Atlassian pricing

About

Atlassian is a publicly traded (NASDAQ: TEAM) enterprise software company, founded in 2002 by Mike Cannon-Brookes and Scott Farquhar, built around team collaboration and software-development tooling. Its portfolio anchors on three products — Jira (project & work tracking), Confluence (team knowledge base/wiki), and Jira Service Management (IT service management) — alongside Bitbucket, Trello, Loom, and its newer AI layer, Rovo. Atlassian scaled for most of its history on a low-friction, self-serve go-to-market rather than a traditional enterprise sales force, relying on free trials and bottom-up adoption inside engineering and IT teams before layering on sales-led motions for its largest accounts.

Atlassian serves everyone from small teams just getting started (its Free tier caps at 10 users for Jira/Confluence and 3 agents for Jira Service Management) up to enterprises running 100,000+ licensed users across as many as 150 isolated sites, competing against Monday.com and Asana in work management, ServiceNow and Freshservice in IT service management, and Notion and SharePoint in knowledge management. Its most consequential recent strategic move is Rovo, an AI layer (search, chat, and agents) rolled out across its Cloud product line as a bundled, credit-metered add-on rather than a stand-alone purchase — layering Atlassian’s AI monetization on top of its long-standing per-seat subscription core.

Pricing summary : How Atlassian prices Jira, Confluence & Jira Service Management

Atlassian uses a hybrid per-seat + AI credit overlay model with 4 dimensions:

1. Per-user/per-agent seat tier: Each product (Jira, Confluence, Jira Service Management) is sold on its own Free → Standard → Premium → Enterprise ladder, priced per user (per agent for Jira Service Management), with Enterprise always quoted and billed annually only.
2. Progressive seat-count discounting: The displayed per-user rate isn’t flat — Atlassian publishes banded monthly rates (Jira Standard runs USD 9.05/user for 1–100 users down to USD 2.45/user above 90,000) so the effective blended price falls as headcount grows; a 350-user Jira Standard site nets a blended ~$7.91/user/month.
3. Rovo AI credits, bundled by tier: Every paid Standard/Premium/Enterprise plan bundles a fixed Rovo AI credit allowance (25/70/150 credits per user per month respectively) plus an indexed-object cap for Rovo Search — Free tiers get no Rovo access at all.
4. Per-resolution AI agent overage (Jira Service Management only): Jira Service Management’s Premium and Enterprise plans separately meter Rovo’s virtual customer-service agent at $1 per AI-handled resolution, on top of the seat fee and Rovo credit pool.

What makes this different: Atlassian bundles its core AI monetization as a credit allotment inside existing per-seat tiers rather than selling Rovo as a stand-alone SKU — the one exception is Rovo Dev, a separate AI coding agent sold per developer seat with its own credit pool and per-credit overage.

Pricing by product

Jira (Cloud plans)

*Atlassian publishes banded monthly per-user rates for Jira Standard/Premium (e.g. Standard: USD 9.05/user for 1-100 users, USD 6.40/user for 251-1,000, down to USD 2.45/user above 90,000) - the plan-card price is the blended average at whatever team size the buyer slider is set to, not a single flat rate.

Confluence (Cloud plans)

Jira Service Management (Cloud plans)

Sales motions across products: PLG / self-serve for Free, Standard, and Premium across Jira, Confluence, Jira Service Management, and Rovo Dev; sales-led for every Enterprise tier (annual-only, contact sales).

Rovo & Rovo Dev (AI add-on pricing)

• Rovo (Search, Chat, Agents, Studio) is bundled automatically into any paid Standard/Premium/Enterprise Cloud subscription of Jira, Confluence, or Jira Service Management, and is not sold as a separate SKU for those apps. Credit allowance scales by tier: Standard = 25 credits/user/mo (100 indexed objects), Premium = 70 credits/user/mo (250 indexed objects), Enterprise = 150 credits/user/mo (625 indexed objects). Admins receive usage alerts at 80% and 100% of the monthly allowance and can opt into extra usage (disabled by default).
• Rovo Dev, a separate AI coding agent for engineering teams, is priced independently at $20 per developer per month, including 2,000 Rovo Dev credits per developer per month; overage bills at $0.01 per credit, with an additional 2,000 credits per user per month of extra usage auto-enabled by default so work is not blocked at the cap. Seats can be purchased for individual developers rather than upgrading an entire site.
• Rovo Customer Service (Jira Service Management AI virtual agent) is bundled into Premium/Enterprise JSM plans at $1 per AI-resolved conversation, with an included allotment of 1,000 assisted conversations per month (or 12,000 per year) and volume-discounted add-on packs beyond that.

