Explore my portfolio

MDL Research Project

Overview

Summary

Role: UX Researcher/Product Designer
Collaboration: Internal Matillion Product Team
Focus: Research planning, facilitation, user testing, analysis
Tools: Calendly, Confluence, Empathy Mapping, Usability Testing, Affinity Diagramming

As part of a strategic research initiative, I investigated the inefficiencies data engineers face when debugging and maintaining pipelines in Matillion’s Data Loader (MDL). Through empathy mapping, usability tests, and heuristic reviews, I provided actionable insights to improve the MDL support system.

Problem Statement

Matillion’s internal support data and user feedback revealed a persistent pain point:
Data engineers were spending excessive time fixing, debugging, and maintaining pipelines.

This raised questions:

  • Is the current support system aligned with the real needs of data engineers?

  • What specific areas of the MDL product are breaking the user flow or causing time loss?

  • How might we improve documentation, UX support, or tooling to reduce this burden?

These questions guided the design of a research study focused on exploring user pain points and delivering clear recommendations.

Research Goals

  • Understand user personas (primarily data engineers) working with MDL

  • Identify knowledge gaps and recurring pain points in debugging/maintenance

  • Evaluate the effectiveness of the MDL support system (documentation, UI, workflows)

  • Derive recommendations for reducing time and friction spent on fixes

Research Process

Empathy Mapping:

  • Built a persona-specific empathy map to visualise what users think, feel, say, and do

  • Focused on the “Data Engineer” role and mapped recurring frustrations around pipeline reliability and complexity

Formulating the Research Question:

“How can we improve the support experience in MDL so that data engineers spend less time debugging, maintaining, and fixing pipelines?”

This helped scope the research plan and select appropriate methods.

 

Secondary Research

Heuristic Evaluation:

  • Reviewed the current MDL support experience (e.g. help centre, in-product help, documentation)

  • Assessed consistency, findability, and clarity of support materials using heuristic criteria

Documentation Review:

  • Analysed Confluence documentation and internal guides to assess completeness and user alignment

Ideation & Design

Early Collaboration:

  • Partnered with engineers to explore feasible ways to simplify complexity

  • Identified a pattern: most confusion stemmed from not knowing how each strategy worked in practice

Key Design Decisions:

  • Introduced an eight-tile layout, each tile representing a pagination strategy

  • Included code snippet previews on each tile for visual understanding

  • Enabled inline configuration once a strategy is selected

  • Designed a preview feature to show users how their selected setup would behave using live API data

  • Prioritised usability principles such as reducing clicks and following Nielsen Norman heuristics

Primary Research

Usability Testing:

  • Planned and conducted 5 moderated usability tests with internal data engineers

  • Used Calendly for participant scheduling

  • Created a structured script focusing on:

    • Discoverability of help materials

    • Steps taken when debugging pipelines

    • Friction points in existing MDL (Matillion data loader) support interactions

*An outline of the script used within the usability tests*

Challenges:

  • Carefully avoided leading questions, ensuring neutral facilitation

  • Required flexibility in adapting the script based on participant reactions

  • Created a safe space for open feedback on internal tools

Outcomes & Insights

Analysis:

  • Used an Affinity Diagram to categorise and prioritise key themes:

  • Lack of in-context support and tooltips

  • Overwhelming and scattered documentation

  • No clear debugging flow or triage checklist

  • Ambiguous error messaging

Key Recommendations:

  • Embed contextual support directly within MDL (e.g. inline hints, alerts)

  • Consolidate documentation into a step-by-step debugging guide

  • Introduce a “Common Fixes” section informed by real user cases

  • Develop a lightweight debug checklist UI to assist engineers during troubleshooting

Reflection & Impact

This project not only informed product support strategy for MDL, but also served as a real-world test bed for:

  • Research planning and execution

  • Stakeholder alignment

  • Quantifying qualitative data into recommendations

Learnings

  • User confidence comes from context and control—clear language and visual examples matter deeply when designing for technical workflows

  • Collaborating early with engineering ensures design feasibility and speeds up development alignment

  • A modular UI design with consistent interaction patterns supports both scalability and usability

Academic Impact

This project directly supported my preparation for an upcoming dissertation on UX in technical enterprise tools, providing hands-on experience in participant recruitment, facilitation, and synthesis.

Next Steps:

  • Share insights with Product and Engineering to define support system MVP upgrades

  • Explore introducing UX metrics (e.g. Time to Fix, Number of Support Searches) to validate success

  • Consider embedding a feedback loop within MDL for users to report unclear errors in real time