Data retention
What the retention policy deletes and keeps, how to read the retention view and the upcoming deletions list, and how to ask for a change.
Retention keeps a workspace from growing without limit. When a retention period is set, old finished runs and old dataset versions that are no longer active are deleted after that many days. The active version of every dataset is always kept, so what your team reads is never affected.
By default a workspace has no retention period: nothing is deleted automatically until one is set.
- Run finishesor a dataset version is superseded
- Period elapsescounted in days from that point
- Sweeplisted under Upcoming deletions first
- Deletedlogs, fetches, screenshots, old rows
Where to find it
Settings → Content → Data Retention. The card is headed Retention policy — Terminal runs and inactive dataset versions are deleted after the configured period. The active dataset version is always preserved.
Workspace members see a read-only view:
| Line | Meaning |
|---|---|
| Automatic deletion is disabled. | No retention period is set. Nothing expires. |
| n days | The retention period. Items are deleted once they are older than this. |
| Only a platform admin can change this workspace policy. | The policy is set by Jsonify on your behalf; see Requesting a change. |
Upcoming deletions
Below the policy, Upcoming deletions lists Items due in less than 30 days, including items waiting for the next daily sweep. Press Refresh to reload it.
| Column | What it shows |
|---|---|
| Type | Run or Dataset version |
| Item | A link to the run or dataset version, with the pipeline or dataset it belongs to |
| Retention basis | The date the period is counted from: when the run finished, or when the version was published |
| Scheduled deletion | Next daily sweep for anything already due, then Tomorrow, In n days, and the date |
The list shows 25 items per page. With no policy it reads Data retention is disabled.; with a policy and nothing due, Nothing is due in the next 30 days.
If a run or version you need is listed, export what you need before the date. A dataset version can be pinned as the active version to keep it; see Dataset versions.
What expires
Under a retention period, two kinds of item are deleted:
Finished runs — runs that completed, failed or were cancelled, counted from when they finished. Deleting a run removes its logs, outputs, step previews, the pages it fetched, its activity, the screenshots and other evidence behind row provenance, and any files stored with it. A run whose fetched pages are still relied on by a newer run is skipped until that run also expires.
Inactive dataset versions — every published version of a dataset except the active one, counted from when the version was published. The rows in those versions are deleted.
What does not expire
Retention is not the same as deleting a workspace. Deleting a workspace removes everything at once and is documented on Workspaces.
When deletion happens
Items are removed by a background sweep after they become due; the settings page refers to it as the daily sweep. An item shown as Next daily sweep is already past its period and goes in the next pass. Changing the period takes effect from the next sweep, so shortening it can make many items due at once; the Upcoming deletions list shows exactly what.
Requesting a change
Members cannot set or change the period themselves. To turn retention on, change the number of days, or turn it off, contact the Jsonify team at support@jsonify.com with the workspace name and the period you want. A member can check the result on the Data Retention tab; the read-only line updates as soon as the policy is applied.
Retention changes are not retroactive in the other direction: once a run or version has been deleted by a sweep, it cannot be restored. Ask for a longer period before it is needed rather than after.
Why keep runs at all
Runs are worth keeping for as long as you might want to answer “where did this value come from?” Every published row links back to the run, step and page that produced it, with a screenshot of the page as collected. Once the run is deleted, the row remains in its dataset version but the origin trail behind it is gone. Teams that audit their data usually keep at least a full reporting cycle of runs; teams that only need the current numbers can keep far less.