I whipped up something quickly to explain the lean approach to standards to some clients today, and thought I would share...
1) when a team starts work it has a set of standards to use that describe how to complete the work. Often these standards are based on standardized work types, but they can also be based on process, team context, or be global
2) the team has two choices, use the standard, or to suggest an improvement
3) if the teams suggests an improvement they must make a prediction of improved performance
4) the team then completes the work, and measures performance
5) if performance is improved than the suggested improvement becomes the new standard forthat team
6)Team standards can be spread to other teams, becoming standards that apply to processes, standard work types, and even globally. This is facilitated by various improvement events such as operational reviews, retrospectives, etc with the assistance of managers and the QMO
7) Managers can help teams by recommending optional practices and suggestions for improvements. Again a prediction should be made, work should executed for a period of time, and the results should be measured.
This approach largely holds regardless of the "kind" of standard we are talking about. Technical, coding, process, architecture etc are at there most effective when they are measurable, and when they can changed through safe-to-fail experimentation. Managers govern the process, and make sure that standards are not just ignored, changes must go through a scientific process.
Managers can also suggest sweeping, global rules, but again we need to measure. With Kanban our measures are based on features. Specifically throughput, lead time, failure to value ratio, and failure intake at the feature level.