The right number of Rovo credits per user is whatever your team's measured usage consumes, not the plan allowance divided by 10. A Jira Premium seat's 70 credits covers about 7 Basic chat messages a month. On Atlassian's current rate card, Search and Summaries are free, every Basic chat message (Quick Answers) costs 10 credits, and Think deeper, advanced agents and the Jira Coding Agent cost a variable amount.
To size your Rovo credit pool reliably, export your organization's usage since August 31, 2026, group users into light, regular and heavy profiles, and compare that demand with the allowance your seats provide. With extra usage billing starting on December 3, 2026, now is the time to do it.
What's new in this update (November 2026)
Empyra is an Atlassian Platinum Solution Partner with 15+ years of Atlassian experience, 80+ certified consultants and 1,000+ customers, including Rakuten, Victoria's Secret, Houston Community College, Sally Beauty and UC Office of the President. This is the sizing method our Atlassian license management services team uses ahead of the December 3, 2026 billing change. Every rule in this guide is checked against Atlassian's How Rovo credits work page, last updated by Atlassian on October 22, 2026.
For the fundamentals of what credits are and how overage works, read our Atlassian Rovo pricing guide. This post turns those rules into a number for your organization.
December 3, 2026 is the day usage-based pricing starts to cost money. From that date, Atlassian bills extra usage for Rovo credits: any usage beyond your organization's pooled monthly allowance is charged at $0.01 per credit. Atlassian states that extra usage is on by default, so work keeps running and billing continues up to any spending cap you set. If extra usage is off, billable features pause until the allowance resets.
Two other changes land on the same day:
Before December 3, do four things:
Every paid Jira, Confluence, Service Collection and Teamwork Collection cloud subscription includes a per-user monthly allowance of Atlassian Rovo credits.
|
Offering |
Standard |
Premium |
Enterprise |
|
Jira |
25 |
70 |
150 |
|
Confluence |
25 |
70 |
150 |
|
Loom (new) |
25 (Business) |
70 (Business + AI) |
150 |
|
Service Collection |
250 |
700 |
1,500 |
|
Teamwork Collection |
250 |
700 |
1,500 |
Rovo credits per user per month. Source: Atlassian, How Rovo credits work.
|
Users |
Jira or Confluence Standard (25) |
Jira or Confluence Premium (70) |
Jira or Confluence Enterprise (150) |
Teamwork Collection Premium (700) |
|
50 |
1,250 |
3,500 |
7,500 |
35,000 |
|
200 |
5,000 |
14,000 |
30,000 |
140,000 |
|
500 |
12,500 |
35,000 |
75,000 |
350,000 |
|
2,000 |
50,000 |
140,000 |
300,000 |
1,400,000 |
Rovo credits per month for one product. Add each product your users hold, because allowances stack in one pool. At 10 credits per Quick Answers message, a Jira Premium user's share covers 7 messages a month.
Three rules decide how these allowances behave in practice:
Most published estimates of Rovo credits per user divide the allowance by a flat 10 credits per interaction. On Jira Premium, that gives 70 ÷ 10 = 7 interactions per person per month. On the current rate card, this shortcut fails in four ways.
Rovo Search, Definitions and Summaries cost nothing. A user who mostly searches and summarizes may barely touch their share.
Atlassian bills each user message in chat as a separate event. A five-message thread in Basic chat, which Atlassian's credits page bills as Quick Answers, costs 50 credits, not 10.
Several features are billed at variable rates:
Think deeper mode in Rovo Chat
Deep research mode, which can run for up to 15 minutes on a single request. Atlassian's credits page doesn't list it by name, so check its line in your usage export and treat it as one of your most expensive features.
advanced agents with model selection
Agents in Jira and the Jira Coding Agent
Confluence slides and Create with Rovo
Request Resolver, Alert Intelligence and the Self-service Agent
Each one's cost depends on reasoning steps, search volume, tools invoked and model power. Atlassian's rate card groups Premium work into three estimated ranges:
|
Task type |
What it involves |
Estimated Rovo credits per task |
|
Light |
Few sources, light reasoning, one output (e.g. AI-generated Confluence slides) |
10–200 |
|
Medium |
Many sources across time, structured reasoning, two or more outputs (e.g. release notes plus stakeholder updates) |
201–500 |
|
Heavy |
Broad aggregation, deep reasoning, multiple outputs (e.g. cross-team agents tracking roadmap risk) |
500+ |
Source: Atlassian's Rovo credits task-type rate card on How Rovo credits work, as of 22nd Oct, 2026.
