Google's documentation now calls this product BigQuery sharing and describes it as formerly Analytics Hub. The console page is still titled Sharing (Analytics Hub), and this guide uses both names. Google describes it as a data exchange platform, and for a buyer it plays the role of a data marketplace: a provider shares a BigQuery dataset, and you receive a read-only reference to it in your own project. The guide explains what a buyer meets, what subscribing creates, what a linked dataset does not allow, and what to ask a provider before relying on a listing. Fokals is delivered direct, by REST API and as bulk files, which a buyer loads into BigQuery with BigQuery's own load jobs.
The objects a buyer meets
Google's introduction to BigQuery sharing uses five terms, and the rest of this guide uses them in the same way.
| Term | What it is |
|---|---|
| Data exchange | A container in which publishers place listings and subscribers browse and ask for access. It is private by default, and a public exchange can be seen and subscribed to by every Google Cloud user. |
| Listing | A reference to a shared resource, offered through an exchange with a description, sample queries, documentation links and a contact. It is private or public, and a public listing can be free or commercial. |
| Shared dataset | The BigQuery dataset a publisher shares. It can hold tables, views, materialized views, routines, table snapshots and other supported objects. |
| Linked dataset | A read-only dataset in your project that points at the shared dataset. Subscribing creates it without replicating data. |
| Subscription | The resource that records the subscriber and represents the connection between publisher and subscriber. |
From listing to first query
A buyer's path has four steps.
- Discover. In the Google Cloud console open the Sharing (Analytics Hub) page, choose Search listings and filter by private, public or organisation listings, by category, by the exchange's location and by provider. Browsing needs
roles/analyticshub.vieweron your project. - Request or purchase. A listing that needs approval or payment shows Request access or Purchase via Marketplace in place of Subscribe. Google says some providers contact you after a request to share their commercial datasets.
- Subscribe. Choose Subscribe, and in the Create linked dataset dialog give a project, a name for the linked dataset and a primary region, with replica regions as an option. You need
roles/bigquery.useron your project androles/analyticshub.subscriberon the listing, the exchange or the publisher's project. - Query. With
roles/bigquery.dataVieweryou can view and query the linked dataset.
A listing sold through Google Cloud Marketplace adds an order first. Google's page on commercial listings has you choose Purchase via Marketplace, set a subscription plan and accept the terms on the order summary, wait for the order to activate, and then return to BigQuery sharing to subscribe. For some listings you submit a form to get a quote, and any project on the same billing account can also subscribe. Access to the linked resource is managed by active Marketplace orders, so ask what happens to it when an order ends.
What a linked dataset allows and what it does not
The page on viewing and subscribing to listings says a linked dataset is read-only, and the introduction adds that you can query its tables and views but cannot add or update objects in it. Google's pricing section says there is no additional cost for managing exchanges or listings, that the publisher pays for storage, and that the subscriber pays for the queries it runs, on demand or on capacity pricing. These limits matter to a research or data team.
- No history of your own. A linked dataset is a reference, so a query reads what the publisher's table holds when the query runs. Google lists snapshots of linked dataset tables, and materialized views over them, as unsupported.
- Egress can be restricted. A publisher can switch off exporting. Google says the copy, clone, export and snapshot APIs are then unavailable, as are
CREATE TABLE AS SELECTand writing query results to a destination table. Ask before you plan on a copy of your own. - Access follows the subscription. Deleting the linked dataset unsubscribes you and leaves the source dataset alone. For a Marketplace listing, access is tied to the order.
- The publisher can see use. Google says publishers can track the jobs run against a shared dataset, consumption by subscriber projects and organisations, and total rows and bytes processed. The listing dialog also says whether the provider logs subscriber emails.
- Roles sit at dataset level. Google says you cannot set IAM roles on individual tables inside a linked dataset, so one grant covers the whole listing.
- Large joins. Google notes that queries which join linked datasets and exceed 1 TB of physical storage might fail.
