Brief
Data Marketplace
A Data Marketplace Design: From User Research and Strategy to MVP
Engineers, data scientists, managers, and stakeholders currently struggle to find and utilize data effectively due to the existence of multiple data management portals. The initial goal is to improve data discovery by creating a more unified access point, enabling seamless data discovery and utilization across the organization.
Search-to-discovery success rate.
Time to find a usable dataset.
Data engineering ticket volume.
Adoption & engagement
Early Research
Data Insight
User Interviews
I started early research by intentionally recruiting across the full ecosystem of the data marketplace: data producers, consumers, and gatekeepers (engineers, analysts, governance leads, and business stakeholders). This ensured the insights reflected how decisions were actually made around data, not just the perspective of the loudest or most technical users.
During sessions, I grounded conversations in real workflows instead of hypotheticals. I asked participants to walk me through their current tools and end-to-end journeys—where they start, what they open next, where handoffs happen, and where they get stuck. This helped me map concrete friction points, workarounds, and dependencies that would later inform key interaction patterns.
As a lead interaction designer, I also used early research to surface governance, privacy, and risk considerations alongside user needs. I made access policies, data quality concerns, and security constraints explicit parts of my discussion guides, so we could frame opportunities that were both desirable for users and realistic for platform, governance, and security teams.
Design Hypothesis
Job-To-be-Done
Focussing on what users are actually trying to accomplish in their day-to-day work: finding trusted data, knowing whether they’re allowed to use it, and moving quickly from discovery to action. Looking through that lens let me separate the jobs users need to get done from the way the organization currently delivers them, which opened up space for a more focused, outcome-driven MVP.
From the interviews, four recurring pain points emerged that shaped the MVP.
Trust
Users struggled to understand whether a dataset was reliable, up to date, and approved for their use case, which led to second-guessing and shadow copies of “trusted” data.
Matching
Even when data existed, users found it hard to match the right dataset to their specific question or decision, often relying on tribal knowledge instead of clear signals in the tools.
Ease of Use
Existing tools felt fragmented and unintuitive, forcing users to jump between multiple systems, remember obscure dataset names, and piece together workflows on their own.
Accessing Data
Accessing data was slow and opaque—users didn’t always know who to ask, what they were allowed to see, or how long approvals would take, which delayed time-sensitive decisions.
If we design a data marketplace that makes it simple for employees to discover, trust, and access high-quality datasets, then they will rely on it as their primary starting point for data-driven decisions, leading to higher adoption, more efficient workflows, and stronger alignment around shared, trusted data.
Design trade-offs and Considerations
Search Experince
Initial Interaction: Setting the Stage
Initial View Upon first access, the user is greeted with a clean and welcoming landing page.
Theme Choice with Light Mode or Dark Mode. This caters to user preference and accessibility. The options will be visually distinct and easily selectable with a clear label.
No Forced Login Users can explore the high-level categories and perhaps a few featured datasets without immediate login.
Navigating and Discovering Data
Prominent Placement A highly visible and centrally located search bar is paramount. It should be the primary entry point for users who know what they're looking for.
Autosuggest and Autocomplete As the user types, the search bar will provide intelligent suggestions based on dataset titles, descriptions, tags, and potentially even categories. This speeds up the search process and helps users discover relevant data they might not have explicitly searched for.
Did You Mean?" Functionality: If the search query yields no results or few relevant results, a "Did you mean?" suggestion will help correct potential typos or offer alternative phrasings.
Categorical Browsing
Well-Organized Categories Datasets will be logically grouped into clear and intuitive categories
Sub-Categories Within each main category, users can drill down into more specific sub-categories to refine their exploration.
Visual Cues Clear icons alongside category names to enhance visual recognition and navigation.
The Craft
Interaction run-through & experience highlights.
Initial Interaction: Setting the Stage
Initial View Upon first access, the user is greeted with a clean and welcoming landing page.
Theme Choice with Light Mode or Dark Mode. This caters to user preference and accessibility. The options will be visually distinct and easily selectable with a clear label.
No Forced Login Users can explore the high-level categories and perhaps a few featured datasets without immediate login.
Navigating and Discovering Data
Prominent Placement A highly visible and centrally located search bar is paramount. It should be the primary entry point for users who know what they're looking for.
Autosuggest and Autocomplete As the user types, the search bar will provide intelligent suggestions based on dataset titles, descriptions, tags, and potentially even categories. This speeds up the search process and helps users discover relevant data they might not have explicitly searched for.
Did You Mean?" Functionality: If the search query yields no results or few relevant results, a "Did you mean?" suggestion will help correct potential typos or offer alternative phrasings.
Categorical Browsing
Well-Organized Categories Datasets will be logically grouped into clear and intuitive categories
Sub-Categories Within each main category, users can drill down into more specific sub-categories to refine their exploration.
Visual Cues Clear icons alongside category names to enhance visual recognition and navigation.
Filtering and Faceting
Refine Search Results On search results pages and category landing pages, a robust filtering and faceting mechanism will be available in a sidebar or expandable panel.
Dynamic and Content-aware Filters, based on relevance of datasets displayed.
Clear Indication of Applied Filters allowing users to easily remove individual filters or clear all filters.
Dataset Preview and Details
Concise Search Results Each search result will display key information at a glance: dataset title, a brief description, relevant tags, the last updated date, and potentially the data format.
Detailed Dataset Page Clicking on a search result or dataset card will lead to a dedicated dataset details page.
Detailed Dataset Page Clicking on a search result or dataset card will lead to a dedicated dataset details page.
Managing Data Selection: The Cart Experience
Clear "Add to Cart" Button Prominently displayed on dataset cards and the dataset details page.
Visual Confirmation Upon clicking "Add to Cart," clear visual feedback will indicate that the dataset has been added (e.g., a notification, a change in button state, an update to the cart icon)
Adding to Cart
Reusable & Reliable data
Create, discover and access data products across multiple use cases to minimize duplicate efforts and reduce costs through a reliable internal data marketplace.
Share data products
Enable large-scale sharing of data and analytics products sourced from disparate source systems, such as data bakehouses and data catalogs.
Governed data sharing
Embed governance mechanisms to provide assurance that data products are shared and used in a compliant manner across all use cases.
Automated data lifecycle
Gain efficiencies in how data products are packaged, shared and consumed by automating and managing data as a product across the lifecycle.