Put that next to the allowance table. A single light Premium task can use more than a Jira Premium user's entire 70-credit monthly share. These are estimates, which is why your own usage export is the only reliable way to price Premium work for your team.
Several sources draw on the same pool without adding a licensed user:
One more catch: Atlassian applies the new rate card only to usage from August 31, 2026 onward. Averages built on earlier usage, including your own, measure the old model.
At Team '26 Europe in October 2026, Atlassian announced several features that make Rovo easier to use, and easier usage means more Rovo credits per user. Atlassian hasn't published per-feature rates for most of them yet, so plan for them as their own lines and watch the Features tab in your usage dashboard.
|
Feature |
Status |
How it uses Rovo credits |
What to do |
|
Code Reviewer (Rovo Dev) |
Moves to Rovo credits on Dec 3, 2026 |
Metered in Rovo credits; no per-review rate published yet |
Add a code review line to your forecast (see Step 3). Keep reviews on "when a PR is created", not on every commit |
|
Rovo Work |
Announced |
Longer, multi-step tasks that can run for minutes to hours and run with the starting user's permissions. Rate not published |
Treat it as heavy Premium work until Atlassian publishes a rate |
|
Scheduled chats |
Announced |
Recurring chats run without anyone sending the message. Rate not published |
Review who schedules what, and treat each scheduled run as at least one chat message |
|
Rovo desktop and mobile apps |
Desktop shipped, mobile rolling out |
Same chat modes as the web, so the same rates apply |
Expect chat volume to rise as access gets easier |
|
Agents in Jira (Rovo, Claude, Cursor, Copilot) |
Available in Jira Cloud |
Agents in Jira and the Jira Coding Agent are variable Premium. Third-party agents are billed by their own vendors, but their Teamwork Graph calls can use Rovo credits |
Track agent rows separately in your export |
|
Planner |
Early access |
Beta features don't count against your allowance unless Atlassian says otherwise |
Keep it as a "watch" line for when it reaches general availability |
If you want visibility across all of this, Atlassian also announced AI spend tracking in the Strategy Collection (open beta) and Model Fit in DX, which flags agent runs that could use a cheaper model. To see where your credits go feature by feature, read our guide to Atlassian AI costs [link the anchor 'Atlassian AI costs' to the AI Costs blog URL]. For the bigger picture of how agents now work across Atlassian, read our guide to Atlassian AI agents and AMP.
Organization admins can see usage in Atlassian Administration → Insights → Platform usage → Rovo credits. The dashboard shows:
For sizing, you need the raw data, so select Export data. The CSV records feature, user, agent, timestamp and credits consumed per event. It covers up to three months or one million events, whichever comes first, and the download link expires after 24 hours.
Filter the export to events dated August 31, 2026 or later. Then build three summaries:
If Rovo is new to your organization and you have no usage data yet, start with our calculator to estimate Rovo credit usage by team size, then run a short pilot with a representative group to replace that range with real numbers.
Rovo credits per user vary too much from person to person to model one by one. Profiles let you forecast the pool as a whole.
Now compare supply (your seats' allowances) with demand (your profiles plus non-seat usage).
Rovo credit sizing worksheet
|
Line |
How to calculate |
Example: 60-user dev team on Jira Premium |
|
|
A. Monthly supply |
Σ (licensed users per product × plan allowance) |
60 × 70 = 4,200 |
|
|
B. Light demand |
light users × median credits |
20 × 40 = 800 |
|
|
C. Regular demand |
regular users × median credits |
30 × 420 = 12,600 |
|
|
D. Heavy demand |
heavy users × median credits (Basic + Premium) |
10 × 420 = 4,200 + Premium |
|
|
E. Non-seat demand |
automation agents + help-center agents + enriched Teamwork Graph / MCP calls |
0 in this example |
|
|
F. Monthly demand |
B + C + D + E |
17,600 + Premium |
|
|
G. Expected gap |
F − A (a negative number means spare allowance) |
13,400 + Premium |
|
|
H. Expected overage cost |
G × $0.01 per credit |
about $134/month + Premium |
|
Line H uses Atlassian's current list price for extra usage: $0.01 per credit, or $10 per 1,000 credits.
