Go-Live, Deployment and Cutover Planning: A Checklist for IT Project Managers
Go-live is the moment the project becomes real for users. It is also when many of the biggest and most visible problems occur: a migration that overruns, a forgotten integration, users who do not know how to log in, a support desk flooded with calls. Careful deployment and cutover planning turns that moment from a gamble into a rehearsed procedure. This lesson covers readiness, the go/no-go decision, cutover planning, rollback, training and hypercare.
Key Terms
- Deployment (release): moving the solution into the live environment.
- Go-live: the point at which real users start using it for real work.
- Cutover: the transition from the old system or process to the new one, including data migration and switching traffic.
- Rollback: reverting to the previous state if the deployment fails.
- Hypercare: a period of heightened support immediately after go-live.
Deployment Strategies
| Strategy | How it works | Advantage | Risk |
|---|---|---|---|
| Big bang | Everyone switches at once | Simple, short transition | High impact if it fails; hard to reverse |
| Phased | Roll out by location, department or module | Lessons from early phases; contained impact | Longer transition; temporary integrations |
| Parallel run | Old and new systems run side by side | Results can be compared; fallback available | Double work; higher cost |
| Pilot | A small group uses the system first | Real feedback with limited exposure | Pilot group may not represent all users |
| Blue-green or canary (technical) | New version deployed beside the old; traffic switched or gradually shifted | Quick rollback | Needs duplicate environments and tooling |
Choose according to business criticality, how reversible the change is, data complexity and user numbers. A payroll system might use a parallel run for one cycle; an internal tool might go big bang.
Go-Live Readiness Checklist
- Solution: testing complete; exit criteria met; no open critical or high defects; performance and security tests passed (see Quality Management and Testing).
- Data: migration rehearsed at least once with timings and reconciliation checks; backup taken.
- Infrastructure: production environment built, monitored and capacity checked; certificates, firewall rules, DNS and access in place.
- Integrations: interfaces tested end to end with third parties, and their teams available during cutover.
- People: users trained; super-users identified; support team trained with runbooks.
- Business: process changes agreed, policies updated, communications sent and blackout dates checked.
- Governance: approvals obtained, change records raised and compliance checks done.
- Contingency: rollback plan tested and decision criteria agreed.
The Go/No-Go Decision
Hold a formal go/no-go meeting shortly before cutover, with the sponsor, business owners, technical lead, test lead, operations and support. Review the checklist item by item, with a clear owner and status. Agree criteria in advance so that the decision is based on evidence and not on the emotional pressure of a deadline. Document the outcome. Going live with a known serious defect because “the date is fixed” transfers the problem to the users and the support team; sometimes the right answer is a short delay.
The Cutover Plan
A cutover plan is a minute-by-minute runbook for the transition. For each step it records the task, owner, planned start and duration, dependencies, verification and rollback point.
| Time | Task | Owner | Check |
|---|---|---|---|
| Fri 18:00 | Announce freeze; stop transactions in old system | Business lead | Confirmation from all departments |
| Fri 18:30 | Take full backup of old database | DBA | Backup verified by restore test |
| Fri 19:00 | Run final data migration | Migration lead | Record counts and totals reconciled |
| Sat 01:00 | Deploy application; smoke tests | DevOps lead | All smoke tests pass |
| Sat 03:00 | Rollback decision point | Project manager and sponsor | Go / roll back |
| Sat 06:00 | Switch DNS; enable user access | Network lead | Login tests from several locations |
| Sat 08:00 | Business validation of key transactions | Super-users | Signed checklist |
| Mon 08:00 | Open for business; hypercare begins | Support lead | Support desk staffed |
Rehearse the cutover at least once, preferably in a production-like environment, and time every step. Rehearsals reveal missing permissions, slow migrations and unclear handoffs. Maintain a contact list of everyone involved with phone numbers, and a clear communication channel such as a dedicated bridge call during the event.
Rollback Planning
Hope is not a plan. Decide before cutover what would trigger a rollback (for example, data reconciliation fails or key transactions do not work by a certain time), who has authority to call it, the steps and how long it takes. Identify the point of no return, after which rollback is no longer feasible, for example once the old system has been decommissioned or users have created new data that cannot be reversed. Test rollback just as you test deployment.
Training and Change Readiness
A system that users cannot or will not use is not successful. Plan role-based training close to go-live, using real scenarios; provide quick reference guides and short videos; identify super-users who can help colleagues; and communicate early why the change is happening and what changes for each group. Measure readiness with short assessments and feedback. A structured approach to the people side of change, often called organisational change management, is as important as the technology.
Hypercare and Handover to Support
Hypercare typically lasts from a few days to several weeks. During this period:
- The project team stays available, with extra support staff on the help desk.
- Issues are triaged daily and fixes are prioritised.
- Key metrics, such as errors, response times, usage and ticket volume, are monitored closely.
- Knowledge transfer to the operations team continues.
Define exit criteria for hypercare, for example “no critical incidents for five working days and ticket volume below an agreed level”. Then formally hand over to operations with runbooks, architecture documents, known-issues lists and support contacts. Continue with Project Closure and Lessons Learned.
Common Mistakes
- Never rehearsing the cutover or the rollback.
- Scheduling go-live at a business peak or just before a holiday.
- Forgetting third-party and integration teams during the cutover window.
- Letting the deadline override agreed go/no-go criteria.
- Disbanding the project team the day after go-live.
- Choose a deployment strategy by risk and reversibility: big bang, phased, parallel run or pilot.
- Use a readiness checklist and a formal go/no-go meeting with predefined criteria.
- A cutover plan is a timed, owned runbook that has been rehearsed.
- Plan rollback, training and hypercare, then hand over to support formally.