Hidden costs : What a multi-product Atlassian bill actually looks like

The $7.91/user headline on the Jira Standard card is only ever paid by one product, at one specific team size, in isolation. Most real customers run Jira plus Confluence plus a Jira Service Management desk, and pay Rovo overage on top. Two representative examples:

A 350-person engineering & IT organization on Premium

Once Confluence and a Premium-tier JSM desk are layered on top of the advertised per-seat Jira number, the blended bill is nearly 2x the Jira-only Premium line by itself — and Rovo Customer Service’s per-resolution overage can add hundreds of dollars a month before any seat count even changes.

A 50-developer team standardized on Rovo Dev

Because Rovo Dev auto-enables an additional 2,000 credits/developer/month of extra usage by default rather than hard-capping the team, a group that regularly exceeds its pool pays a predictable ~30% premium on top of seat fees instead of being blocked mid-sprint. It’s a textbook credit-based billing trade-off: bill predictability up front, in exchange for an overage line that compounds silently unless someone is watching the admin usage dashboard.

Want to estimate your own Atlassian bill? Use the Atlassian pricing calculator to model your monthly cost based on team size, product mix, and Rovo credit usage.

Pricing evolution : From a flat per-user fee to a two-stage self-hosted sunset and bundled AI credits

Cadence

Tracked range: 2019–2026. Quarters not listed above were verified stable (0 price changes, 0 SKU additions).

Notable changes

• 2020-09 — Jira Cloud’s flat $10/mo (10 users) + $7/user/mo (11-100 users) model from 2019 is replaced by a Free/Standard/Premium three-tier ladder, with no Enterprise tier yet.
• 2020-10-19 — Atlassian announces it will stop selling new self-hosted Server licenses; the plan draws a 380-point, 298-comment Hacker News thread.
• 2021-02-02 — Server end-of-sale takes effect; Cloud Enterprise launches as a fully quotable, unlimited-instance top tier the same day.
• 2023-10-18 — First annual October Cloud price increase (~5% average).
• 2024-02-15 — Server reaches full end-of-support; Data Center list prices rise 23-29% depending on product.
• 2024-10-09 — Rovo reaches general availability priced at $20/user/month as a separate add-on.
• 2024-10-16 — Jira Service Management introduces its first consumption-based AI pricing: $0.30 per Rovo virtual-agent resolution above a 1,000/month free allowance.
• 2025-02-11 — Second annual Data Center price increase (15-25% list, 30% Advantage).
• 2025-04-09 — Rovo’s per-user add-on charge is discontinued; credits bundle free into Premium/Enterprise.
• 2025-09-09 — Atlassian announces Data Center will fully end of life on March 28, 2029.
• 2025-10-15 — Rovo bundling reaches Standard; Rovo Dev reaches general availability; third annual Cloud price increase.
• 2026-02 — Third annual Data Center price increase (flat 15%).
• 2026 (exact date unknown) — Jira Service Management’s Rovo Customer Service overage rate rises from its October 2024 introductory $0.30/resolution to $1.00/resolution — more than 3x the launch rate — though Atlassian has not published the exact effective date.
• 2026-08-17 — Atlassian begins collecting customer metadata by default to train its AI models, reversing its earlier no-training pledge; the change draws a 604-point, 136-comment Hacker News discussion.

The self-hosted sunset in detail

Atlassian’s move away from self-hosted deployment happened in two distinct stages nearly a decade apart. Server — the original self-hosted product — was announced for discontinuation in October 2020: new license sales stopped February 2, 2021 (the same day Cloud Enterprise launched), and support fully ended February 15, 2024. Data Center, the more expensive self-managed successor Atlassian had positioned as Server’s replacement, looked like the permanent home for regulated and large on-prem customers — until September 9, 2025, when Atlassian announced Data Center will also fully end of life on March 28, 2029, with new-customer sales ending March 30, 2026 and existing-customer expansions ending March 30, 2028. Every Data Center price increase since 2024 (23-24% in 2024, 15-25% in 2025, a flat 15% in 2026) has landed in the years immediately preceding this wind-down, which is the clearest example in this corpus of pricing being used as a deliberate migration lever rather than a simple revenue lever.

