ADL.ai
Transcript &
Data Management
Data Management
Goal
Develop a platform to help facilitate the process of annotating transcripts with the customer’s in-house taxonomy tagging system. Users should also be able to view auto-generated summaries of transcripts and tag fellow SMEs.
Challenge
Starting the project with a blank canvas and a broad goal of creating a new transcription feature, I guided the strategic design process from 0->1. This involved crafting product strategy, conducting user research, concept testing, creating visual designs, and finally, developing prototypes.
Business Goals
Introduce a new product feature for the COE Research team within a year. Improve workflow efficiency by increased feature engagement.
User Goals
Standardize and streamline research workflows.
Improve analysis gathering and insight generation.
Discovery
The extensive discovery phase combined different interview and research methods to identify end user pain points.
1. We conducted 1:1 interviews with the analyst team to understand their existing processes and to identify process improvements.
2. Held online workshops with the product champion team to gain their insights about their day to day problems.
3. Synthesized our findings to craft our problem statement and socialize learnings with the customer’s technical and product stakeholders.
1. We conducted 1:1 interviews with the analyst team to understand their existing processes and to identify process improvements.
2. Held online workshops with the product champion team to gain their insights about their day to day problems.
3. Synthesized our findings to craft our problem statement and socialize learnings with the customer’s technical and product stakeholders.
Lack of a centralized platform to manage, store and analyze transcripts.
Reduced levels of collaboration.
Varying levels of tech proficiency and software preferences.
Disparate data storage and handling.
Non-standard research workflows.
Unclean data and content inconsistencies.
Defining the Core UX
After exploring various concepts, we mapped out the core experience flows and defined features for the minimum viable product experience.
1. Viewing and managing transcripts within the database
2. Uploading files for transcription.
3. Facilitate faster research by auto-generated and annotations.
User Testing
Prototyped mid-fi of brainstormed concepts and presented them to users in order to determine what solutions users prioritize and want.
1. Start from a higher level of confidence: users wanted the tool to have a first pass at markup using their in-house taxonomy tagging system.
2. Easily link transcripts to existing records such as POIs, events and groups. Create new instances when no existing matches are available.
3. View comments, tags and notes on a transcript in one place without the need to shuffle tabs.
1. Start from a higher level of confidence: users wanted the tool to have a first pass at markup using their in-house taxonomy tagging system.
2. Easily link transcripts to existing records such as POIs, events and groups. Create new instances when no existing matches are available.
3. View comments, tags and notes on a transcript in one place without the need to shuffle tabs.
After each session, we shared a SUS to further understand what technical improvements could be made.
Final Designs
After several iterations and rounds of user testing, we shipped the first version.
Transcripts Overview and Search
Transcripts Page
Database Manager
Outcomes
The first version of this tool was released to the team in March 2023. After a quarter post-release, the team reported an overall increase in the ease of bulk data management, increased insights generation and workflow efficiency. Stakeholders also reported increased collaboration, reduced errors and lowered costs of training.
