Data products in Knowledge Catalog (formerly Dataplex Universal Catalog) let you package related catalog
assets into a single, curated unit to solve specific business problems. You can
then add context and contracts to help data consumers discover, evaluate, and
request self-service access to trusted data.

## Key capabilities

Data products provide the following key capabilities:

- Organize catalog assets into a logical unit that solves a specific business problem and speeds up time to insights.
- Distribute data assets with context that includes a description, documentation, and aspects.
- Establish trust with contracts that let data producers provide assurance to data consumers.
- Provide a self-service workflow for data consumers to evaluate data products and get access to data.

## Key concepts

This section describes the key concepts and terminologies related to data
products.

### Data product

A curated, logical grouping of data assets, formally packaged to be discoverable,
trusted, and accessible for solving specific business problems.

### Asset

A pointer to a physical data resource, such as a BigQuery dataset,
table, or view. A data product contains one or more assets.

### Access group

Access groups simplify permission management for your data product. They map
user-friendly roles (such as `Reader` or `Analyst`) to underlying
Google Groups or service accounts. This abstraction enables data product
owners to manage access at a conceptual level, and helps data product consumers
request the appropriate level of access.

- Data product owners configure access groups and can optionally assign a default
  IAM role, which is automatically applied to all supported
  assets in the data product. If a default role isn't assigned, owners can
  assign specific IAM roles to assets directly.

- Data product consumers use these groups to request access to the data product.

### Data product owner or data producer

The individual or team responsible for the creation and management of data
products. This includes managing quality, access, and documentation.

### Data product consumer

The individual, team, or AI agent that consumes data products to generate
insights.

### Contract

An agreement between the data product owner and its consumers. This
agreement sets clear expectations by defining specific terms for how the data
will be provided and used, such as its refresh schedule and quality standards.

## Example use case

Consider a data scientist analyzing an ecommerce business. Their goal is to
find the average order value (AOV) by traffic source and see if there's a
correlation between user age and order size. To do this, they need to combine
data from multiple tables, such as `order_details`, `user_traffic`, and
`user_demographics`.

In a conventional setup, this process creates friction. To generate insights,
the data scientist must first discover the correct tables within the
organization's vast data landscape, then contact each data owner, justify their access request,
and wait for approval.

With data products, data owners can streamline this experience by packaging
the relevant assets into a single product named "Ecommerce Business Data". This
package includes the following:

- Assets

  - BigQuery tables `order_details` and `user_traffic` (containing historical order data and traffic sources)
  - BigQuery view `user_demographics` (providing user details with PII excluded)
- Access groups

  - Predefined `Reader` and `Writer` groups to streamline access requests
- Contract

  - A contract defining the data refresh frequency (for example, weekly at 8:00 AM PST)
- Context

  - Documentation with sample queries and other details
  - Additional metadata to depict data sensitivity

Data scientists can now discover this data product as a single logical unit.
This lets them confidently generate insights to answer questions like,
'What is the average order value for each traffic source?'---ultimately
revealing which sources generate the highest value customers.

## Data product user flow

The data product lifecycle in Knowledge Catalog involves two key user
journeys: one for the data product owner (or producer) who creates and manages
the data, and one for the data product consumer who discovers and uses it.

### Data product owner journey

This journey focuses on packaging, securing, and governing the data products to
ensure they are trusted and accessible.

- Create: Define the data product and include assets. This involves the
  following actions:

  - Configure the unique name, project, region, and description.
  - Add assets such as BigQuery tables, datasets, or views.
  - Configure access groups (for example, `Analyst` or `Reader`) and map them to underlying Google Groups or service accounts to simplify permission management.
  - Optionally, assign a default IAM role to an access group to automatically grant permissions across all supported assets in the data product.
  - Assign the necessary IAM roles to these access groups for the specific assets.
  - Add a contract (a system aspect) to formally communicate the agreed-upon data refresh cadence, frequency, and threshold.

  For more information, see
  [Create data products](https://docs.cloud.google.com/knowledge-catalog/docs/create-data-products.md.txt).
- Manage: Update the data product and ensure discoverability. This involves
  the following actions:

  - Update basic details, assets, permissions, and supplementary aspects (metadata), and rich text documentation.
  - Grant access to consumers to discover and request access to data products.

  For more information, see
  [Manage data products](https://docs.cloud.google.com/knowledge-catalog/docs/manage-data-products.md.txt).

### Data product consumer journey

This journey focuses on quickly finding trusted data and gaining the necessary
permissions to use it.

- Discover: Find relevant, trusted data for a specific business problem. This
  involves the following actions:

  - Use the [Knowledge Catalog Search](https://docs.cloud.google.com/knowledge-catalog/docs/search-assets.md.txt) with keywords or natural language to find the packaged data product.
  - Review the data product's overview, assets, contract, and other aspects to determine its fitness for use.

  For more information, see
  [Search for data products](https://docs.cloud.google.com/knowledge-catalog/docs/use-data-products.md.txt#search).
- Request access: Ask the data product owner for permission to access the data.

