User count hygiene moved 20 to 35 percent of the bill, which is more than the discount moved in any of the agreements reviewed
Atlassian rarely gives a large discount and rarely needs to. The renewal is decided by directory sync rules somebody set years earlier and nobody has read since.
Prepared by Redress Compliance · August 18, 2026 · Atlassian Cloud advisory. 25 to 30 agreements reviewed, 2024 to 2025.
Executive summary
Licensed user counts ran 20 to 35 percent above monthly active users once directory sync inflation was audited. Every one of those seats bills at the same rate as a seat somebody actually works in.
Estates sitting just above a tier boundary saved 10 to 20 percent by cleaning users back under the cliff before renewal. Crossing a band reprices the whole population, not the marginal user.
Marketplace app spend grew to 25 to 40 percent of platform cost in mature estates, scaling on the same padded counts the platform bills.
The negotiation is arithmetic before it is commercial. Structure moved enterprise deals even where the per user rate barely moved at all.
What actually sets an Atlassian bill?
Atlassian prices each product per user on banded tiers. Jira, Confluence and Jira Service Management each count their own users on their own bands, with rates published on the Jira pricing page.
Tier boundaries are the hidden mechanic. Crossing a band reprices all users rather than the marginal ones, so an estate near a cliff pays disproportionately for growth it did not plan. The full band architecture sits in the Atlassian cloud pricing guide.
Few estates need identical counts on each product
Cloud Enterprise sits at the top of the range, adding unlimited instances, data residency and advanced governance, documented on the enterprise cloud page. Estates running one instance with light compliance needs often fit Premium.
| Lever | Works when | Typical movement |
|---|---|---|
| User audit before quoting | You hold 90 days of activity data | 20 to 35 percent off billable seats |
| Tier boundary management | The estate sits just above a band | 10 to 20 percent from recrossing down |
| Plan right sizing | Enterprise features go unused on Premium workloads | Per user rate relief |
| Multi year term | Traded for written renewal caps | Protection against list increases |
Where does the padding actually hide?
Audit billable users against product activity 90 days before renewal, because directory sync quietly licenses everyone the identity provider knows. The cloud licensing documentation defines who counts as billable.
- Directory sync inflation: every synced account with product access bills, active or not.
- Cross product over assignment: users granted all products by default when they use one.
- Departed users: offboarding flows that remove identity but not product access.
- Service accounts: automation identities holding full seats.
When the resulting drop crosses a tier boundary downward, the saving compounds across every remaining seat rather than stopping at the seats removed.
The Atlassian renewal kit
The user audit worksheet, the tier boundary model, and the renewal cap language that survives Atlassian's redlines.
Get the brief →What 25 to 30 Atlassian agreements showed
Across roughly 25 to 30 Atlassian Cloud agreements reviewed between 2024 and 2025, user count hygiene and tier placement decided more spend than negotiated discounts. Three patterns recur.
- Licensed user counts ran 20 to 35 percent above monthly active users once directory sync inflation was audited.
- Estates sitting just above a tier boundary saved 10 to 20 percent by cleaning users back under the cliff before renewal.
- Marketplace app spend grew to 25 to 40 percent of platform cost in mature estates, scaling on the same padded counts.
Atlassian bills follow directory sync decisions made years ago. The identity configuration, not the seller, usually sets the renewal baseline.
- Percentile standing for your exact deal size and industry, from real closed transactions
- Scenario simulation before the call, so you can test alternative terms and see the financial impact of each
- A negotiation playbook, talking points, and a two page executive brief on day one
Why do marketplace apps compound the problem?
Per user apps scale on the same populations the platform bills. Padding you never removed is therefore paid for twice, once on the platform line and again across every installed app.
Review the apps on the platform calendar
App spend reached 25 to 40 percent of platform cost in the mature estates benchmarked. That is large enough to deserve its own review, and it moves automatically when the counts move. The mechanics sit in the marketplace app scaling guide.
Watch the briefing · 6:31Converting Off Atlassian Data Center Before the 2029 DeadlinePricing the renewal, the cloud equivalent and the migration as one comparison.
Which terms are worth the negotiation?
Renewal caps, true down rights and app governance, negotiated under the cloud terms of service framework. Atlassian raises list prices regularly, so the cap clause is not theoretical.
Three clauses carry the term
- Renewal cap: a written ceiling on per user rate increases at renewal.
- True down rights: the contractual ability to reduce user counts at anniversary.
- Coterming: aligned product end dates, so the negotiation happens once with full leverage.
The tier gate and the Data Center migration window are the other half of this negotiation, and they are covered in the Atlassian tier and migration guide. Exit timing sits in the cloud migration guide, and the paper itself in the contract negotiation guide.
What the audits measured, 2024 to 2025
Two cuts of the engagement file size the opportunity before any discount is discussed.
The gap between licensed users and users with product activity over 90 days, across the estates audited.
Where the cleaned count carried the estate back under a band boundary it was sitting just above.
Treat the ranges as benchmarks rather than promises. Your directory sets your baseline, and the file describes what disciplined buyers achieved against the same vendor playbook.
Your first five moves
- Export billable users per product and compare against 90 days of activity logs, which is where the 20 to 35 percent gap becomes visible.
- Fix the directory sync rules so only active product users hold seats, including service accounts and departed staff whose identity was removed but whose product access was not.
- Map the cleaned counts against tier boundaries and clean back under the nearest cliff, worth a further 10 to 20 percent where the estate sits just above one.
- Verify the Enterprise plan features are genuinely used, and right size to Premium where they are not. The plan comparison prices the difference.
- Trade multi year term for written renewal caps and true down rights, then audit app spend on the same calendar. The Atlassian practice runs the count before the quote arrives.
Frequently asked questions
How is Atlassian Cloud Enterprise priced?
Per user, per product, on banded tiers, with Cloud Enterprise as the top plan adding unlimited instances, data residency and governance. Each product bills its own user count, so the estate total is the sum of separately tiered populations.
Do inactive users count in Atlassian billing?
Generally yes. Directory synced accounts with product access bill whether or not they log in, and audits in the 2024 to 2025 file found 20 to 35 percent of billable users inactive over 90 days.
Does Atlassian give enterprise discounts?
At scale, modestly, and mostly on structure rather than headline rate. Cleaned counts, tier boundary management and written renewal caps moved more money than discount asks in the agreements reviewed.
What happens at a user tier boundary?
Crossing a tier reprices every user, not only the marginal ones. Estates sitting just above a boundary saved 10 to 20 percent by auditing users back under the cliff before renewal.
Is Cloud Enterprise worth it over Premium?
When you genuinely need multiple instances, data residency or advanced governance, yes. Estates running a single instance with light compliance requirements often fit Premium at a materially lower per user rate.
How much do marketplace apps add?
App spend reached 25 to 40 percent of platform cost in the mature estates benchmarked. Per user apps scale on the same padded counts the platform bills, so the padding is paid for twice.
When should the user audit happen?
Ninety days before renewal, so the activity data exists and the cleaned count lands before Atlassian sets a baseline. An audit run after the quote is an argument against a number already anchored.
What is a renewal cap worth?
It is the protection against the list increases that followed in the file reviewed. Atlassian raises published prices regularly, so a written ceiling on per user increases is the term that survives the term.
Do service accounts really consume seats?
Automation identities holding full product access bill like any other user in most configurations. They surface in almost every audit because nobody offboards a robot.
Has the migration leverage gone?
Largely. With server retired, Atlassian knows the estate is not going back, so Data Center exit timing is the remaining pressure point rather than the threat to stay put.
