Research Tooling Suite
Lead Developer
2018 - Present
In 2018, my team and I started working with Research Ops to build out a suite of tools to support our research process. Today, this application consists of three encapsulated applications offering lab video analysis, participant recruiting and management, and repository of studies and insights.
At a glance
8278 videos - indexed and transcribed
361 studies
414 participants
Our needs
Although each of our components had niche needs to fulfill, they all shared common business and user needs.
Security
At Workday, we handle a great deal of sensitive information, not only PII but often confidential business data. We wanted to ensure that recordings and personal data didn't leave the walls of our secure infrastructure in AWS. This meant providing a robust feature set so that researchers wouldn't need to export data to third parties or maintain a separate copy of data.
Governance and compliance
Top of mind as we were preparing for a higher volume of participants was safeguarding their PII and allowing for swift purging of personal data upon request. To respect our privacy policy, we needed to purge recordings and participant entries periodically.
Centralized and seamless
From recording a session to analysis, we wanted to provide a seamless experience for our researchers through one, centralized system to reduce context switching. We wanted to track and archive every artifact produced as part of the research process as well as a complete history of research participation.
Video analysis
The first piece of our suite was born out of natural needs as we started ramping up more robust support for our UX research labs globally and inviting more participants.
Automatic uploading and streaming from labs
We built an integration with the Pearl Epiphan devices in our labs to automatically upload recordings to S3. We then processed the MP4 video through MediaConvert to prepare them for HLS streaming in the client. The media files were then served through S3 and CloudFront and played in a modified Video.js player.
Transcription analysis
Upon being uploaded from either Zoom or our labs, the videos also get transcribed through AWS Transcribe. Transcriptions can then be accessed alongside the video player, highlighted following the time track. Researchers can highlight sections of note.
Library organization
We implemented faceted search using ElasticSearch to sort through the thousands of videos produced as part of our research (8000 videos since the inception of the project). Researchers could also assign videos to themselves for later reference.
Participant management
Prior to 2019, Workday primarily performed research on our own employees or an exclusive set of partners. With the addition of a Research Ops Program Manager, we had resources to start supporting inviting the public to participate in our research. However, we also needed infrastructure to support the long term volume of participants and participation in our process. We completed interviews with our researchers to determine their specific needs.
Marketing website
We used Gatsby and Workday's Canvas Kit to build research.workday.com. This site primarily served as a landing page to then invite interested participants to complete our SurveyGizmo survey.
SurveyGizmo integration
Once a participant completes the survey, the data gets ingested via a AWS Lambda function that sanitizes and records it into a AWS RDS database.
Robust filtering
One of our key value propositions is to allow researchers to quickly and easily find participants that align with their criteria. We determined the most commonly used criteria from observation and interviews and built powerful filtering for them.
History & tracking
As researchers complete sessions with participants, we track each stage of their participation as well as any incentives that are issued. All researchers are able to see participants' statuses.
Policy enforcement
We automatically remove participants from our pool based on contact and incentive policies we have in place. For example, participants are removed from the pool based on a certain amount of incentives received and we don't want to contact participants if we've had contact within the past few weeks.
Study library
Based on our ethnography, the study is the core model that ties all of our research together. Our researchers can document the details of their research engagements as well as top insights that come about. Our video library and participant database have a deep integration with our studies. Both stakeholders and researchers can then search through the entire index of our studies.
Dashboard
Stakeholders and researchers can follow studies to receive updates and track their status on their dashboards.
Search
Key to our insight vision was powerful search accessible to everyone. Users can search and filter through studies by product area, keywords, or researcher name. The index is powered by ElastiCache and ElasticSearch.
Consolidation
We accepted that we wouldn't be able to replace all the tools researchers use to perform and document their research. We wanted the study page in our tool to be a centralized space where other links and documents could be attached.
Technical specs
Frontend
React
GraphQL
Apollo
AWS AppSync
Emotion
Cypress
Jest
TypeScript
Backend
AWS
Lambda, RDS, AppSync, ElastiCache, Cognito, MediaConvert, Transcribe, SQS, S3, CloudFront, WAF, DynamoDBElasticsearch
Serverless Framework
Node.js