What’s unique : Three AI pricing mechanics inside one company

1. Three genuinely different AI monetization patterns, all live at once. Most vendors pick one AI pricing mechanic and apply it everywhere; Atlassian runs three simultaneously. Rovo’s core search/chat/agents are bundled with a per-seat credit allowance baked into the existing Jira/Confluence/JSM subscription (no separate line item). Rovo Dev, the standalone coding agent, is seat-plus-credit-overage — its own $20/developer/month SKU with a per-seat base and metered overage. Rovo Customer Service, JSM’s virtual agent, is pure outcome-based pricing at $1 per AI-resolved conversation, with zero relationship to seat count. Few companies in this corpus run all three archetypes — bundled-credits, seat-plus-overage, and per-outcome — under one brand at the same time.

2. AI credits as a loss-leader upsell lever, not a revenue line. Atlassian doesn’t monetize Rovo’s core credits directly — it gives them away to push customers into higher-credit tiers. Premium and Enterprise plans carry roughly 3x and 6x the Standard credit allowance respectively, and Atlassian has publicly said customers who adopt Rovo grow their subscription ARR at more than twice the rate of non-adopters. The AI feature is the upgrade incentive; the seat-tier ladder is where the money is actually collected.

3. Pricing as a migration lever, not just a revenue lever. Atlassian’s Data Center list-price increases (23-24% in 2024, 15-25% in 2025, 15% in 2026) have landed in the exact years surrounding two forced self-hosted exits: Server’s 2021/2024 sunset and Data Center’s own newly announced 2029 end-of-life. The price gap between self-hosted and Cloud isn’t incidental — it’s a deliberate lever widening every year self-hosted remains available, a pattern worth understanding alongside the broader usage-based pricing fundamentals this corpus tracks.

4. A published, banded per-seat rate card instead of a black-box quote. Unlike most Enterprise-gated SaaS, Atlassian publishes its actual banded per-user rates (e.g., Jira Standard: $9.05/user for 1-100 users, down to $2.45/user above 90,000) rather than hiding volume discounts behind “contact sales.” The team-size slider on every pricing page recalculates the blended rate live against that published table — a rare level of pricing transparency at Atlassian’s scale.

Strengths & weaknesses

Billing UX : How Atlassian’s pricing calculator and admin usage controls work

• Team size slider — an interactive users/agents slider on each product’s pricing page recalculates the displayed per-user price live against Atlassian’s published volume bands.
• Monthly/Annually toggle — switches billing cadence and surfaces “SAVE UP TO 17%” messaging on every paid tier; Enterprise pricing is only reachable after switching to Annual.
• Maximum Quantity Billing (MQB) — for monthly subscriptions, Atlassian bills on the highest seat count assigned at any point in the billing cycle; adding seats mid-cycle is prorated, but removing seats does not reduce that cycle’s bill.
• Cloud Calculator — a separate multi-step tool (atlassian.com/software/pricing-calculator) lets buyers select multiple products at once (Jira, Jira Service Management, Jira Product Discovery, Confluence, Bitbucket, Compass, Atlassian Guard) and estimate a combined bill.
• Rovo Platform usage dashboard — an admin panel shows total Rovo AI credit allowance versus consumption over time, broken out by app and by user, with proactive email/in-product alerts at 80% and 100% of the monthly allowance.

Strategic wins : Why Atlassian’s AI and migration bets paid off

1. Bundling Rovo’s core credits into existing seat tiers, instead of a new per-user AI add-on

Atlassian tried the industry-default move first — launching Rovo in October 2024 as a $20/user/month add-on — and reversed it within six months once adoption lagged. Folding Rovo’s credits into Premium and Enterprise for free in April 2025 (and Standard in October 2025) removed the “should we even pay to try AI” purchase decision entirely, which is exactly the per-user-license-to-AI-bundle shift playing out across the whole SaaS market. Atlassian’s own reporting that Rovo adopters grow ARR more than 2x faster than non-adopters suggests the bundle is working as an upsell engine rather than a cost center.

