Atlassian and the new usage meters — what changed in September 2026
Until spring 2026, Atlassian billed primarily per seat. The model was predictable: 50 Jira Premium users equalled one fixed invoice line. That model has changed. On September 1, 2026, Atlassian announced a usage billing layer stacked on top of existing subscriptions. Meter tracking begins immediately; billing for overages starts December 3, 2026.
- Rovo credits: $0.01 per credit beyond the included allowance. Each AI agent invocation consumes multiple credits.
- Automation steps: $0.50 per 1,000 steps. The old model counted triggered rules; the new one counts every action executed inside a rule.
- Loom AI: transcriptions, summaries, and auto-chapters consume Rovo credits.
- Assets objects (JSM) and CSM agent resolutions: two additional meters affecting support teams.
- Catalogue price increase: approximately +7% on Standard and Premium plans from October 13, 2026 for new subscriptions and renewals.
For a 50-person team using Jira Software Premium + Confluence Premium + Loom + Rovo enabled by default, projections published by several Atlassian partner consultants (Praecipio, SPK & Associates, Sourcesense) converge on increases of 40 to 150% depending on automation usage intensity. The highest figure applies to DevOps teams with automation pipelines firing thousands of times per day.
Additionally, Atlassian confirmed the end of sales for new Data Center subscriptions on March 30, 2026. Existing customers can renew until March 2028; full end of life is planned for March 2029. It is therefore no longer possible for a new organisation to choose Data Center as an on-premises hosting path through Atlassian.
Overview: which tool replaces what?
The table below summarises open-source alternatives by Atlassian feature, with their category, licence and maturity.
Scroll the table
| Atlassian Feature | Self-hosted Tool | Migration Effort |
|---|---|---|
| Jira Software – issues and sprints | Plane | Medium |
| Jira Software – lightweight issues + CI/CD | Gitea Issues + Gitea Actions | Low |
| Confluence – wiki and documentation | Docmost | Medium |
| Loom – video recording and sharing | Nextcloud Talk / PeerTube | High |
| Jira – internal databases / registries | Grist | Low |
| Confluence – tables and databases | NocoDB | Low |
| Jira Service Management – support tickets | Zammad (outside ServOrbit catalogue) | High |
Gitea for issues and CI/CD
Gitea is a lightweight Git forge written in Go. Its official Docker image (gitea/gitea) runs on 512 MB of RAM for teams of fewer than 20 people, and on 1 GB for about a hundred active users. It includes an issue tracker with labels, milestones, Kanban projects, and since Gitea 1.21, an Actions module compatible with GitHub Actions syntax.
- Minimum resources: 1 vCPU, 512 MB RAM, 10 GB disk (without CI artefacts).
- Docker image:
gitea/gitea:latest— mount the/datavolume on a persistent disk. - Migration from Jira: Jira CSV export → import via the Gitea REST API (community
jira2giteascript). Issue links and labels are preserved; Jira workflows are not. - Marginal cost on a ServOrbit VPS: zero licence cost, only compute resources matter.
Gitea is particularly suited to development teams that want to keep code and issues in a single tool without paying separately for GitHub Advanced Security or Jira. The issue tracker lacks Jira's depth for portfolio management, but it broadly covers the sprint use case for a small team.
Docmost for team wikis (Confluence replacement)
Docmost is an open-source real-time collaborative page editor launched in 2024 that reached production maturity in 2025. It offers page hierarchies, spaces (equivalent to Confluence spaces), inline comments, user mentions, and a full-text search engine.
- Minimum resources: 1 vCPU, 512 MB RAM. PostgreSQL database required.
- Docker image:
docmost/docmost:latestwith an officialdocker-composeincluding PostgreSQL and Redis. - Migration from Confluence: Confluence HTML Space export → conversion via
confluence-to-markdownthen Docmost import. Complex Confluence macros (Gliffy diagrams, Jira integrations) are not automatically migrated. - What does not migrate: custom Confluence templates, native Jira integrations (Issues on a page), and third-party macros.
Plane for Jira sprint management
Plane is the most direct self-hosted competitor to Jira Software. It supports cycles (sprints), modules (epics), multiple views (board, list, gantt, spreadsheet), custom states, and priorities. Its interface is modern and its onboarding curve is noticeably shorter than Jira's.
- Minimum resources: 2 vCPUs, 2 GB RAM (Plane runs multiple services: API, worker, beat, frontend).
- Docker images:
makeplane/plane-backend+makeplane/plane-frontend+makeplane/plane-worker. An officialdocker-composeis provided in the GitHub repository. - Migration from Jira: Plane natively imports Jira JSON exports (via
Settings > Import). Issues, statuses, priorities and assignees are imported. Custom workflows and computed fields must be recreated. - Key strengths: public roadmap option, full REST API, outbound webhooks, native GitHub/GitLab integration.
