Atlassian usage-based pricing: Decisions to make by December 3

When does this happen?
For most meters, billing for usage beyond included allowances begins December 3, 2026.

What does it mean for you?
The change impacts your organization in two ways:
1.  If extra usage remains enabled, you may incur charges after allowances and any purchased capacity are exhausted
2.  If extra usage is disabled or capped too tightly, specific capabilities, including business-critical workflows, can pause.

The right decision depends on what is consuming capacity, which workflows rely on it, and what business value the usage creates. A blanket decision to allow or block overage can solve one problem while creating another.

What should you do before December 3, 2026? 

• Check usage: Open Atlassian Administration and review Insights, then Platform usage, for every applicable meter

• Assign ownership: Confirm who owns organization administration, billing administration, budget decisions, and each operational use case

• Configure limits: Check whether extra usage is enabled and whether each meter has an appropriate limit

• Map impact: Identify critical workflows (automations, pipelines, AI) that would pause if limits are reached, especially automations, pipelines, imports, external context calls, and customer-facing AI

• Check hidden drivers: Review sandboxes, loops, high-frequency rules, and bulk imports, duplicate Assets objects, failed pipelines, and retained packages

• Model costs: Compare optimization, usage packs, pay-as-you-go, plan changes and Collections against forecasted growth

• Evaluate toolchain: Audit overlapping AI and automation capabilities and tools across your broader stack

Atlassian usage is one part of your larger AI investment decision. You may be paying for overlapping assistants, agents, automation, model access, and context retrieval across Atlassian, GitHub, GitLab, cloud platforms, and other tools. These capabilities are often connected within the same software delivery process, and usage in one system can trigger activity or consume capacity in another.

Before setting limits or buying more capacity, identify and map the business-critical processes that depend on these tools. That means understanding which systems and integrations support each process, how usage moves across them, who owns the workflow, and what happens if one capability pauses. Without that view, a reasonable cost-control decision in one platform can interrupt approvals, deployments, service interactions, data updates, or other work elsewhere in the SDLC.

Eficode can help you baseline usage, identify operational risk, optimize agents and automations, compare licensing and capacity options, and evaluate Atlassian usage in the context of your wider AI and development tool investment.

• Most eligible paid plans include monthly allowances for applicable meters

• Allowances are pooled across the organization, so teams and sites can draw from the same balance

• Extra usage is enabled by default for most meters and is billed after the included and prepaid capacity is used

• Administrators can monitor usage, set limits, and receive alerts at 80 percent and 100 percent of allowance

• If a meter pauses, the underlying subscription remains active, but the affected capability can stop

• Atlassian usage should be evaluated alongside overlapping AI, automation, and developer tools across the wider toolchain

Which products are eligible to contribute Rovo credits?

Only paid subscriptions for the following products contribute monthly Rovo credits to the organization pool.

• Teamwork Collection

• Service Collection / Jira Service Management (JSM)

• Jira

• Confluence

(Products like Jira Product Discovery (JPD), Bitbucket, and Free plans do not contribute Rovo credit allowances directly, but users on those products can draw from the organization's shared pool).

Which products and plans provide the most credits?

• Teamwork Collection (Highest Allowance): Collections provide the largest per-seat allowance; up to 10x higher allowances than standalone apps on the equivalent edition tier.

• Higher Edition Tiers (Enterprise & Premium): Allowances scale directly by tier (Enterprise > Premium > Standard). Upgrading edition tiers significantly increases the monthly per-user credit allocation.

• Multi-Product Stacking: Subscribing to multiple eligible standalone apps (e.g., Jira + Confluence) stacks each product's allowance into the single organization-wide pool.

• September 1, 2026: Expanded meter visibility became available in Atlassian Administration

• October 26, 2026: Monthly Bitbucket allowances move from workspace-level to organization-level pooling

• December 3, 2026: Extra-usage billing begins for most expanded meters; annual Bitbucket plans begin paying for usage beyond included allowances

A pricing estimate is useful, but it will not reveal every operational dependency. These are the scenarios to check before setting a limit or buying more capacity.

One team can consume capacity, another team expects

Allowances are pooled at the organization level. A high-volume site, app, team, or use case can draw from the same meter as other teams. Confirm ownership and budget responsibility before usage grows.

Automation is measured by steps rather than complete runs

Every executed trigger, condition, action, branch, and loop can count. A high-frequency rule can consume significant capacity even when it rarely changes a work item. Enterprise plans also move from unlimited automation to defined allowances.

A pause can create missing business actions

If automation stops, events that occur during the pause are not automatically queued and replayed after capacity returns. The impact may be missing approvals, notifications, updates, or integrations rather than a simple delay.

The people who receive alerts may not own the workflow

Organization and billing administrators receive allowance and limit alerts. They may not know which automations, pipelines, imports, or AI services are business critical, so escalation ownership needs to be agreed in advance.

Rovo credits can be used outside an Atlassian screen

External tools such as ChatGPT, Claude, Cursor, or an internal copilot can consume credits when they retrieve enriched Atlassian context through the Rovo MCP server or Teamwork Graph interfaces.

Basic and advanced AI interactions do not cost the same

Basic billable Rovo interactions use a fixed credit amount, while more intensive experiences, such as Think Deeper, use a variable amount. Agent design, reasoning mode, repeated execution, and cross-system context can all change consumption.

Sandbox work can draw from production capacity

Rovo credits and automation steps used in sandboxes count against the organization allowance. Testing, demonstrations, and development activities can therefore affect the same pool used by live workflows.

Assets do not behave like a monthly action allowance

Assets measure active stored objects. Imports, APIs, integrations, and automations can add objects quickly, and the count remains high until objects are removed.

Successful service outcomes are chargeable from the first resolution

Customer Service Management AI agent resolutions currently have no allowance included. A live request resolved end-to-end without a person is chargeable; requests handed to a human are not counted as successful AI resolutions.

Bitbucket exposure depends on more than build volume

Annual plans begin paying for usage beyond included allowances on December 3. Review hosted build minutes, recurring and unsuccessful pipelines, Git LFS growth, package retention, and network transfer separately.

Start with the meter, then connect consumption to the workflow and its owner. Five questions will expose most of the immediate risks.

1. Which meters apply to your products and plans?

2. What does Atlassian Administration show for current usage, forecast, and reset date?

3. Which teams, sites, sandboxes, automations, agents, integrations, and external AI tools contribute to that usage?

4. What stops if extra usage is disabled or the limit is reached?

5. Who owns the budget, receives the alerts, and decides whether to optimize, cap, or fund additional usage?

This sequence separates a manageable cost question from an operational risk. It also prevents a billing administrator from making a reasonable financial decision without seeing the effect on a customer-facing or business-critical workflow.

Start with the value and dependency of the use case, then compare the controls available for that meter.

December 3 is close enough that organizations should establish their baseline and decision owners now. Waiting for the 80 percent or 100 percent alert leaves less time to determine what is driving usage, test an optimization, secure a budget, or protect a critical workflow.

A review now gives you a clearer cost range, safer guardrails, fewer operational surprises, and stronger evidence for where AI and automation investment should continue across your SDLC.