2. Splitting Rovo Dev and Rovo Customer Service into billing models that match how they’re actually consumed

Rather than forcing every AI surface through one mechanic, Atlassian priced Rovo Dev on seats-plus-credit-overage (matching how individual developers burn tokens continuously) and Rovo Customer Service on pure per-outcome pricing (matching how a support desk only cares about tickets actually resolved). That’s a more disciplined value-metric exercise than most vendors attempt across even one AI product, let alone three.

3. Publishing the real banded rate card instead of hiding volume pricing behind a sales call

Most SaaS vendors gate any volume discount behind “contact sales.” Atlassian’s team-size slider recalculates the actual blended per-user price live against a published table (Jira Standard: $9.05/user at 1-100 users down to $2.45/user above 90,000) — a level of self-serve pricing transparency that materially shortens the usage-metric evaluation a buyer has to do before committing.

4. Telegraphing the self-hosted sunset years in advance, in two clearly separated stages

Server’s wind-down (announced 2020, sold until 2021, supported until 2024) and Data Center’s own newly announced 2029 end-of-life give affected customers multi-year runways rather than an abrupt cutoff — a rare example of a vendor sequencing a forced-migration pricing lever with enough lead time that customers can actually plan around it instead of being blindsided.

Areas to improve : Where the AI credit story gets harder to trust

1. Publish the Rovo Customer Service overage-rate change instead of letting it surface only in screenshots

The per-resolution rate for JSM’s Rovo virtual agent rose from its October 2024 introductory $0.30 to $1.00 by 2026 — more than a 3x increase — with no dated blog post, changelog entry, or investor notice found for the change itself (only the before/after prices). A vendor charging per AI outcome should treat that rate the way it treats seat-tier prices: a dated, citable pricing-change announcement, not a silent update to a support-page table.

2. Give Free/Standard/Premium customers a real opt-out from AI-training data collection, not just Enterprise

Starting August 17, 2026, Atlassian collects Jira/Confluence metadata by default to train its AI models, with no opt-out available below Enterprise — a policy reversal from its earlier no-training pledge that drew a 604-point Hacker News thread. Tying a core data-governance right to the highest, sales-quoted tier (rather than offering it as a toggle on any paid plan) reads as pricing power exercised over privacy, not a technical constraint.

3. Resolve the tension between “free AI credits” messaging and the seat-tier upsell they’re funding

Rovo’s headline is “included at no additional cost,” but Premium and Enterprise carry 3x and 6x the Standard credit pool — meaning the real price of more AI usage is upgrading the whole team’s seat tier, not a transparent AI line item. Disclosing an explicit “cost per additional 1,000 Rovo credits by tier” the way prepaid credit models elsewhere in this market do would let buyers actually compare AI cost per tier instead of reverse-engineering it from the plan-comparison table.

4. Shorten the gap between three AI products and one mental model for total AI spend

A customer must separately track Rovo core credits (per seat, bundled), Rovo Dev credits (per developer, metered), and Rovo Customer Service resolutions (per outcome, metered) to know their real AI bill. A single consolidated AI-usage view spanning all three — not just the existing per-product Rovo dashboard — would match the usage-based billing consolidation buyers increasingly expect from multi-product AI vendors.

Key takeaways

1. A failed pricing experiment is recoverable if you reverse it fast. Atlassian launched Rovo as a $20/user/month add-on and walked it back to a free bundled credit allowance within six months once adoption lagged. Pricing teams should treat a new AI SKU’s launch price as a hypothesis to test, not a permanent commitment.
2. Different AI products can use different pricing mechanics under one brand. Rovo core (bundled credits), Rovo Dev (seat + overage), and Rovo Customer Service (per-outcome) each match a different consumption pattern rather than forcing one mechanic company-wide. Match the meter to how the specific feature is actually used, not to organizational convenience.
3. Published banded rate cards build more trust than “contact sales” at every volume tier. Atlassian’s live team-size slider against a public per-user rate table removes the black-box feeling most enterprise software pricing creates, even though Enterprise itself still requires a quote.
4. A pricing lever can double as a migration lever, but the timing has to be defensible. Atlassian’s Data Center price increases (23-24% in 2024, 15-25% in 2025, 15% in 2026) landed in the exact years surrounding two forced self-hosted exits — a deliberate sequencing that only works because customers were given years of advance notice.
5. Silent AI overage-rate changes erode the trust that outcome-based pricing depends on. JSM’s Rovo Customer Service per-resolution rate roughly tripled from $0.30 to $1.00 with no dated public announcement found for the change itself. Outcome-based pricing’s core promise — “pay for value delivered” — requires the price of that value to be as visible as the seat-tier price is.

