- Pilot at two sites: your best and your worst. If it only works at the best one, it is not a standard yet.
- Launch with the check already built. A standard with no check attached is an announcement.
- Week three is where standards die. Schedule the follow-up before the launch, not after.
Pilot at the extremes, not the middle
The instinct is to pilot at a good site, where it will go well. That tells you almost nothing. Pilot at your best site and your worst one, because the gap between those two results is what the rollout will actually look like.
If the standard only works where the team was already strong, it is not a standard, it is a description of what good people already do.
Launch the check at the same time as the standard
Standards that arrive as a document decay quietly. Standards that arrive as a scheduled check with a named owner and a piece of evidence survive, because there is a moment each day when somebody has to do something.
This is the single biggest predictor of whether a rollout holds. Not the quality of the document, not the training, but whether the check exists from day one.
Compliance with a new standard peaks in week two and falls in week three. Schedule the follow-up visit before you launch, because after the dip it costs far more to recover.
Expect the week three dip
Compliance is high in week one because it is new, high in week two because somebody is watching, and falls in week three when attention moves to the next thing. Almost every rollout follows this curve.
Because it is predictable, it is manageable. Put the week three review in the diary before the launch. A visit in week three is worth three visits in week eight.
Kill standards that are not working
An estate carrying twenty standards that are half-followed is in worse shape than one carrying eight that are followed completely, because the team has learned that standards are optional.
If a standard is not holding after two honest attempts, either it is wrong or it is not important enough. Withdraw it publicly rather than letting it rot, and the remaining ones keep their authority.
Frequently asked questions
How many standards can a site absorb at once?
One meaningful change at a time. Two if they affect different parts of the day. More than that and none of them stick.
Should head office write the standard or the sites?
Head office writes the draft, the pilot sites break it, head office rewrites it. A standard written entirely centrally usually fails on a detail nobody at head office knew about.
How long before a standard is a habit?
Assume eight to twelve weeks of active checking before it survives without attention, and expect to re-check it after any period of high staff turnover.