- Pilot for long enough to hit a weekend and a stock reorder, or you have not tested it.
- Write the prep, storage and waste implications down, not just the recipe or spec.
- Launch the check with the change: a spot check in week one catches the sites that quietly did not.
Pilot long enough to hit the hard parts
A three day pilot tests novelty. It does not test what happens on a Saturday, or when the first reorder lands, or when the person who was trained is on holiday. Those are where rollouts actually fail.
Two weeks including a weekend and at least one stock cycle is the usual minimum for a change of any operational weight.
Document the consequences, not just the change
Most launch packs describe the new thing. Fewer describe what it does to prep time, to storage space, to the order guide, to waste, and to the shift where it is busiest.
Those consequences are what determine whether a site can actually run it, and they are usually discovered at the twelfth site rather than the second because nobody wrote them down after the pilot.
Pilot across a weekend and at least one reorder. Most of what breaks a rollout is invisible in the first few days.
Sequence the estate deliberately
Going live everywhere on the same day feels decisive and removes your ability to learn. Going site by site takes forever. A sensible middle is waves: a small first wave with your strongest sites, then the bulk, then the sites that need support.
Between waves, actually change the pack. A wave structure with no revision is just a slow simultaneous launch.
Check in week one, not month two
Every rollout has sites that quietly did not do it: the signage never went up, the old spec is still being made, the training never happened on the late shift. In week one that is a quick correction. In month two it is a habit.
A short, specific spot check in the first week costs very little and is the difference between a launch and an announcement.
Frequently asked questions
How many sites should be in a pilot?
Two or three, chosen to be different from each other. A pilot at three similar sites tells you about one kind of site.
Should the pilot sites be your best ones?
Include one strong and one average. A pilot that only runs at strong sites will not surface the problems the rest of the estate will hit.
How do I know the rollout actually happened?
Ask for a photo of the change in place at each site during week one. It takes a minute per site and is the only reliable answer.