Intelligent Mapping: Accelerating Data Integration with AI
Overview
-
My Role: UX Manager
-
The Problem: Creating data integration mappings is traditionally a complex, manual, and time-consuming task. Users spend a significant amount of time determining the logical sequence of data transformations and manually mapping hundreds of fields between sources and targets.
-
The Solution: This project aimed to fundamentally enhance the mapping experience by embedding Machine Learning capabilities directly into the workflow. We introduced two core intelligent features:
-
Next-Step Recommendations: An engine that suggests the most likely next transformation for a user to add to their mapping, guiding them through the process.
-
Automated Field Mapping: A feature that automatically matches and maps fields between transformations, eliminating tedious manual work. These features were designed to accelerate the mapping process, reduce errors, and make data integration more accessible to a broader range of users.
-
Scope and Goals
Strategic Goals: Our primary objective was to leverage AI to make the mapping process faster, smarter, and more reliable. We aimed to:
-
Drastically reduce the time and effort required to create a complete data integration mapping.
-
Improve the accuracy of mappings by minimizing manual field-matching errors.
-
Lower the learning curve for new or less-technical users, empowering them to build complex mappings with confidence.
-
Increase user satisfaction and productivity for one of the platform's most critical tasks.
My Core Responsibilities:
-
Initial Concept Design: I worked with the Product Manager to create the first set of high-level designs.
-
Team Leadership & Design Direction: I managed and mentored the UX designer responsible for the project, guiding the detailed interaction and visual design work to ensure it met a high standard of usability and aligned with the product's vision
Process
Discovery & Conceptualization:
-
We started by doing a competitive analysis, identifying patterns of visual chart editors offering “next shape” interactions.
-
We analyzed the existing mapping editor with a focus on existing interactions and gestures allowing users to add and modify mapping elements (aka “transformations”).
-
I guided the creation of low-fidelity wireframes and prototypes to explore different ways of presenting recommendations (e.g., in-context pop-ups) and handling the review of auto-mapped fields.
-
Early concepts were tested with internal users to gather initial feedback on the approach and its potential value.
Design & Iteration:
-
Based on feedback, I guided the design team in refining the concepts into high-fidelity, interactive prototypes.
-
A key challenge was designing the "review and accept" flow for auto-mapping. We needed to clearly communicate which fields were matched with high confidence versus lower confidence, allowing users to quickly approve, reject, or modify the AI's suggestions.
-
We designed the user interface to be non-intrusive, providing assistance when needed but allowing expert users to work without interruption if they chose to.
Collaboration & Refinement:
-
I fostered a close partnership between my designer, product management, and engineering to define the concept.
-
I continued to guide the designer through the detailed design phases.
-
We maintained a tight feedback loop with the ML engineering team to ensure the user experience was aligned with the model's capabilities.
Next Transformation Recommendation
Recommendations
Recommendations Menu Alternatives
Presenting a new interaction that enhances the mapping canvas by displaying a list of available transformations based on the previous one, along with AI-based recommended transformations.
Design explorations for the Next Transformation menu
Field Auto-Matching With Smart Match
Auto-Match Menu
Introducing a new feature that allows users to select from various auto-mapping options, including conventional auto-match methods as well as an innovative AI-driven approach.
Auto-Match
Auto-mapped fields are highlighted to enable the user to examine the mapping and make any necessary adjustments.