Atlassian Data Center to Cloud: the true migration cost.
Moving from Atlassian Data Center to Cloud typically raises annual subscription cost 20 to 40 percent before a single service hour is bought. This note prices the whole move — the per-user subscription delta, migration services, app re-licensing, internal labor, dual-running, and the end-of-sale timing that removes the option to stand still.
The Cloud tier price is the smallest line in an Atlassian Data Center to Cloud move. The budget is won or lost on migration services, app re-licensing, internal labor, and dual-running — and on capturing Atlassian's migration incentives and a written uplift cap in the same deal, before the baseline price is set.
01 Key findings
The subscription rises 20–40% for most estates. Converting a perpetual-license-plus-maintenance Data Center estate to per-user Cloud tiers usually lifts annual subscription cost before any services — and the per-user model scales linearly, so large estates see the steepest rise.
Five components, not one. A complete budget covers subscription change, migration services, app re-licensing, internal labor, and productivity or downtime. The two most often omitted — app re-licensing and internal labor — are the two that most often blow the number.
Customization is the cost multiplier. A vanilla Jira and Confluence estate migrates cheaply with the Cloud Migration Assistant; a heavily customized one can cost several times more, because every bespoke workflow, field, and integration must be reviewed, mapped, and tested.
The move is effectively one-way. Atlassian has ended new Server sales and is steering roadmap and innovation to Cloud. That removes the option of standing still and makes the migration a moment of genuine commercial weight, not a routine transaction.
The migration deal sets the price for years. Atlassian is motivated to close the move and will offer discounts and dual-running grants. Securing the incentive and a renewal uplift cap in the same agreement is the difference between a one-time saving and a durable one.
02 Why Cloud costs more than the tier price
Data Center is an annual license tied to a user tier, often run on infrastructure you already own. Cloud is a per-user subscription on Standard, Premium, or Enterprise, billed for every licensed user every year. For many estates the per-user Cloud cost exceeds the Data Center license plus the infrastructure it replaces — the source of the 20 to 40 percent increase.
The increase is not universal. Small estates that ran Data Center on costly infrastructure can come out flat or cheaper on Cloud once hosting, patching, and administration savings are counted. Large estates with thousands of users frequently see the steepest rise, because the per-user model scales linearly while the old license tiers flattened at the top. Model your own estate, and net the infrastructure and administration savings against the subscription rise to see the real change.
03 The five cost components
A complete migration budget has five parts, and leaving any one out produces the classic underestimate. The table sets them out with the typical driver of each.
| Cost component | What it covers | Main driver |
|---|---|---|
| Subscription change | Cloud per-user tier versus Data Center license | User count and tier |
| Migration services | Data move, validation, cutover | Estate size and complexity |
| App re-licensing | Marketplace apps re-subscribed on Cloud | Number and type of apps |
| Internal labor | Project, testing, training, change management | Customization and integrations |
| Productivity / downtime | Cutover disruption and ramp | Cutover approach and timing |
The two components most often missed are app re-licensing and internal labor. Marketplace apps are licensed separately on Cloud and frequently cost more there; a heavily extended Jira or Confluence estate can carry an app bill that doubles at migration. Internal labor — the time of administrators, project managers, and testers — is real cost even when it never reaches an invoice, and it scales with how customized the estate is.
04 Migration budget composition
Buyers fixate on the tier price because it is the visible line, but it is rarely the largest. The bars show an illustrative first-year migration budget for a mid-size, moderately customized estate — the shape, not the absolute figures, is the point.
Services, apps, and labor together dwarf the tier line. A heavier customization or app footprint pushes services and re-licensing higher; a clean, rationalized estate collapses them. The single most effective cost lever is shrinking the footprint you move before you price the subscription.