Don't have a full export yet? The Rovo credits calculator gives a monthly planning range by team size and usage profile. Use it for a first estimate, then refine it with this worksheet. It covers Rovo credits only, so size automation steps separately.
A pool sized for this month's demand will usually be short within a quarter. Plan for three kinds of growth:
Pick a headroom percentage that matches the variance you actually saw, and apply it to line F. Use this padded figure for your spending cap. Keep line H unpadded, so finance sees expected spend while the cap sets the ceiling.
If line G is positive, you have three options.
Keep extra usage on, with a spending cap. Atlassian's documentation is inconsistent on whether extra usage is enabled by default, so confirm the setting in Atlassian Administration first. Set the cap at (padded demand − supply) × $0.01 in Atlassian's extra usage settings. This suits small or uncertain gaps. Admins get alerts at 80% and 100% of the allowance either way.
Buy prepaid usage packs. If the gap is steady, packs make the budget predictable, and Atlassian offers volume discounts on annual packs. Pack sizes and discounts depend on your agreement, so ask your Atlassian partner for a quote.
Move to a higher-allowance collection. Service Collection and Teamwork Collection include 10× the Rovo allowance of standalone Jira or Confluence on the same plan. Compare the license price difference against twelve months of line H. Note that list prices change on October 13, 2026, as covered in our Atlassian cloud price increase 2026 breakdown, so quote the collection at the rate you'll actually renew on.
One caution when you clean up licenses at the same time: reclaiming inactive seats lowers license cost but also shrinks your pool, because every seat carries its allowance. Run your Atlassian license optimization and your Rovo sizing as one exercise, not two.
Want a second pair of eyes on your usage export? Empyra's license management team will read your export, build the worksheet with you and recommend a spending cap before December 3.
These scenarios show how Rovo credits per user play out in three common setups. They use Atlassian's published allowances and the $0.01 overage rate. The usage figures are illustrative assumptions chosen to show how the method behaves. They are not benchmarks or client data. Premium usage is left as a line to fill from your own export.
700 Rovo credits per user looks generous, but it can be outrun by ticket volume rather than headcount. Model routing and customer-facing agents from your request forecast. Our Jira Service Management consulting team often starts here, because these are the AI features ITSM teams switch on first.
Add code reviews from December 3. Say 40 developers open 8 pull requests each a month, so 320 reviews. Atlassian hasn't published a Rovo credit rate per review yet. Its Rovo Dev documentation says 2,000 Rovo Dev credits typically cover 20 to 75 reviews, or about 27 to 100 credits per review. If one review costs the same in Rovo credits (our assumption, not Atlassian's), 320 reviews add 8,500 to 32,000 credits, or about $85 to $320/month more.
At 70 Rovo credits per user, two Quick Answers messages a day already use six times the allowance. Premium work and code reviews widen the gap fast. This is where flat-rate estimates mislead most.
This example has 250 people holding 200 Jira Premium seats and 150 Confluence Premium seats.
New with Loom: if 100 of these people also have Loom Business + AI seats, they add 100 × 70 = 7,000 credits to the pool. Supply rises to 31,500, the gap falls to 16,500 and expected overage to about $165/month. Count every Loom seat in line A.
The real decision here is a cap plus usage packs versus a move to Teamwork Collection, which raises the Premium allowance to 700 per user. You can only make that comparison with a real license quote placed next to twelve months of projected overage.
This example has 250 people holding 200 Jira Premium seats and 150 Confluence Premium seats.
The real decision here is a cap plus usage packs versus a move to Teamwork Collection, which raises the Premium allowance to 700 per user. You can only make that comparison with a real license quote placed next to twelve months of projected overage.
As an Atlassian Platinum partner, Empyra covers the full picture:
Sizing and cap strategy. Our license management team builds your worksheet from your real export, sets spending caps and weighs usage packs against a collection upgrade.
Reducing consumption. Our Atlassian consulting services team reworks automation rules and agent configurations that consume credits without adding value.
AI rollout and governance. Our Atlassian AI consulting practice helps teams choose between Quick Answers and Think Deeper, set usage ownership, and roll out agents with governance in place.
The flat "allowance ÷ 10" estimate is simple, but it undercounts conversational chat and overcounts free usage. A sizing built on your own export, split into user profiles plus non-seat demand, gives you a number you can defend before December 3, 2026.
The fastest next step: Book a Rovo usage audit with Empyra's Atlassian team.