UBP implications

1. AI credit bundling is becoming the default on-ramp for AI features inside existing SaaS products. Rather than charging a separate AI subscription, Atlassian folds Rovo credits into the seat tiers customers already buy, using higher-tier credit allowances (not a distinct AI price) as the monetization lever — a pattern other incumbent SaaS vendors adding AI are likely to converge on.
2. A single vendor can validate three different usage-based pricing archetypes at once, which is useful evidence for the field. Bundled-credit, seat-plus-overage, and per-outcome billing are usually studied as competing philosophies; Atlassian runs all three simultaneously and profitably, suggesting the “right” AI pricing model is a function of the specific feature’s consumption pattern, not a single company-wide doctrine.
3. Forced platform migrations (cloud-only, self-hosted sunset) show that usage-based and packaging pricing levers can be combined with deployment-model levers. Atlassian’s Data Center price increases only make sense in the context of its multi-year self-hosted sunset — a reminder that usage-based pricing strategy doesn’t operate in isolation from deployment, packaging, and product-lifecycle decisions happening at the same time.

Sources

• Jira pricing (accessed 2026-09-01)
• Confluence pricing (accessed 2026-09-01)
• Jira Service Management pricing (accessed 2026-09-01)
• Atlassian Cloud pricing calculator (accessed 2026-09-01)
• Jira licensing & pricing docs (accessed 2026-09-01)
• Rovo licensing FAQs (accessed 2026-09-01)
• Rovo Dev pricing (accessed 2026-09-01)
• Data Center end of life (accessed 2026-09-01)
• Future pricing hub (accessed 2026-09-01)
• Atlassian blog (accessed 2026-09-01)

Bottom line

Atlassian’s per-seat Jira/Confluence/JSM ladder is a familiar SaaS structure, but its AI layer is the more interesting story: three distinct Rovo products — bundled credits, seat-plus-overage, and pure per-outcome billing — running simultaneously under one brand, funded by a deliberate loss-leader upsell strategy, and shadowed by a fast-moving self-hosted sunset and a 2026 AI-training data policy that together show how much a pricing page can obscure about where the real leverage sits.

Want to compare Atlassian against other hybrid and AI-pricing models? Browse the pricing blueprint.

Pricing timeline : Major events on a vertical axis

Each milestone below corresponds to a public pricing change, product launch, or material adjustment. Major events use a filled marker; minor adjustments use a faded one.

Baseline: Free/Standard/Premium/Enterprise with Rovo AI credits

Sep 2026

Captured baseline structure: Jira, Confluence, and Jira Service Management each sell Free/Standard/Premium/Enterprise per-user tiers with progressive volume discounting, and every paid tier bundles a Rovo AI credit allowance (25/70/150 credits per user/month); JSM additionally meters its Rovo virtual agent at $1 per resolution on Premium/Enterprise — up from the $0.30 introductory rate — and Rovo Dev sells separately at $20/developer/month.

Default AI-training data collection takes effect; no opt-out below Enterprise

Aug 2026

Atlassian began collecting Jira/Confluence metadata by default to train its AI models for roughly 300,000 customers, reversing its earlier pledge not to train on customer data. Free, Standard, and Premium customers cannot opt out; only Enterprise can decline both metadata and in-app content collection. The policy drew a 604-point, 136-comment Hacker News discussion.

Third annual Data Center price increase (flat 15%)

Feb 2026

Atlassian applied a flat 15% Data Center list-price increase across all tiers, with legacy Advantage-pricing customers facing 18-40% increases depending on tier — roughly a 41% average cumulative Data Center increase since February 2024.

