Data sharing gives another organisation read access to tables you hold, in place, without sending it a copy. The recipient queries the provider's data, and the provider controls the access and can end it. Cloud data platforms and open protocols implement it.
How it differs from a copy
With files or an API, the recipient ends up holding its own copy. With a share, it queries the provider's tables directly, so updates appear as soon as the provider makes them and no transfer job runs.
Two forms exist. A platform-native share, such as Snowflake Secure Data Sharing, works between accounts on that platform, and Snowflake's documentation states that no data is copied between accounts and that shared objects are read-only. An open protocol, such as Delta Sharing, named OpenSharing since June 2026, is built so that a recipient can read a share from tools other than the provider's own, as the project describes.
What it changes in a licence
The mechanism does not replace the agreement. Settle four points:
- Revocation: the provider can end access, but tables the recipient built from the share stay with it, so the licence must say what happens to them.
- Derived data: query results and tables created from a share are where the limits on embedding and redistribution apply.
- Revision: a share shows the provider's current tables, so if the provider corrects old rows, a query you ran last month gives a different answer today.
- Evidence: record what was shared and when, because no file arrives that you could archive.
In Fokals data
Fokals is delivered direct. The buyer pulls the data through the REST API or bulk exports and loads it into the platform of choice, where the tables are the buyer's own copy under the licence. Daily and weekly datasets are written once and never revised, so a copy you load stays consistent with the source and a query run last month returns the same answer today. See the delivery page for how Fokals data is pulled.
Related terms
- Data marketplace: the venue where shared and delivered datasets are listed and bought.
- Data warehouse: a common home for the tables a share exposes.
- Data licence: the agreement that still governs a dataset received through a share.
- Third-party data: data collected by one organisation and supplied to another, by share or by file.
Frequently asked questions
Is data sharing the same as an API?
No. An API returns records in response to requests, and the caller stores its own copy. A data share lets the recipient run queries against the provider's tables directly, so no copy is made on the way in. The recipient can still create its own tables from the results, which is why the licence, not the mechanism, sets what may be done with the data.
Can a data share be revoked?
Yes. On the platforms that offer sharing, the provider can end access to a share, or to objects in it, and the recipient stops seeing the data. What the recipient has already copied into its own tables stays where it is, so a licence should say what happens to such copies when access ends.
What is the difference between data sharing and a data marketplace?
Data sharing is a delivery mechanism: how one party gives another access to tables. A data marketplace is a venue: where datasets are listed, found and bought. Some marketplace listings are delivered by data sharing, but a share can also be set up directly between two organisations with no marketplace involved, and many datasets reach buyers by other routes.