Editorial note. The tier example is Atlassian's own, quoted from its cloud licensing page as of September 2026; headline Standard and Premium list prices move with Atlassian's periodic increases, so the example's rates may not match today's calculator to the cent. Atlassian is not affiliated with Sonic AI.
Most per-seat products charge every seat the same. Atlassian does not, and it is careful about how it does not: the per-user price drops as the seat count grows, but the lower rate applies only to the seats inside that tier — not retroactively to all of them. That is graduated pricing applied to seats instead of usage, and Atlassian's monthly billing adds a second rule worth copying: you are billed for the highest seat count assigned at any point in the cycle.
What Atlassian charges
From Atlassian's licensing page, for Jira Cloud Standard at 450 seats on a monthly plan:
| Tier group | Per user / month | Seats | Subtotal |
|---|---|---|---|
| Seats 1 – 100 | $8.60 | 100 | $860.00 |
| Seats 101 – 250 | $7.30 | 150 | $1,095.00 |
| Seats 251 – 450 | $6.10 | 200 | $1,220.00 |
| Month | average $7.06 | 450 | $3,175.00 |
In Atlassian's words: "your monthly price is calculated using tier groups, each with a distinct per-user price." And on quantity: monthly subscriptions use Maximum Quantity Billing — "you are charged for the highest number of seats ('maximum quantity') assigned to each product at any point during your billing cycle." Annual plans are sold in user tiers (up to 100, up to 250, and so on) and paid up front.
The model underneath
- Seat-based graduated: contiguous seat bands, each with its own per-seat rate; the month's count is split across bands and each slice is priced separately. Not volume — 450 seats is not 450 × $6.10.
- Quantity rule: high-water mark. The billable count is the maximum assigned during the cycle, derived from daily seat snapshots, not the count on invoice day.
- In arrears on monthly, since the maximum is only known at cycle end; in advance on annual, with the tier chosen up front.
- Product-level: each Atlassian product (Jira, Confluence, Jira Service Management) has its own tier table and its own maximum quantity on the same invoice.
Compare Slack, which prorates a seat by the days it existed, and Salesforce, which bills the contracted count. Atlassian's rule is a third answer to "how many seats this month" — the peak — and it is common in developer tooling because seats churn daily.
What a month looks like
A customer whose Jira Standard seat count moved 380 → 412 → 395 during September. Maximum quantity: 412.
| Line | Seats | Rate | Amount |
|---|---|---|---|
| Jira Standard — tier 1 (seats 1–100) | 100 | $8.60 | $860.00 |
| Jira Standard — tier 2 (seats 101–250) | 150 | $7.30 | $1,095.00 |
| Jira Standard — tier 3 (seats 251–412) | 162 | $6.10 | $988.20 |
| September (max quantity 412) | 412 | avg $7.14 | $2,943.20 |
The customer's admin sees 395 seats today and an invoice for 412. The invoice has to say why — the peak date and count — or that is a support ticket every month.
Where graduated seats go wrong
- Volume by accident. An engine that supports "tiers" but prices all seats at the reached tier bills 412 × $6.10 = $2,513.20 — $430 short. The other way round, a customer who negotiated volume pricing gets over-billed by the same logic.
- High-water mark needs daily snapshots. "Maximum during the cycle" is only computable from a per-day seat count; a roster read on invoice day misses the peak.
- Bands on the invoice. Customers expect the tier rows, not a blended $7.14 they cannot verify.
- Annual tier boundaries. An annual customer on the "up to 250" tier who reaches 251 has crossed into a new tier mid-term; the contract decides whether that is a prorated upgrade now or a true-up at renewal.
- Several products, several tables. Jira and Confluence have different tier rates and separate maximums, on one invoice.
Billing this shape on Sonic AI
Jira's tier table is a seat-based graduated list price on Sonic: contiguous seat bands with a first and last seat and a per-seat rate each, and the engine splits the period's billable count across them so the invoice shows each band as its own row. The billable count comes from daily seat snapshots — the meter's aggregation picks the rule (the maximum for Atlassian's monthly plans, a contracted count for annual) — and the snapshot import records the peak date so the invoice can cite it. Each product is its own seat meter on the same billing schedule, and an annual tier crossing is a schedule phase change prorated by calendar units when the contract says so.
See usage & seats, the glossary on graduated vs volume pricing, and seat-based billing, explained.
If your seats are tiered
- Does the lower rate apply only to seats in the band (graduated) or to all seats (volume)? Write the word in the contract.
- What is the quantity rule — peak, average, end of period, or contracted?
- Does the invoice show each band and the peak date?
- On annual plans, is a tier crossing a prorated upgrade or a renewal true-up?
Bring a seat export to a demo — we will import the snapshots and show the tiered invoice with the peak cited.
Similar pricing models
Graduated per-seat tiers are shared across the Atlassian suite — Confluence, Jira Service Management, Bitbucket — and by GitLab, Miro and Smartsheet at the enterprise end, where the per-seat rate steps down by band on the order form. Flat per-seat with activity-based counts is Slack; flat per-user editions on annual terms is Salesforce; the same graduated bands applied to usage instead of seats is AWS S3.
Related terms: graduated pricing, seat-based billing, seat snapshot, in advance vs in arrears.