Skip to Content

Jira Cloud backup, Atlassian, a Marketplace app, or nothing at all

September 11, 2026 by
ONETEEM

Three real options, the published numbers behind each one, and a recommendation you can match to your own user count.

The question behind the question

A team lead called us the week after she had connected Jira to another system through an integration with write access. The integration worked. That was the problem. She had watched it update a batch of fields in seconds and realised the same mechanism could rewrite a few thousand work items just as fast and just as quietly. So she did not ask whether Jira has a backup. She asked whether she could go back to yesterday.

That is a different question, and a better one. A backup is a schedule. Recovery is what you actually get back, and how long you wait for it.


Option one, Atlassian Backup and Restore

The native product covers Jira, Jira Service Management and Confluence Cloud. Backups run daily or weekly according to a policy, and policies can be managed centrally across several sites. Recovery means a full site restore from a hot backup in twelve hours or less. Atlassian is direct about the purpose, which is recovery after customer caused incidents such as accidental deletion or data corruption.

Four things to know before you put it in a budget.

It is a paid add on, priced per app and per user, available only on Premium or Enterprise plans with annual billing. On Premium you buy it per site. On Enterprise it covers the sites belonging to that Enterprise plan.

There is no self service purchase. It comes through Atlassian sales, Atlassian support or a partner, and it can be bought off cycle, charged pro rata.

The size limits at general availability are hard ones. Jira including Jira Service Management is supported up to 300 GB of app data and 7 million attachments, Confluence up to 32 GB and 7 million attachments. Above those limits it cannot be purchased. It is also not available in the Atlassian Government environment.

And only the entities in the published tables are backed up and restored. For Jira that list is broad, covering users and groups, company managed and team managed projects, archived projects, project configuration, workflows, screens, fields, comments, attachments, work logs, links, filters, boards and automation. The gaps are specific. Jira Product Discovery projects sit outside it, along with product access settings, application links, the Jira audit log, Goals, email channel authentication and customer notification logs. In Jira Service Management the exclusions bite harder, covering knowledge bases, customer notifications, raw incoming emails, alerts and on call schedules, and several Assets features such as the list view and import history. Third party app data appears nowhere on the supported list, and since only listed entities are protected, we read that as not covered. That is a conclusion from the published list rather than a sentence Atlassian writes.

The price is published as an annual estimate, where the Jira family means Jira and Jira Service Management together. At 1,000 users it is about 7,000 US dollars per year for the Jira family and about 5,000 for Confluence. At 5,000 users, 30,000 and 21,000. At 10,000 users, 55,000 and 39,000. At 50,000 users, 206,000 and 122,000. Per head that is roughly 7 US dollars per user per year for the Jira family at 1,000 users, falling to about 4 at 50,000. The table begins at 1,000 users. Below that you need a quote, and we will not guess one for you.

Pick this if you are on Premium or Enterprise, you have a thousand users or more, and the scenario that keeps you awake is a site level catastrophe.


Option two, a Marketplace app

These apps exist because a site restore is a blunt instrument. If one automation rule wrecked one project on Tuesday, rolling the whole site back to Monday is a second incident. The vendors sell granularity, and mostly they sell app data coverage with it. Prices are per user and month, the tier up to ten users is a flat monthly fee, tiers are progressive, and an annual subscription costs ten monthly payments.

http://GitProtect.io for Jira, from Xopero Software, is Cloud Fortified. It is free up to ten users, then 4.00 US dollars per user and month from 11 to 100. The vendor lists coverage of projects, users, work items, roles, workflows, comments, attachments, boards, versions, custom fields, votes, audit logs and notifications, and describes granular restore plus disaster recovery, cloud to cloud data mobility, AES encryption and SAML.

Rewind Backups for Jira comes in two editions. Standard starts at 40.00 US dollars flat per month up to ten users, then 4.00 from 11 to 100. Advanced starts at 52.00 flat, then 5.20 from 11 to 100. The listing describes automated daily backups, granular restore of individual items to any backed up point in time, one year of retention, unlimited backup storage, a choice of data residency region including Frankfurt and Zurich, syncing backups to your own AWS, Azure or Google storage, coverage of Jira Service Management Assets and configuration, and configurable alerts with SIEM integration. Jira Service Management and Jira Work Management are included. Which feature sits in which edition is worth checking on the listing before you commit.

Revyz Command Center for Jira, now part of http://Spin.AI and marked as a bestseller, prices both editions identically. Up to ten users is 32.00 US dollars flat per month, then 3.20 from 11 to 100. The listing covers backup and restore of work items, backup and restore of Assets objects, versioning and deployment of Jira and Assets configuration, configuration drift analytics, site cleanup, configuration health reports, an approval workflow for deployments, and protection of third party app data with Xray as the named example. A separate bring your own storage edition also exists.

All three offer a 30 day free trial for teams of any size. If you already run central backup tooling across the estate, HYCU is worth a look as a further vendor here.

Pick this route if you are below a thousand users, or you need to repair one object rather than one site, or a Marketplace app holds data the business runs on.


Option three, no backup product at all

A legitimate choice, and plenty of small teams make it deliberately. Every plan includes a manual site export under Settings, System, Import and Export, Backup manager. It gives you work items and field content, boards and sprints, users and groups, comments, media if you select it, space configuration and Advanced Roadmaps.

What it will not give you is automation rules, third party apps and their data, Opsgenie based features, Assets, app access settings, or Jira Product Discovery views, insights and votes. Automation rules can be exported and imported separately by hand, which is a task somebody has to remember. Exports run at any time, but attachments, avatars and logos can only be included once every 48 hours. Importing a backup means importing into a site, so it is not a route to bringing back a single work item. And since 22 January 2026 the Backup manager only restores from backups that are 30 days old or less.

Pick this if your Jira holds nothing you could not rebuild in a week, and somebody owns the calendar reminder.


What it actually costs

The figures below are calculated from the published Marketplace tiers using the conventions above. Treat them as arithmetic rather than list prices, and confirm your own number with the Marketplace calculator.

Two things fall out of those numbers, and both surprise people.

Below a thousand users the Marketplace apps are the only option with a public price, because the native table does not start there. For a team of 50 or 250 the decision is therefore not native against Marketplace, it is which app, at a cost landing somewhere between 1,600 and 7,400 US dollars a year.

From a thousand users upward the picture inverts. At 5,000 users the native product costs roughly 30,000 against roughly 100,000 for any of the three apps, a factor of three, and it holds at 10,000. The common assumption that Atlassian sells the expensive option and the Marketplace sells the cheap one is exactly backwards at scale.

Which does not settle it. The apps are not losing that comparison on price, they are winning a different one. They restore one object instead of one site, they cover app data, and they work on any plan at any team size. Whether a factor of three is worth giving that up depends on which incident you are insuring against.


Five ways, side by side

The same five routes on the axes that decide the choice.


Which route for which situation


What the difference looks like

One rule went wrong in one project on Tuesday afternoon. This is what each route does about it.


Backup is one of the few IT decisions where the entire value is created before anything goes wrong. And when something does go wrong, the difference that matters is not whether you have a backup, it is whether you can repair one object or only roll back an entire site.

If you want a second pair of eyes on your own numbers, or help testing a restore before you need one, we are glad to look at it with you.

All figures are published prices or calculations from published tiers at the time of writing, and the Marketplace calculator with your real user count is the final word.

Sources

ONETEEM is an Atlassian Marketplace Partner, which is worth knowing when you read a comparison like this one. Prices checked in September 2026.

Say hi 👋🏼 to "Atlassian Service Accounts"