OneLake Data Catalog
The OneLake catalog is the central hub to discover, manage, govern and secure every data item across Fabric. We set it up so your people find the trusted dataset in seconds — and your Copilot and agents ground on endorsed, governed sources rather than the nearest available table.
Quick enquiry
Get started with Data Catalog
Tell us where you are. Amit replies within one business day — no slides, no pitch, no obligation to proceed.
- A senior practitioner reads every enquiry
- First value in 6 weeks, not a 50-slide roadmap
- Your details are never shared with third parties
What we hear from operators
The problems we solve
Nobody can find the right dataset
In a growing Fabric estate, the same data exists in five workspaces and nobody knows which is current. So teams rebuild what already exists, and two reports on the same question disagree because they were built on different copies. A catalog with search, tags and descriptions is what turns "does this exist?" from a Slack message into a two-second lookup.
No one knows which number to trust
Without endorsements, every dataset and report looks equally official, so a user picks whichever they find first — often the wrong one. Certification and promotion in the catalog mark the trusted, owned assets clearly, so people and agents use the endorsed source instead of guessing. Trust has to be visible to be useful.
Governance posture is invisible
When you cannot see which items are sensitive, undocumented or over-shared, you cannot fix it — and sensitive data quietly spreads. The catalog's governance view surfaces the posture of the data you own and the actions that improve it, turning governance from an annual audit into an ongoing, visible practice.
Agents and Copilot ground on ungoverned data
The reason AI over your data gives confident wrong answers is often that it grounded on the nearest table rather than the right one. A governed catalog with endorsed sources is what lets Copilot and Fabric Data Agents reason from trusted, classified data — which is why the catalog is a foundation for AI, not just a convenience for humans.
Who this is for
Who onelake data catalog is built for
The roles that feel the problem first — and what we build for each of them.
Chief Data Officer
The problem
Cannot see the governance posture of a sprawling Fabric estate, so risk is managed by audit after the fact.
What we build
The OneLake catalog governance view plus Purview classification, surfacing sensitive, undocumented and over-shared items and the actions to fix them.
Governance posture visible and improving continuously
Data Governance Lead
The problem
Users cannot tell the trusted dataset from a stale copy, so the wrong number keeps ending up in decks.
What we build
Certification and promotion endorsements on owned assets, with tags, descriptions and ownership across the catalog.
Trusted sources clearly marked and used by default
Head of BI / Data
The problem
Teams rebuild datasets that already exist because discovery is a Slack message, not a search.
What we build
A curated, searchable OneLake catalog embedded in Teams and Excel so the current dataset is a two-second lookup.
Duplicated and conflicting assets stop multiplying
CIO
The problem
AI over company data gives unreliable answers and IT cannot see what it is grounding on.
What we build
A governed catalog of endorsed, classified sources that Copilot and Fabric Data Agents ground on, queryable via APIs and the Fabric MCP.
AI grounded on trusted, governed data
Measurable outcomes
What changes after implementation
Specific shifts from delivered onelake data catalog work — the before, and the after.
Discovery: "does this exist?" → two-second catalog search
Teams find the current, trusted dataset instead of rebuilding it, so duplicated and conflicting assets stop multiplying.
Trust: every asset looks official → endorsed sources clearly marked
Certification and promotion make the owned, trusted assets obvious, so people and agents use the right source by default.
Governance: annual audit → visible, ongoing posture
The governance view surfaces sensitive, undocumented and over-shared items and the actions to fix them, so posture improves continuously.
AI grounding: nearest table → endorsed, governed sources
Copilot and Fabric Data Agents reason from certified, classified data, which is the single biggest lever on whether their answers can be trusted.
By market
Data Catalog — market-specific pages
Each page below covers what onelake data catalog looks like specifically in that market — the local ERP landscape, compliance context, and the operational patterns we actually see there.
Singapore & Malaysia
United Kingdom
North America
Technology stack
Related services
Common questions
Data Catalog — frequently asked
What is the OneLake catalog?
The OneLake catalog is the single place in Microsoft Fabric for data professionals and business users to discover, manage, govern and secure the data they can access across OneLake. It has an Explore experience for finding data with search, tags and endorsements, and a Govern experience that shows the governance posture of the data you own and the actions to improve it. It covers the full range of Fabric item types, including reports and dashboards.
Is the OneLake catalog the same as Microsoft Purview?
They are complementary, not the same. The OneLake catalog is the discovery and governance hub inside Fabric for your Fabric data; Microsoft Purview provides broader enterprise data governance, sensitivity labelling and compliance across your whole estate, including sources beyond Fabric. In practice we use the OneLake catalog for in-Fabric discovery, endorsement and posture, and Purview for sensitivity classification and enterprise-wide governance, and connect the two.
How does the catalog help Copilot and AI agents?
AI over your data is only as good as what it grounds on, and the catalog is how you make the trusted sources the ones agents find. Endorsed, well-described, classified items in the catalog let Copilot and Fabric Data Agents reason from certified data rather than the nearest available table — which is the most common cause of confident wrong answers. The catalog is also queryable through public APIs and the Fabric MCP, so agents can discover data programmatically.
What is the difference between certified and promoted endorsements?
Promotion is a lighter endorsement — an owner or contributor marking an item as ready for others to use. Certification is a stronger, controlled endorsement, usually reserved for authoritative assets and granted under a defined process by a smaller group. Using both well means users can see at a glance which assets are the trusted, owned ones and which are simply shared, which is what stops people building on the wrong copy.
How do we keep sensitive data secure while making data discoverable?
Discovery and security are separated in the catalog: people can find that an item exists and who owns it without being able to open data they are not entitled to. We apply row-level and object-level security and Purview sensitivity labels so access is governed, and use the governance view to find and fix over-shared items. Making data discoverable does not mean making it open — it means making the right people able to find and request the right data.
What are the prerequisites to set up the OneLake catalog well?
A Fabric estate with data worth cataloguing and clear ownership of the important domains. The catalog itself is part of Fabric, so there is nothing to install — the work is curation and governance: assigning owners, adding tags and descriptions, setting up the endorsement process, and applying sensitivity. It works best alongside a governed semantic model and, where you are heading toward agents, a Fabric IQ ontology, so discovery, governance and business meaning reinforce each other.
Related FAQs
Go deeper on the questions buyers ask
Start with a conversation, not a proposal
First call is 30 minutes with Amit. We ask about your systems, your team, and your most pressing operational problem. You get a clear view of where the gap is and what closing it looks like. No slides. No pitch deck. No obligation to proceed.