Rovo bundling reaches Standard; Rovo Dev reaches GA; third October price rise

Oct 2025

Rovo's free credit bundle extended to Standard-tier Cloud subscriptions, and Rovo Dev (the standalone AI coding agent) reached general availability. Atlassian's own Investor Relations pricing notice confirms Cloud list prices rose again effective this date: Standard +5%, Premium +7.5%, Enterprise +7.5-10%, Bitbucket +10%.

Data Center itself gets an end-of-life date: March 28, 2029

Sep 2025

In a blog post titled "The next chapter for Atlassian and our customers," Atlassian announced Data Center (Jira Software, Jira Service Management, Confluence, Bamboo, Crowd) will fully end of life on March 28, 2029, with new-customer sales ending March 30, 2026 — the second and final stage of Atlassian's self-hosted sunset, after Server's 2021/2024 wind-down.

Rovo's per-user add-on pricing ends; credits bundle free into Premium/Enterprise

Apr 2025

Atlassian discontinued Rovo's separate $20/user/month add-on charge; Rovo credits became bundled at no additional upfront cost into Premium and Enterprise Cloud subscriptions, rolling out through July 2025.

Second annual Data Center price increase (15-25%, ~30% advantaged)

Feb 2025

Per Atlassian's FY25 Investor Relations notice, Data Center list prices rose 15% (<=1K users/250 agents), 20% (2K-5K users/251-1K agents), or 25% (>5K users/1K+ agents), with Jira Software "Advantage" legacy-pricing customers seeing a flat 30% increase across all tiers.

Rovo reaches GA as a $20/user/mo add-on; JSM adds consumption AI pricing

Oct 2024

Rovo (search, chat, agents) launched in general availability priced separately at $20/user/month annual ($24/month monthly), with usage-based pricing originally planned for mid-2025 (TechCrunch, 2024-10-09). The same week, Atlassian raised Cloud list prices again (Standard +5%, Premium/Enterprise +10%, JSM +8-20%) and introduced Jira Service Management's first consumption-based AI pricing: 1,000 free Rovo virtual-agent assisted conversations/month (or 12,000/year), then $0.30 per conversation above that, per Atlassian's own FY25 Investor Relations pricing notice.

Server end-of-support; Data Center prices jump ~23-24%

Feb 2024

Server reached full end-of-support with no further security patches or bug fixes, and Atlassian simultaneously raised Data Center list prices roughly 23-24% (Jira +24%, Confluence +29%, JSM +23%), with legacy "Advantage" discount customers seeing realignments up to 27-30% — widening the self-hosted-vs-Cloud price gap to accelerate migration.

First of the annual October Cloud price increases

Oct 2023

Atlassian raised Cloud list prices roughly 5% on average across Standard, Premium, and Enterprise for Jira and Confluence, and 5-20% for Jira Service Management — the first in what became a yearly October increase cadence.

Server end-of-sale takes effect; Cloud Enterprise becomes a full tier

Feb 2021

Atlassian stopped selling new on-premises Server licenses and launched Cloud Enterprise as a fully quotable, unlimited-instance top tier the same day (TechCrunch, "Atlassian stops selling on-prem server licenses, adds new enterprise pricing tier," 2021-02-02).

Server sunset announced; an Enterprise column appears

Oct 2020

Atlassian announced it would stop selling new self-hosted Server licenses, targeting end-of-sale February 2021 and end-of-support February 2024 — a 380-point, 298-comment Hacker News discussion. Within weeks Jira Cloud's pricing page added a fourth Enterprise column (unlimited sites, included Atlassian Access, data residency).

Free/Standard/Premium three-tier ladder replaces the flat model

Sep 2020

By September 2020 Jira Cloud had moved to a Free (10 users)/Standard ($7/user avg)/Premium ($14/user avg) three-tier structure; no Enterprise tier existed yet.

Baseline: flat/per-user Cloud pricing, no free tier

Mar 2019

Jira Cloud priced at a flat $10/month for up to 10 users and $7/user/month for 11-100 users, with a banded annual price list published up to 2,000 users; no Free tier and no Standard/Premium/Enterprise ladder existed yet.