Plane does not replace Jira Service Management for external customer support, nor Jira Product Discovery for strategic portfolio management. For a development team managing sprints, backlogs, and releases, it covers the essentials.
Grist and NocoDB for internal databases
Confluence and Jira are often used to host structured registries: internal customer bases, asset catalogues, project tracking tables. These use cases migrate better to dedicated tools than to a wiki or issue tracker.
- Grist: open-source relational spreadsheet with Python formulas, multiple views (grid, card, calendar, chart), and fine-grained permission controls. Docker image:
gristlabs/grist. Minimum resources: 1 vCPU, 512 MB RAM. - NocoDB: Airtable-style interface layered on any relational database (PostgreSQL, MySQL, SQLite). Supports auto-generated REST and GraphQL APIs, webhooks, galleries, and forms. Docker image:
nocodb/nocodb. Minimum resources: 1 vCPU, 256 MB RAM. - Migration: CSV or JSON export from Confluence Databases or Jira custom fields, direct import into Grist or NocoDB.
- Typical use case: IT asset registry, contract tracking, internal product catalogue, prospect contact base — anything that lived in Confluence Table macros or Jira custom fields.
Five-step migration plan
These five steps allow a gradual exit from Atlassian without service interruption, by validating each tool under real conditions before the final cut-over.
Step 1
Usage audit: identify in Atlassian Administration which meters are already in the red (Rovo credits, automation steps). Export the list of active Confluence spaces (last modified < 12 months) and the active Jira backlog (issues created or commented on in the last 90 days). This defines the real scope to migrate.
Step 2
Export: from Jira, export as JSON via
Settings > System > Backupor per project viaProject settings > Export. From Confluence, export each space as HTML viaSpace settings > Export. Keep timestamped exports — they also serve as a legal archive.Step 3
VPS installation: deploy your chosen tools from the ServOrbit catalogue. Each application has a pre-configured Docker image with an nginx reverse proxy and automatic TLS certificate. Allow 30 minutes to 2 hours per tool depending on configuration complexity.
Step 4
Import and reconfiguration: use native import scripts (Plane for Jira,
confluence-to-markdownfor Docmost) then rebuild automations that are not portable. This is the longest phase — allow 1 to 3 days for a team of 20 to 50 people.Step 5
DNS cutover: update your internal or public DNS records to point to the new server. If you used
*.atlassian.netdomains, update all internal links (bookmarks, CI/CD integrations, webhooks). Keep the old Atlassian instance in read-only mode for 30 days before cancelling.
Cost calculation: VPS vs per-seat subscription
For a team of 25 people using Jira Software Premium + Confluence Premium, the Atlassian Cloud price after the October 2026 increase is approximately $16,400 per year (excluding Rovo and Loom). Adding estimated usage meters for an active DevOps team (automation steps, Rovo credits), the annual projection exceeds $20,000.
- A ServOrbit VPS sufficient for Plane + Docmost + Gitea (4 vCPUs, 8 GB RAM, 100 GB SSD): from 99 DH/month, under $600 per year.
- Three-year total cost of ownership: one-time migration (estimated 5–10 person-days), ongoing maintenance (Docker updates, backups), zero licence cost.
- Typical savings for 25 users over 3 years: between $55,000 and $70,000 depending on Atlassian usage level.
- The calculation changes if your team uses Jira Service Management for an external customer help desk: replacement tools are less mature in this segment.
What does not migrate easily: Jira Service Management with a public customer portal, Atlassian-native integrations (Opsgenie, Statuspage, Atlas), complex multi-product automations (Jira + Confluence + Slack via Automation for Jira), and advanced Portfolio/Advanced Roadmaps reporting. If your primary use is ops incident management or programme portfolio management, estimate the reconfiguration cost before deciding.
Conclusion: act before December 3, 2026
The new Atlassian meters begin billing on December 3, 2026. You have three months to measure your actual consumption in Atlassian Administration, simulate your post-meter invoice, and decide whether self-hosted migration makes sense for your context. The tools available in the ServOrbit catalogue — Plane, Gitea, Docmost, Grist, NocoDB — cover the core Jira + Confluence scope for development and business teams. The technical migration is documented in two dedicated guides: one on migrating Jira to Plane, the other on migrating Confluence to Docmost.
- For more on the SaaS vs self-hosted decision: see our complete 2026 cost comparison.
- For the step-by-step Jira → Plane migration: see our Atlassian EOL to self-hosted migration guide.
- For the Confluence → Docmost migration: see our dedicated guide.
