BACK

Yiyang Zhang

19:02:17 UTC

JusticeText is a legal-tech platform that helps public defenders manage video evidence.

internship

0 words
0 m

JusticeText helps public defenders search, annotate, and organize hours of police video and transcript evidence. In summer 2024, I joined the team as a product design intern, where I focused on redesigning a critical but underperforming feature: the Notebook.

client

JusticeText

time

2024

role

UX Design Intern

mentor

Sharon Tung

client

JusticeText

time

2024

role

UX Design Intern

mentor

Sharon Tung

Notebook was one of the least-used feature on the platform.

Notebook is a key workflow within a larger product meant to support active case preparation. It was designed to deepen engagement during video review. When usage stayed low, it signaled that users weren’t fully activating the platform’s core value, impacting adoption, retention, and long-term product stickiness.

The original Notebook didn’t match how users reviewed video, or how they thought.

There was no clear place to start writing. Notes were saved in the order they were added, not in the order things happened. The list had no structure, no visual cues, and no connection to the footage itself.

Original Notebook Interface

But fixing it wasn’t simple. We couldn’t disrupt workflows for users who were already using the platform.
JusticeText already had adoption across public defenders’ offices. That meant we couldn’t introduce entirely new interaction models or change how users navigated review. The redesign had to work within existing mental models, focusing on UI clarity, structure, and affordance, without touching the core playback or annotation logic.

So I focused on the biggest points of friction.

Rather than reinventing how the feature worked, we looked for where it was actively getting in users’ way. Three issues stood out:

Lack of Clear Entry Point

The old interface offered little visual guidance. A line of text was the only prompt, leaving users without visual affordance and unsure how notes would behave or where they would go.

No Link Between Notes & Video

While notes were timestamped in the backend, the UI didn’t make that connection visible. Users had no feedback about what they were anchoring or how to return to a moment.

Notes Ordered by Save Time, Not Video Time

In the original Notebook, notes were listed in the order they were added—not in the order events occurred in the video. This broke the flow of review and made it harder to reconstruct a timeline of events.

From Notes to Narrative

To understand why the Notebook wasn’t working, I stepped through typical review sessions end-to-end. I recreated common workflows using internal accounts and observed how users interacted with the tool across real cases.

Start the video

Watch for relevant moments

Take quick notes mid-playback

Return later to scan, recall, or export

I focused on three key actions:

  • Taking notes mid-playback: Users often paused briefly or typed while video played, expecting notes to be saved accurately and contextually.

  • Returning to notes later: Many tried to skim through past notes to find a moment, but the lack of timeline order made it slow and unreliable.

  • Trying to reconstruct a timeline: In longer sessions, users couldn’t tell which note belonged to which part of the video. It forced rewatching or external tracking.

What I found was consistent: defenders used the Notebook to track what happened, not just what they were thinking. But the UI didn’t reflect that intent. Notes appeared out of order, looked identical regardless of source, and gave no indication of what they were linked to.

In short: users didn’t trust the Notebook to capture what they saw.

They didn’t know if a note was linked to the video, if it saved correctly, or where it went. What should’ve been a lightweight way to track key evidence felt more like writing into a black box.

That led to three clear design decisions:

Make the starting point obvious

The input field needed to be always visible, clearly labeled, and visually tied to the video. Users should know exactly where to type, what will happen, and when it’s saved.

Reflect the timeline, not the typing order

Notes had to appear in the order events happened—not when they were saved. This change would help users reconstruct a narrative and reduce rewatching.

Introduce structure without adding steps

Each note needed enough hierarchy to be scannable (timestamp, speaker, content), but without requiring additional tagging or setup. The goal was clarity without complexity.

Before & After

I focused the redesign on the areas that caused the most confusion during review. Here’s how the new design resolved them:

No Clear Entry Point → Labeled, Persistent Input

Before

After

Before

  • A single unlabeled text line

  • No clear typing indicator

  • Users weren’t sure where to start or if the feature was working

AFTER

  • A fixed, clearly labeled input aligned with playback controls

  • Supportive empty state graphic prompting users to add their first note

  • Immediate visual feedback for first-time and returning users

Saved Notes Out of Sequence → Timeline-Based Sorting

Before

After

Before

  • Notes were saved in the order they were added

  • Reviewing felt disjointed—important moments appeared out of context

  • Defenders had to rewatch video to understand sequence

AFTER

  • Notes are automatically sorted by their video timestamp, not when the user typed

  • Each note is anchored to its corresponding moment in playback

  • Reading the list now reflects the actual timeline of events

Structured, scannable note cards

Before

  • Flat list of text with no hierarchy or visual cues

  • Hard to skim or compare notes

  • Everything looked the same, regardless of content or context

AFTER

  • Notes are displayed as modular cards with timestamp, speaker, and content

  • Spacing and type hierarchy optimized for scanability

  • Easier to revisit, organize, and trust what was saved

Usage went up, because the friction went down.

JusticeText users could now take structured notes while watching video, confident that what they captured would be clear, organized, and easy to revisit.

  • Notes are now consistently ordered by video timestamp

  • First-time users immediately understand where to type and what will happen

  • Internal stakeholders noted improved clarity and better alignment with real defender & paralegal workflows

People weren’t ignoring the feature—they didn’t trust it.

This project taught me how small interface gaps, like unclear input or misaligned ordering, can block adoption entirely, even when the feature is critical.

By focusing on clarity, alignment with mental models, and reducing friction during note-taking, I helped reposition the Notebook as a tool defenders could actually trust.

Simple structure, shown at the right moment, can unlock complex thinking.

© 2025

by Yiyang Zhang

Pittsburgh, pa

20

°C

© 2025

by Yiyang Zhang

Pittsburgh, pa

20

°C