In the inquest after a difficult IT migration, the temptation is to focus on the weekend itself. In our experience, that is almost never where it went wrong. The weekend is just where the consequences became visible.
There are four earlier moments that matter much more. Get them right and the cutover is dull. Miss them and no amount of weekend heroics will save it.
Moment one: the discovery you do not want to write down
Every migration discovery finds at least one thing the team would rather not surface — a system nobody owns, a licensing position that does not quite work, a dependency that complicates the timeline. The instinct is to keep going. The right move is to stop, document it, and have the awkward conversation now.
Issues that are written into a RAID log in week two are project management. The same issues, discovered on cutover weekend, are crises.
Moment two: the pilot wave that "mostly went well"
Pilots are not for proving that the migration works. They are for finding the issues that were not in the discovery. A pilot that "mostly went well" with a handful of niggles is not a pilot that has finished — it is a pilot whose findings need closing out before a bigger wave runs.
The discipline is unromantic: every pilot issue gets a root cause, a fix and a verification step before the next wave is scheduled. Skip this and the second wave will fail in exactly the same way, only louder.
Moment three: the comms plan that nobody believes
When users feel surprised on cutover weekend, the migration is in trouble. Surprise is almost always a comms failure that started weeks earlier — meeting invites that were "informational", floor walkers nobody asked for, training that did not feel relevant.
A good comms plan is short, repetitive and arrives in the channels people already use. Three messages, three times each, in the channels people actually read. Less clever than most comms plans; far more effective.
Moment four: the rollback that is theoretical
Every migration plan has a rollback section. Most of them have never been tested. Walk through it line by line a week before cutover, in a room, with the people who would actually execute it. If anyone hesitates, the rollback is not real.
You may never use it. The point is that you can — and that knowledge changes how confidently the team handles surprises on the weekend.
A quiet weekend is a leading indicator, not a lucky one
When a cutover goes calmly, it is almost never luck. It is the cumulative effect of four earlier conversations going well. That is the work; the weekend is just the receipt.
