Start with an inventory and a named recipient
Collect the agreed deliverables in one inventory. For each item, record its location, recipient, and status. Include source code, documentation, design files, data structures, and operational configuration. A statement that all files were sent is not a substitute for that inventory.
Separate rights in project-specific work from the licenses governing libraries, fonts, images, and outside services. Record the relevant terms in line with the project agreement. List anything still outstanding and why, so that accepting the work has a clear boundary.
Source and accepted release
Identify the repository, branch or release that matches production, together with the checks performed on that version.
Current documentation
Installation, configuration, deployment, and operating instructions should match the delivered system, not an earlier architecture.
The next operator
Name the owners of accounts, service charges, updates, and incident reporting.
Have the receiving team follow the setup instructions
Use an appropriate test environment and ask the receiving team to start from a fresh copy of the source. The handover instructions should identify the runtime, dependencies, required configuration, database preparation, and commands for testing and building.
Document variable names and their purpose without including real secrets. Use suitable test data and note where additional access is required. When the receiving team discovers a missing step, put the answer back into the documentation before closing that acceptance item.
Check access before removing the outgoing team
Make an account register covering the domain, DNS, hosting, repository, database, and external services. Identify account owners, administrators, billing responsibilities, and notification recipients. Confirm that the company’s team can reach the required resources using its own accounts.
Transfer secrets through the agreed restricted channel, not through source files or operating manuals. At the end of the handover, consider rotating credentials and removing access that is no longer needed. Check automated systems and service dependencies before changing a credential.
| System | What to verify | Evidence |
|---|---|---|
| Domain and DNS | Account owner, renewal responsibility, record editors | Access to the correct domain account |
| Hosting and database | Roles, billing ownership, notification recipients | Access to the identified environments |
| Source and CI/CD | Repository administration and release permissions | Ability to access source and follow the release process |
| External services | Owners, billing, and service constraints | A complete account and responsibility register |
Rehearse deployment and the route back
The deployment instructions should explain the checks required before release, who approves it, the commands or pipeline used, and how to confirm normal operation afterward. Rehearse in a suitable non-production environment before establishing the production procedure.
Restoring an earlier code version does not necessarily restore the database with it. For migrations and data-format changes, document compatibility, irreversible steps, and the tested recovery approach. Do not reverse a database change without accounting for new data created since the release.
Before release
Check the version, approval, configuration, permissions, backups, and acceptable impact window.
After release
Verify important user journeys, system health, and the agreed integrations.
If checks fail
Name the person who decides whether to stop or revert, with a data-checking procedure and a communication path.
Verify recovery, not just the presence of backups
Record what is backed up, where it is stored, who can access it, and the retention policy. Then restore into an isolated test environment. Check that the required data and supporting components are present, and record the actual time taken and any manual steps.
Acceptable interruption and data loss are business decisions. Use the recovery exercise to assess whether the prepared process fits those goals. The existence of a backup alone is not evidence for a promised recovery time.
Agree outstanding work and ongoing support
List known defects, accepted limitations, and work that will continue separately. Give each item an owner and a way to close it. Distinguish defects against the agreed scope from requests for new features so that support and subsequent estimates remain clear.
Record the agreed correction period, maintenance scope, reporting channel, and response terms. Identify who handles renewals and dependency updates. Handover does not automatically include round-the-clock operation or new features.
A checklist for the next team conversation
- Deliverables match the agreement and the accepted release
- The receiving team can install and build from the documentation
- Important accounts have company owners and administrators
- Secrets are transferred and managed through an appropriate channel
- Deployment and code/data rollback have been rehearsed
- Recovery results and limitations are recorded
- Outstanding work and post-handover responsibilities are agreed
A useful handover provides evidence that the next team can continue: both when making ordinary changes and when the system is unavailable.
Talk through your project