Seats Reclaim for Jira — user guide
Last updated: 17 October 2026
What it does
It finds people who are consuming a paid Jira seat without using Jira, tells you what they cost, and removes their product access — reversibly.
It does not delete accounts, and it does not deactivate Atlassian accounts at organisation level. Nothing can do that from inside Jira; the apps that claim to are asking you for an organisation API key created by hand at admin.atlassian.com. Seats Reclaim never asks for one, because removing group membership is what actually frees the seat — and unlike deactivation, it is undoable.
Requirements
- Jira Cloud.
- You must be a Jira administrator. The app acts with your permissions, so if you cannot do something by hand, the app cannot do it either.
- Nothing else. No account, no token, no configuration.
Getting started
Jira → Settings → Apps → Seats Reclaim.
- Group that grants the access. Pick the group people get their Jira seat from. Under the picker the app tells you which products that group actually grants, read from your site. If it says “This group does not grant product access”, removing people from it will not save you anything — and the app says so instead of pretending.
- Consider inactive after. Six months is a good first run; twelve is the usual answer at renewal. Anything under 30 days is refused, so nobody can sweep up a team that was on holiday.
- What a seat costs you. Jira’s API does not expose your price per seat, so this is a question, not a guess. Pick your Jira plan for list prices, or choose “I pay a different price” and type what you actually pay — your discount and your currency give you a number that is yours, not ours.
- Preview. Nothing is changed. On a large group this takes a minute: every person is measured individually, and the shortlist gets four extra safety checks each.
Reading the preview
The green block is what you would get back per month and per year. Then two lists — and every single member of the group is in one of them, with a reason. Nobody disappears silently.
“never — no trace on this site” means exactly that: not a failed lookup, but an actual absence of any assigned, reported, created or logged work in the whole history of the site.
Reasons you may see in the “will not be touched” list:
- Active — worked on Jira issues inside your window.
- App account — an integration, not a person. Removing it breaks something.
- Administrator — holds the Administer Jira permission.
- Protected group — in a site-admin, org-admin or app group.
- Open work assigned — has unresolved issues; removing them would orphan that work.
- Keeps access elsewhere — in another group granting the same product, so removing them here would save nothing.
- Group frees no seat — the group you picked grants no paid product.
- Already deactivated — not consuming a seat anyway.
- That is you — the app never removes the person running it.
- Not measured — a query failed. A failed measurement is never read as “inactive”: nobody loses access because a query failed. Run the preview again.
- Check did not complete — one of the safety checks could not run. The Detail column names which one, and the HTTP status.
Export the preview (CSV) gives you the whole decision — names, reasons, and a stable reason code — to paste into a spreadsheet and send to whoever has to approve it.
Removing access
Confirm, and the app works in small batches, stopping at the very first error. If something goes wrong halfway you get a partial change that is fully recorded and fully reversible — never a half-finished mess with no record.
Right before each removal the app checks again, on the server, that the person is still not an administrator and still has no open work. Minutes can pass between the preview and the button, and in those minutes things change. Anyone caught by that check is left alone and listed in the log with the reason.
Undoing
History → Undo. Everybody goes back into the group.
If only some people could be put back, the log says “partly undone — N people still out” and the button stays, pointed at the remaining ones. It will never tell you an operation is undone when it is not.
Undo works even if your subscription has lapsed. Nothing you have already done is held hostage.
When something does not work
Diagnostics → Run the checks. Every call the app needs, tried right now against your site, with the raw HTTP status. If one is red, that is the reason — not a guess. Send us that table and we can usually answer without a call.
Diagnostics → Write probe. If removals fail, this takes you out of a group you pick and puts you straight back, showing the raw HTTP status of both calls. Pick a group that grants no product access, so the probe changes nothing real.
Frequent answers
“It says 0 people, is it broken?” Probably not. Open the exclusion list: it will tell you whether everyone is active, protected, or unmeasured. Zero is a legitimate answer.
“Why is my colleague Keeps access elsewhere?” They are in a second group that grants the same product. Removing them from this one costs them access and saves you nothing. Remove them from the other group too — or from that one instead.
“The seat count did not go down.” Atlassian’s billed-user count updates on its own schedule. Check again shortly. If the person really is out of every group that grants the product, the seat is free.
“Can I get the removals back after deleting the log?” No. Deleting the log deletes the record of who was removed, and Undo needs it. The app warns you before doing it.
Contact
Open a ticket on our help portal or write to support@pieranlab.cloud. Monday to Friday, 09:00–18:00 Italian time, first response within 2 business days — details in the support policy.
Please include the Diagnostics table and, if it is about a specific preview, the exported CSV — with the columns you are allowed to share.
See also the privacy policy and the security statement.