Go-Live Without Surprises: What to Check Before You Launch
Most troubled go-lives fail on preparation, not technology. What to check in the weeks before, on launch day, and in the two weeks after.
Go-live day packs months of work into a few hours. When it goes well, nobody remembers it. When it goes badly, people remember it for years: systems down, users unable to work, data that doesn’t reconcile, and an improvised rollback at midnight.
The good news is that most troubled go-lives don’t fail because of technology. They fail because of preparation, and preparation can be checked.
Weeks before: prepare the decision
Write down the go-live criteria. Before setting the date, agree on what has to be true to launch: tests passed, zero critical defects, users trained, backups tested, support ready. Without explicit criteria, the calendar makes the go-live decision.
Test with real data and real users. Tests run with sample data by the development team won’t uncover what a user with real cases will find on day one. User acceptance testing with key users prevents more problems than anything else.
Rehearse the data migration. Migration is one of the most common sources of surprises: incomplete data, mismatched formats, load times far longer than expected. Rehearse it end to end at least once and time it.
Write and rehearse the rollback plan. Rolling back has to be a decision, not an improvisation. Until what time can you roll back? Who decides? How long does it take? If it’s never been rehearsed, you don’t know whether it works.
Prepare the people. Training, communication, and a network of go-to users in each area. Technology that’s ready with users who can’t use it is a failed go-live by another name.
Go-live week: decide with data
Hold a formal go/no-go meeting. One or two days before, review each criterion with its owner. If one isn’t met, the question is simple: can it be fixed in time, is the risk explicitly accepted, or do we postpone?
Freeze changes. In the final days, only fixes for critical defects get in. Every last-minute change is a new, untested risk.
Build an hour-by-hour runbook. Each activity with its owner, duration, and check. Include explicit decision points to proceed or roll back, and a cutoff time.
Set up the coordination channel. One room, bridge, or chat, with a lead running the runbook and one person with the authority to decide.
Launch day
- Follow the runbook and log each completed step.
- Verify each stage before moving on: versions, data counts, smoke tests.
- Honor the decision points. If a criterion fails, apply what was agreed, not what seems best under pressure.
- Tell users and leadership the outcome as soon as you’re done.
After: the first two weeks
Go-live doesn’t end when the system is up. The first two weeks decide whether people adopt it:
- Hypercare. The project team on hand, with short response times and a brief daily check-in to review incidents.
- Adoption tracking. How many users are really on the new system? Where are they getting stuck?
- Formal handover to operations. With the documentation, procedures, and knowledge support needs so it doesn’t depend on the project team.
An independent review helps
People who have spent months on a project lose perspective on its risks. An outside look, two weeks before the date, usually finds what the team no longer sees. You can start with our free go-live plan template or request a go-live readiness review, which gives you prioritized risks and a clear recommendation in two weeks: go, adjust, or postpone.