We went back through thirty-one operations engagements from the last eight years — every one where some form of standardization was the objective. Eleven of them had followed a previous attempt by somebody else that had not worked.
The failed attempts had a great deal in common, and almost none of it was about the quality of the process design.
The standard was usually fine
In nine of the eleven cases we could find the earlier work. The target processes were sensible. Several were better than what we eventually implemented, because the earlier team had more time for design and less time spent arguing.
What they had in common was the direction of travel: designed centrally, validated with a steering group, then rolled out to sites that had not been part of designing it.
The local optimum problem
Here is the thing that makes standardization different from most change programs. In a multi-site business that grew by acquisition, each site has usually spent years optimizing for its own constraints — its equipment, its labor market, its customer mix.
So when you impose a common standard, some sites get better and some get demonstrably worse. Not "worse until they adapt" — actually worse, permanently, because the standard cannot accommodate their constraint.
The people resisting a standardization program are frequently right about their own site. That is precisely what makes it hard.
Change management training tells you to address resistance by communicating the vision more clearly. This does not work when the resistance is empirical rather than emotional.
What the successful engagements did differently
Three things showed up in every case that worked:
- A shared baseline came first. Before any design work, four to six weeks measuring the same things across every site on a consistent basis. In most businesses this has genuinely never been done.
- The sites designed the standard. Not consulted on it — designed it, together, in a room, with the baseline in front of them.
- The pilot was at a weak site. Piloting where things already work proves nothing and convinces nobody.
On the baseline
The measurement phase is the part clients most often want to cut, and it is the part that determines whether the rest works. It is unglamorous, it produces no recommendations, and it typically takes a fifth of the engagement.
What it produces is a set of numbers that nobody in the room can dispute. Once that exists, the conversation stops being about whose process is better and starts being about what the numbers show — which is a conversation that can actually conclude.
What this means if you are planning one
If a previous standardization attempt failed at your company, the instinct is to conclude that the design was wrong or that people need more convincing. Check the baseline question first. In our sample it explained more than everything else combined.