  For more information, see
  [Request access to data products](https://docs.cloud.google.com/knowledge-catalog/docs/use-data-products.md.txt#request-access).
- Use: Access the underlying assets to generate insights. This involves the
  following action:

  - Upon approval, you can access the product and its assets. For example, if
    the asset is a BigQuery table, you can navigate to the
    BigQuery studio and query the data directly.

  - For applications and development workflows operating outside of
    Google Cloud, you can expose the data product using an
    external metadata gateway. For more information, see
    [Use the Knowledge Catalog remote MCP server](https://docs.cloud.google.com/knowledge-catalog/docs/use-remote-mcp.md.txt).

  For more information, see
  [Consume data products](https://docs.cloud.google.com/knowledge-catalog/docs/use-data-products.md.txt#consume-data-products).

## Assets supported

A data product can be composed of one or more data assets. The following data
assets are supported:

- BigQuery datasets
- BigQuery tables
- BigQuery views
- BigQuery routines
- BigQuery models
- BigQuery external tables
- BigQuery graphs
- Gemini Enterprise Agent Platform datasets
- Gemini Enterprise Agent Platform models
- Agent Platform feature groups
- Agent Platform feature online stores
- Agent Platform feature views
- Cloud Storage buckets
- Iceberg REST Catalog catalogs
- Iceberg REST Catalog namespaces
- Iceberg REST Catalog tables
- Looker (Google Cloud core) user-defined dashboards
- Looker (Google Cloud core) user-defined dashboard elements
- Looker (Google Cloud core) Looks
- Looker (Google Cloud core) Explores
- Looker (Google Cloud core) LookML projects
- Looker (Google Cloud core) LookML models
- Looker (Google Cloud core) LookML views
- Bigtable instances
- Bigtable tables
- Bigtable materialized views
- Bigtable logical views
- Bigtable authorized views
- Spanner databases
- Gemini Data Analytics data agents

## Limitations

- Location: Data products and their underlying assets must reside in the same Google Cloud location.
- BigQuery models and graphs: Access to BigQuery models and graphs within a data product is managed through [IAM conditions](https://docs.cloud.google.com/iam/docs/conditions-overview.md.txt) applied to the parent dataset's IAM policy. Sharing BigQuery models and graphs is subject to the [limitations of IAM conditions](https://docs.cloud.google.com/iam/quotas.md.txt#limits).
- Looker (Google Cloud core) assets: Using Looker (Google Cloud core) assets in
  data products has the following limitations:

  - Service accounts aren't supported: [Google Group mirroring](https://docs.cloud.google.com/looker/docs/looker-core-oauth-authentication.md.txt#mirroring) in Looker (Google Cloud core) only supports human user identities. Access groups configured on Looker (Google Cloud core) assets can't include a service account principal.
  - Default access group IAM roles aren't supported: Default IAM roles assigned at the access group level aren't applied to Looker (Google Cloud core) assets and are marked as `unsupported`. You must explicitly configure access group permissions on each Looker (Google Cloud core) asset.
  - Manual permission synchronization inside Looker (Google Cloud core): Role names configured on a Looker (Google Cloud core) asset in a data product are informational placeholders for consumers. Data producers must manually configure and maintain [folder access levels](https://docs.cloud.google.com/looker/docs/admin-panel-users-content-access.md.txt#access_levels) and [Looker roles](https://docs.cloud.google.com/looker/docs/admin-panel-users-roles.md.txt) for the mirrored group inside the Looker (Google Cloud core) instance.
  - LookML dashboards aren't supported: LookML dashboards and LookML dashboard elements aren't supported as data assets, even though they might appear in the list of assets that you can add in the Google Cloud console. To include a LookML dashboard in a data product, copy it as a user-defined dashboard (see [Converting from LookML to user-defined dashboards](https://docs.cloud.google.com/looker/docs/converting-lookml-to-user-defined-dashboard.md.txt)).
- Quotas and limits: For a complete list of API rate limits and capacity quotas,
  see
  [Quotas for data products API requests](https://docs.cloud.google.com/knowledge-catalog/docs/quotas.md.txt#data-product-quotas).

- [Zonal buckets](https://docs.cloud.google.com/storage/docs/rapid/rapid-bucket.md.txt) aren't supported.

## What's next

- Learn how to [create a data product](https://docs.cloud.google.com/knowledge-catalog/docs/create-data-products.md.txt).
- Learn more about [managing data products](https://docs.cloud.google.com/knowledge-catalog/docs/manage-data-products.md.txt).
- Learn how to [search for data products](https://docs.cloud.google.com/knowledge-catalog/docs/use-data-products.md.txt#search).
- Learn how to [request access for data products](https://docs.cloud.google.com/knowledge-catalog/docs/use-data-products.md.txt#request-access).
- Learn how to [use VPC Service Controls with data products](https://docs.cloud.google.com/knowledge-catalog/docs/vpcsc-data-products.md.txt).