PROJECT

Redesigned internal data pipeline: reduced processing time by 67%

AIIMDB is an internal tool that utilizes Directus to hold all data for the USC Annenberg Inclusion Initiative.

TIMELINE

Oct 2024 - Jul 2025

ROLE

Product Manager, UX Designer

CLIENTS

USC Annenberg Inclusion Initiative

LOCATION

Los Angeles

DATA & PROCESS

Project Overview

As part of the Annenberg Inclusion Initiative, the leading think tank studying diversity in entertainment, I led a redesign of our data collection workflow for capturing gender and race/ethnicity of individuals in film and television. Our original process was heavily reliant on Google Sheets and Google Docs, creating inefficiencies, data inconsistencies, and bottlenecks across teams.

Drawing from direct management of researchers and coordination with our data management team, I applied user-centered design thinking to reimagine a more efficient, centralized, and empowering system using our headless CMS (Directus).

My Role

UX Researcher: Identified friction points through embedded management
UX Designer: Designed the workflow structure and interface protocols
Product Manager: Led rollout, hired and trained students, documented SOPs

This case study exemplifies how user-centered design can optimize internal workflows, empower teams, and improve data fidelity in high-impact research environments.

The Problem

While our previous process collected nearly 40,000 records, it was hindered by critical workflow issues:

WORKFLOW DELAYS AND WAIT TIME FOR RESEARCHERS

Lack of research assistant autonomy led to wait times and inconsistent assignments

QUALITY CONTROL ISSUES IN GOOGLE SHEETS

Disconnected evidence documentation stored in Google Docs and not tied to internal database

HEAVY RELIANCE ON DATA MANAGEMENT TEAM

Dependence on data team to scrape and upload Internet Movie Database (IMDb) data in chunks

Field User Research

Over six semesters, I supervised a rotating team of 10 to 24 research assistants per term during a 2 year data collection effort on Oscar nominees and film/TV crew. I gathered user data through daily check-ins, troubleshooting, and in-depth interviews.

DAILY CHECK-INS

TROUBLESHOOTING

IN-DEPTH INTERVIEWS

Identifying Pain Points

Usability Issues

Research assistants were confused by column headers and duplicate versions of Sheets
Evidence links were often misplaced, undocumented, or non accessible
Several people working on a Google Sheet at once decreased the load time significantly which led to accidental crossover and deletion in work
Research assistants had no context of the entire process which led to frequent mistakes

Internal Team Issues

Data management team spent significant time exporting IMDb scraped data and manually inputting into Google Sheets
Project lead was responsible for maintaining sheet structure and usability
Project lead had to track down former students for access and fix broken evidence links
Data management team parsed data for inclusion criteria and manually uploaded data to Directus which was costly in time

Design Goals

Based on my observations and feedback, I determined the following user and system goals:

1. MINIMIZE DEPENDENCIES

Students shouldn’t have to wait on scraped data to begin work

2. CENTRALIZE WORKFLOWS

All data entry and evidence logging should live in one system

3. SUPPORT AUTONOMY

Students should be empowered to collect, judge, and input data

4. ENSURE CONSISTENCY AND QUALITY

The interface should guide users toward structured, high-quality submissions

To address the above challenges, I conceptualized a new process

I created a flow diagram to visualize a new process that would give more autonomy to research assistants and remove the need for time consuming internal team intervention.

I led the implementation of a new model: the Production Tracking Team.

I created and trained a specialized group of student researchers who could independently manage the entire data collection pipeline. These students were responsible for scraping IMDb data by film (e.g., theatrical releases in 2025), filtering credits based on our internal inclusion criteria (such as department heads), and entering individuals directly into Directus with complete demographic information and accompanying evidence.

To support this new workflow, I wrote a comprehensive SOP and worked with our developer to introduce several UX improvements. Production Trackers were provided their own account to create, edit, and track their entries. Each entry included its own evidence section to store evidence directly in the database. This approach gave students full ownership over the process from start to finish, eliminating the need for middlemen and dramatically increasing efficiency and accountability.