Data Marketplace

Collate Data Marketplace is the consumer surface inside the Collate platform where teams discover trusted data products, request the access they need, and approve those requests through a comments thread. The marketplace homepage is shaped to each persona, and access requests can be column-level, time-bound, justified, and reversible. Every action runs against the same metadata graph that powers Discovery, Lineage, Observability, and Governance.

A data catalog organizes everything that exists in your data estate. The marketplace surfaces only the curated, trusted data products that are ready for consumption, with a simpler search-and-request flow tuned to consumers. Catalog and marketplace work together: the catalog is the source of truth for all metadata, the marketplace is the consumer experience layer on top.

Two things make it different. The marketplace itself adapts to your role: analysts, stewards, and engineers each see a homepage shaped to their work, and admins can rearrange the widgets without writing code. The access requests adapt too: ask for specific columns instead of the whole table, set the access to expire in a week or six months, write a reason for why you need it, and work out the details with the approver in a comments thread before anything gets approved. Other marketplaces give you one surface for everyone and a yes/no approval button. Collate adapts to both the person and the request.

A dataset is a single table or file. A data product is the bigger thing: a curated bundle owned by one team, designed for other teams to use. It declares what feeds it (input ports), how you consume it (output ports), and which datasets, dashboards, or models are inside. The marketplace handles both as separate request types because what matters when you're asking for a single dataset is different from what matters when you're asking for a whole product.

Yes. A Data Access Request supports column-level access: pick the dataset, select column-level instead of full, and list the columns you need. The approval grants access only to the columns named, not the whole table. Column-level access is captured alongside full access in the unified audit log.

Approvals run as a conversation. When a steward opens a pending request, they see the request context (dataset or data product, requester, duration, reason) alongside a comments thread. The steward can ask questions, tag domain owners with @mention, or request a different access scope before deciding. The decision and the conversation are both captured in the audit log.

Yes. Designated administrators use a drag-and-drop widget editor to configure each persona's marketplace homepage. Widgets include search, recent searches, new data products, new domains, my requests, pending approvals, and announcements. Save the layout once and every user assigned that persona inherits it. Default personas (business user, data steward, data engineer) ship with the product and your team can extend or define new personas.

Access auto-expires at the end of the duration the requester chose. Durations range from one week to six months. The audit log captures the original grant, the expiration, and any revocations. Consumers who need continued access submit a new request, with the prior history visible to approvers.

The marketplace is one surface inside the unified Collate platform, alongside Discovery, Lineage, Observability, Insights, and Governance. The data products consumers browse are the same data products data engineers define elsewhere in Collate. The lineage and quality signals attached to those products are the same signals available across every other Collate surface. Approvals flow through the same role-based access controls that govern every other Collate action.

Yes. Domains in Collate Data Marketplace follow the data mesh pattern: you can label them Aggregate, Consumer-aligned, or Source-aligned, and nest sub-domains under parent domains to match how your organization is structured. Data products are first-class objects with input and output ports, so a domain team can publish a product and other teams can find it, evaluate it, and request access through the marketplace. It's where the teams that produce data and the teams that need it actually meet.