05 Services, apps & hidden costs
The data move ranges from a near-automated lift for a clean, standard estate to a multi-month project for a large, customized one. Atlassian's Cloud Migration Assistant handles standard projects, users, and content. The expense sits in the non-standard material: complex permission schemes, custom fields and workflows, large attachment volumes, and integrations that must be rebuilt against Cloud APIs.
Marketplace apps are a cost line in their own right. On Data Center they are licensed to the user tier; on Cloud they are per-user subscriptions that bill alongside the core products. Many cost more on Cloud, and some Data Center apps have no Cloud version at all, forcing a replacement that carries its own migration and learning cost. Inventory every app, confirm which are used, check each has a priced Cloud version, and drop the rest before migrating.
Migrating the estate as-is doubles the bill. Every dormant project, unused app, and legacy customization carried into the per-user Cloud model now bills or adds services cost. Audit and rationalize users, apps, and customizations before the move, not after — each item removed is one less to map, test, and re-license.
06 Timing and Data Center end-of-sale
Atlassian has ended new Server sales and is steering its roadmap toward Cloud, so the question is less whether to move than when and on what terms. That reality cuts both ways: it removes the option of standing still indefinitely, but it also makes the migration a moment of genuine commercial weight, where the terms negotiated govern cost for years.
Dual-running is the timing cost most often underestimated. Few large estates cut over in a single weekend; most run Data Center and Cloud in parallel while content is moved, validated, and users are trained — paying for both platforms at once. Atlassian often grants a free or discounted dual-running period, but its length and coverage are negotiable, and a window too short forces either a risky big-bang cutover or paying twice.
Do not let end-of-sale pressure set the price. The deadline is real, but a rushed, under-prepared move accepts the standard Cloud uplift and forfeits the migration incentives on offer. For regulated estates, let compliance — not the sales pitch — set the tier: map residency, security, and audit needs to the specific Cloud tier that satisfies them, rather than buying Enterprise across every user for a feature a subset requires.
07 Migration cost framework
Four inputs decide whether the migration is controlled or a budget surprise. Work them before any commitment.
Footprint & customization
Rationalize users, apps, and customizations first. Every workflow, field, and app removed is one less to map, test, and re-license — and a smaller subscription to boot.
Tier matched to need
License to the tier compliance and function actually require, only for the populations that need it. Over-tiering the whole base on Premium or Enterprise is a common, expensive error.
Incentives & uplift cap
Atlassian offers Cloud discounts and dual-running grants to close the move. Secure the incentive and a written renewal uplift cap in the same agreement, when leverage is highest.
Cutover & dual-running
Phase the cutover to limit disruption, but size the dual-running window to the plan. Decide and cost both before signing, not mid-project.
08 Our recommendation
If Data Center ran on costly infrastructure, Cloud may land flat or cheaper once hosting and admin savings are counted. Build the five-part model for your own numbers rather than trusting the 20–40% average.
Customization is the multiplier. Audit workflows, fields, integrations, and apps and cut the dead weight before you move — you migrate a smaller, cheaper footprint and emerge with a platform that is easier to govern.
The per-user model scales hard against you at volume. Treat the move as a negotiation: capture migration incentives and a dual-running grant, and lock a written uplift cap so the baseline does not rise unchecked every renewal.
09 Migration timeline
The sequence that controls cost puts preparation and negotiation ahead of the technical move — not the other way around:
Prepare, then negotiate, then cut over Recommended
Rationalize the estate, build the five-part cost model, and match the tier to need. Negotiate incentives and the uplift cap while Atlassian is motivated to close, then phase the cutover with a costed dual-running window.
Cut over, then reconcile Weaker
Treat the move as purely technical, migrate as-is, and accept the standard price. Complexity, dormant apps, and an uncapped uplift all carry into the per-user model — and are far costlier to unwind once the baseline is set.
Plan the move on real numbers
We model the full migration cost and negotiate the incentives and uplift cap before you commit.
The Licensing Edge
Weekly vendor and licensing intelligence for enterprise IT leaders. 3,000+ subscribers.