.

The first window was tight and handlers were hard to reach. We began with subject-matter experts while recruiting handlers and leads. Auto/Motor claims took priority.

RECONSTRUCTED METHOD / NO PARTICIPANT MATERIAL

The guide changed as we learned.

PAIR SESSIONS

A moderator and observer interviewed together. Both took notes.

EVENING DEBRIEF

The two pairs compared surprises and contradictions.

NEXT-DAY GUIDE

We adjusted questions for the following sessions.

The guide changed as we learned.

PAIR SESSIONS

Moderator and observer interviewed together.

EVENING DEBRIEF

The pairs compared surprises and contradictions.

NEXT-DAY GUIDE

We adjusted the next sessions.

THE BRIEF CHANGED MIDWAY

Stakeholders narrowed the brief to recovery and subrogation. We limited the findings to relevant sessions and paused for an interim readout.

Why the surveys came later

The interviews gave us the process language to ask sharper questions. One survey checked whether more handlers recognized the patterns; another sought operational metrics from managers and leads.

2

THE CHECK

When the accounts did not match

When the accounts did not match

When the accounts did not match

At the interim review, operations leaders challenged parts of what handlers had told us. They described a more standardized, better-supported process and pointed us to people who might clarify the difference.

At the interim review, operations leaders challenged parts of what handlers had told us. They described a more standardized, better-supported process and pointed us to people who might clarify the difference.

That disagreement changed the research plan. We returned to original participants with narrower questions and used the handler survey to check whether the same pattern appeared more widely.

RECONSTRUCTED DECISION PATH

A disagreement became the next research question.

FOLLOW UP

Original participants repeated their accounts when we asked narrower questions.

SURVEY CHECK

The handler survey supported the pattern we had heard.

REPORT

We recorded both accounts and the evidence checked, without assigning a cause to the difference.

Public method diagram. No client or participant evidence shown.

A disagreement became the next research question.

FOLLOW UP

Original participants repeated their accounts.

SURVEY CHECK

The survey supported the pattern.

REPORT

Both accounts and checks were recorded.

The report recorded both accounts and the evidence checked. The study did not establish why those accounts diverged.

3

THE HANDOFF

Make the evidence easy to check

Make the evidence easy to check

Make the evidence easy to check

Four researchers were interviewing, analyzing, and writing an interim report at the same time. I built a redacted source trail so the growing material stayed usable.

Four researchers were interviewing, analyzing, and writing an interim report at the same time. I built a redacted source trail so the growing material stayed usable.

That trail mattered after leaders challenged the handler account. We needed to test what people had said in context, then show the evidence behind a finding.

RECONSTRUCTED METHOD / NO SOURCE MATERIAL

From interview record to report claim.

This reconstructed path shows how we moved from sensitive interview material to a finding in the report, with a researcher source check between them.

Organize and redact

I grouped transcripts by session and role, then removed identifiers locally before working with the material in Claude.

Retrieve and verify

Claude through AWS Bedrock helped locate relevant passages with participant codes and nearby context. Researchers checked the material against the source.

Connect in the report

The Figma report set each insight beside its pain point, supporting evidence, relevant metric, and recommendation.

WHEN A FINDING WAS CHALLENGED

We could reopen the passage and its context before finalizing the report. Claude located material; researchers decided what it meant.

Reconstructed workflow. No client or participant material shown.

From interview record to report claim.

ORGANIZE

Group and redact source material.

VERIFY

Find passages, then check context.

CONNECT

Link evidence to findings and recommendations.

Claude through AWS Bedrock helped locate passages. Researchers checked the source and decided what it meant. I revised the final wording and figures with the team; we completed the report and shared it with leadership.

INITIAL DIRECTION / NOT A SHIPPED DESIGN

A unified workspace, guidance where decisions happen, and less manual retrieval and entry. Handlers would still verify and decide.

REFLECTION / RESEARCH PRACTICE

What I’ll carry into the next study

What I’ll carry into the next study

What I’ll carry into the next study

Claude made our interviews searchable while four of us interviewed, analyzed, and wrote in parallel. I still had to protect participant details, keep context with each passage, and judge whether it supported a finding. I now treat a checkable evidence trail as part of the design work.

Corporate research rarely follows a neat sequence. Access came in bursts, the brief narrowed, and we owed stakeholders a readout while still learning. The source trail helped us adapt and stay accountable to what people said.

WHAT I’D CHANGE

Start the checks earlier.

Next time I’d request operational metrics earlier, start tagging and redaction checks from the first session, and test the search method with the team sooner. I’d protect more time for synthesis before the readout.

Interviews would still shape the handler survey. They gave us better language for the questions.

DELIVERY

Research complete.
Report shared with leadership.

Research complete.
Report shared with leadership.

Research complete.
Report shared with leadership.

COMPLETE

Fieldwork, current-state map, and final report.

SHARED

Findings and initial recommendations shared with leadership.

NOT CLAIMED

A shipped design or measured recovery outcome.

Client and participant details are omitted under NDA. All method visuals are reconstructed.