Questions to put to a provider
A listing shows what the provider chose to publish: a description, sample queries, documentation links and regions. For company data, these questions fill the gaps.
| Question | Why a buyer asks it |
|---|---|
| Is egress restricted on the listing? | If it is, you cannot copy, snapshot or export, so you cannot keep a record of the data as it stood on a date. |
| Are published rows ever revised, and how is a correction marked? | A linked dataset shows a table as it is now, and a back-test needs to know what was known on each date. |
| Which identifiers does each row carry? | Joins to a security master or a CRM need stable keys such as ISIN, LEI, ticker with market code, or a domain. |
| In which regions is the listing available? | Google says to colocate a linked dataset with your other data to limit egress and ease joins across datasets. |
| What does the licence allow? | A subscription gives access, while internal use, embedding in a product and redistribution are rights the licence grants or withholds. |
| How was the data sourced? | A compliance review asks, and a sourcing statement answers. |
Fokals answers the same questions for its own data. Every observation is dated, and daily and weekly datasets are written once, after the period closes, and never revised. Each company has one stable company ID, and a listed company also carries its ticker, MIC, ISIN, LEI and share-class FIGI, as the data dictionary sets out. Sourcing is first-party company sources and public records, processed in-house, and the sourcing document describes it. Licensing is by written agreement for internal use, embedding in a product or redistribution, and the comparison of data marketplaces with direct licensing sets the two routes side by side.
When the data is delivered direct
A provider that delivers by API and bulk files, as Fokals does, puts the copy in your hands. You download an export, land the files in a Cloud Storage bucket, load them into a dataset in the region you choose and keep every file. The delivery page describes the API and the exports, and the guide to loading company data into BigQuery gives the steps. According to Google's pages, the two routes differ as follows.
| Subscribing to a listing | Loading files you license directly | |
|---|---|---|
| What is in your project | A read-only linked dataset that points at the publisher's shared dataset | Tables you create from the files |
| Storage | The publisher pays for it | You pay once the data is loaded, and for the files you keep in Cloud Storage, as Google's pricing page states |
| Updates | Whatever the publisher changes appears in your queries | You choose when each file is loaded |
| Your own history | Snapshots are unsupported, and copying stops if egress is restricted | You keep every file you load |
| Setup | Subscribe in the console | Create a bucket, a dataset and a load job |
A listing is built for data in place without managing loads, with the publisher keeping storage current. Loading files is built for a record of your own, a schema you control, and one pipeline for every provider whatever route it delivers by. The comparison of BigQuery sharing with Snowflake Marketplace covers the choice between platforms.
Where this guide stops
This guide reflects Google's pages as read on 4 October 2026, and regions, roles and limits change, so check the linked pages before you sign. It covers the buyer's side of BigQuery sharing and leaves publishing, data clean rooms and sharing Pub/Sub topics to Google's documentation. For company data, the datasets to start from are on the data page.
Frequently asked questions
What is BigQuery Analytics Hub called now?
Google's documentation calls it BigQuery sharing and describes it as formerly Analytics Hub. The console page is still titled Sharing (Analytics Hub), the setup steps still refer to enabling the Analytics Hub API, and the roles keep the analyticshub prefix, such as roles/analyticshub.subscriber. Google's pages still use the terms data exchange, listing and linked dataset.
Does a linked dataset copy the data into my project?
No. Google says subscribing creates a read-only linked dataset that serves as a pointer to the shared dataset, without replicating data. You can query its tables and views but cannot add or update objects in it. The publisher pays for storage and you pay for the queries you run. If you need a copy of your own, check whether the provider restricts egress.
What roles do I need to subscribe to a listing?
Google lists roles/bigquery.user on your project and roles/analyticshub.subscriber on the publisher's listing, exchange or project to subscribe, roles/analyticshub.viewer to discover listings, and roles/bigquery.dataViewer to view and query a linked dataset. Deleting a linked dataset needs roles/bigquery.admin on your project. Your administrator grants these through IAM.
Can I buy commercial data through BigQuery sharing?
Yes, where a provider has a commercial listing. Public listings can be free or commercial. A commercial listing either asks you to request access, after which the provider contacts you, or is integrated with Google Cloud Marketplace, where you choose Purchase via Marketplace and set a subscription plan. Google notes that publishers and subscribers of Marketplace listings must be in a supported Marketplace Agency Jurisdiction.
How do I get Fokals data into BigQuery?
Fokals is delivered direct, by REST API and as bulk files in CSV, JSON or JSON Lines. You download the files or call the API, land the output in Cloud Storage and load it into your own BigQuery tables, which gives you your own copy and your own history. Licensing is by written agreement for internal use, embedding in a product or redistribution.
Can I keep my own history of a shared dataset?
Not through the linked dataset itself. Google lists snapshots of linked dataset tables as unsupported, and a publisher can restrict copy, clone, export, snapshot and CREATE TABLE AS SELECT. Where egress is allowed, you can query the data into your own tables on a schedule. Where it is restricted, you need the provider to deliver files or an API that you can load.
What this page says about the products it names was checked against their public documentation on 4 October 2026. Product and company names are trademarks of their owners. Fokals is not affiliated with them or endorsed by them.