Designing Yext Connectors, A Framework for Automated Data Ingestion at Scale
A user-first approach to Extract, Transform and Load data into the Yext Knowledge Graph
ROLE
Lead Product Designer
COMPANY
Yext
DURATION
JAN - MAR 2021
As Product Design Lead,
I designed Yext Connectors, a scalable data ingestion framework inspired by ETL workflows. This product enables users to seamlessly integrate structured data from multiple sources into the Yext platform.
COMPANY OVERVIEW
Yext centralizes business information through its Knowledge Graph CMS and AI tools, enabling seamless content syndication across digital platforms.
After completing work to migrate to a multi-entity type, knowledge graph database powered system we began to focus our efforts on one of the larger known critical issues with the platform, data ingestion and bulk updates .
PROBLEM
Enterprise users struggled with bulk updates, uploads, and data edits in Yext, facing error-prone manual file uploads, slow single-entry additions, and costly API integrations.
Data gathered from 17 enterprise User interviews, and quantitative Snowflake data .
CORE USER TYPE
Admin Users manage large-scale data in Yext’s Knowledge Graph, overseeing thousands of entities
Prioritize efficiency, advanced controls and flexible data management
Being in the Yext Platform is not their primary responsibility
Their raw data lives in multiple formats, in multiple placesConfigure the platform to execute brand data needs
Construct user permissions and approval flows
Key decision makers in software adoption
Analysis from 17 customer interviews informed a catalogue of features which resembled an ETL framework
Data Ingestion & Integration
Bring in data from multiple sources, including files, APIs, and third-party integrations.
Multi Source Data Support
Bulk Uploads
API Integrations
Web Crawling
Third-Party Integrations
Scheduled Data Syncing
Data Transformation & Structuring
Clean, transform, and map data before loading it into the Knowledge Graph
Data Transform & Cleaning
Flexible Data Mapping
Automated Data Loading
Prebuilt Data Connectors
No-Code Configuration
Real-Time Validation
Monitoring, Compliance & Automation
Track errors, ensure compliance, and route to approval worklows.
Error Handling & Troubleshooting
Notifications & Alerts
Approval-Based Workflows
Audit Logging & Compliance
Competitive research into ETL platforms found that while many offer robust features, directly mimicking their approach could introduce complexity and reduce usability for core Admin Users.
INFINITE CANVAS
Cluttered UI and complexity compromise persistent data views
LIVE DATA PREVIEWS
Enables real-time validation, reducing errors throughout the flow
TRANSFORM FRAMEWORK
Extensive ETL transforms served as a key reference
To manage the complexity of building an ETL product in Yext, I structured the project into two parallel design workstreams. This allowed for iterative testing and refinement.
CORE DESIGN ELEMENTS
Layout
Task Flow
Card Component
Live Data Preview
ADAPTING ETL FRAMEWORK
To discuss the ETL approach I created a breadboard to visualize key flows as wireframes and interactions.
Core Design Elements
Layout
To enhance clarity and usability, I designed a split-panel layout that allowed users to navigate tasks effortlessly while keeping their configurations and data in view
1
Dedicated panel to represent progress within the ETL framework
2
Ample space for users to interact with configuration settings
3
Persistent data preview area offers real-time feedback and validation
4
The lower action bar supports multiple functions for flexibility
Task Flow
Card Component
Refined task flow by abstracting the ETL process, reducing visual noise, and introducing a card component to improve click target clarity.
TIMELINE
Timeline demonstrated a linear progression, but visual elements reduced readability.
RADIO CARD
Cards improved click targets, but excess visual noise continued to reduce clarity.
TASKFLOW CARD
Abstracted ETL framework, retained card design, and simplified visual elements.
Live Data Preview
Live data preview provided real-time feedback, helping users refine transformations, validate data, and reduce errors.
1
Selecting a task flow card updated the data preview, reflecting changes in real time.
2
Column indicators highlighted applied transformations for clear visibility.
3
A status bar enabled users to refresh sample data and monitor source status.
Adapting ETL Framework
I iterated on design elements in parallel with the development of the ETL framework, adapting it for the Yext platform.
Extract
Sources were represented as selectable cards grouped by type. Selecting the source would continue the user to configuration settings for the source
Extract
Users selected data fields for their table. A one-click action quickly added all columns for files, APIs, and native sources.
1
Default selectors were quickly applied to extract key data
2
Extracted data was instantly displayed in a structured format.
3
Users could manually defined selectors for greater control over.
Transforms
A modal and taskflow approach was evaluated for applying transforms.
1
An overlay was considered but obscured previews and duplicated work, increasing complexity.
2
Selected an integrated taskflow transform design with clear linear progression within the ETL flow.
3
The design leveraged live data previews to show transform effects in real time.
3
Once applied, the transform condensed into a summary tile.
Extract
The final stage of the ETL process required users to select an entity type, then map columns to fields. Run modes would determine how entities would be updated or added to the Knowledge Graph.
Extract
Standard table views provide a clear overview of connector details and run summaries. Run detail pages allow users to drill down into entity creation, updates, and deletions.
RUN SUMMARY
RUN DETAIL
A competitive product designed in 12 weeks, with continued support and development
PROJECT OUTCOMES