Alex Bai
All selected work

JPMorgan Chase / Mixed-methods research

Turning a low score into a useful decision.

A two-week satisfaction study that connected survey feedback to reliability, workflow control, and engineering priorities.

My role

UX Researcher

Context and timing

2 weeks · Q4 2025

Contribution

Four recommendations adopted into the backlog

102survey respondents
2,935active users invited
4backlog priorities adopted
Illustration of feedback being organized into product priorities
Harmony CSAT project cover illustration from the original portfolio. View full size ↗

The question

Which experience problems should the team act on first?

Harmony supported enterprise governance workflows across multiple user roles. A recurring platform-health survey created an opportunity to connect satisfaction and effort measures with existing feedback from support channels, in-app comments, and interviews. I designed, deployed, analyzed, and presented the study within two weeks.

Survey designCSAT and effort measuresComment analysisStakeholder prioritization
Recruitment and participation: invitations, completed surveys, comments, and follow-up opt-ins.
Recruitment and participation: invitations, completed surveys, comments, and follow-up opt-ins. View full size ↗

My approach

How I worked through it.

01

Define the audience precisely

Partnering with data and product teams produced an invitation list of 2,935 active users and a survey spanning six primary personas. One hundred and two people responded.

02

Segment without losing the sample context

I reviewed satisfaction and effort by role and business area, then analyzed open-text responses. Around two-thirds of respondents provided qualitative feedback, creating useful leads for deeper inquiry.

03

Connect findings to engineering’s priorities

I compared survey themes with earlier support and research evidence, then facilitated prioritization around reliability, data loss, workflow control, and audit needs.

Qualtrics dashboard snapshot with 91 responses. The final readout below reports 102 responses; the two views reflect different reporting snapshots.
Qualtrics dashboard snapshot with 91 responses. The final readout below reports 102 responses; the two views reflect different reporting snapshots. View full size ↗
The analysis workbook connects open-text feedback with themes and severity assessments.
The analysis workbook connects open-text feedback with themes and severity assessments. View full size ↗

The sample behind the score

102respondents
2,935invitations
~3.5%response rate
Useful signals for prioritization and follow-up. The response rate limits how confidently findings can be generalized to all active users.

Evidence to decisions

What mattered. What it changed.

Reliability was an experience problem

Loading failures and lost form data made the platform difficult to rely on.

Design directionPrioritize autosave and recovery alongside performance improvements.

Users needed control over their own workflows

Participants described difficulty withdrawing mistaken requests or understanding stalled work.

Design directionAdd withdrawal options and clearer status visibility.

Aggregate scores concealed role differences

Feedback varied across personas, highlighting needs that the dominant application-owner workflow did not cover.

Design directionUse role-specific follow-up research before generalizing small subgroup results.

Final readout: 102 responses, with the top-box definitions used for satisfaction and effort scores.
Final readout: 102 responses, with the top-box definitions used for satisfaction and effort scores. View full size ↗
Participant feedback grouped into opportunities around reliability, navigation, workflows, and data.
Participant feedback grouped into opportunities around reliability, navigation, workflows, and data. View full size ↗
Satisfaction and effort scores by line of business, compared across survey waves.
Satisfaction and effort scores by line of business, compared across survey waves. View full size ↗
Effort scores over time, including differences between user roles.
Effort scores over time, including differences between user roles. View full size ↗

Contribution and outcome

Four recommendations adopted into the backlog.

The study informed four priorities adopted into the engineering backlog: autosave, workflow withdrawal, deduplication, and export. Sixty-six respondents opted into follow-up research, giving the team a concrete path from the survey to targeted discovery.

How to read the outcome

102 responses out of 2,935 invitations is approximately 3.5%. Findings describe respondents and carry nonresponse and small-subgroup risks. Backlog adoption is documented; shipped improvements and subsequent satisfaction gains are not claimed.

Six satisfaction drivers used to frame product priorities.
Six satisfaction drivers used to frame product priorities. View full size ↗

Reflection

Triangulation strengthened the explanation of recurring problems, but it did not make the sample representative. The right use of this study was to prioritize follow-up and identify actionable friction.
Let’s talk through the work

Next case study

Harmony journey mapping