Here is my current view on guidelines and standards (aggregated from Lean, experiences and other sources):
- Current best way of working, No such thing as “Best practice” because that only kills motivation
- I don’t want guidelines that prevent people from thinking
- Try to make them executable
When do we use guidelines?
1) When new to a domain
2) As a basis for improvment (where are we now and where is our target condition)
3) When you don’t want to make a decision and want to be “safe”
The problem with guidelines is that as soon as you are familiar to the domain of the problem, you will most likely not read any guidelines on the subject. The value from written content will be minimal, because you will not read it. The questions is also how much attention you will have on new guidlines that differs from your current way of working?
That is why I really like the idea of making guidelines executable, so when you are not following the current best way of working, you will be notified.
What kind of guidelines are not possible to make executable? How much effort is it worth creating them?
Johan Nordin's Weblog 
Inexecutables, some examples:
1. High level guidelines/principles which guide the development of methods and other guidelines (“We value people over process”, f’rinstance).
2. Any guideline which includes a value judgement (there must be an architecture document describing stakeholders, NFR’s and their realization to a sufficient level for the solution lifecycle — you can’t perform that valuation without intelligence).
3. Any guideline which intends to influence behaviour and have effects outside the domain of IT (“Meetings without an agenda must be adjourned”).
4. Any situation where the cost of automatic verification is exorbitant (e.g. with need for complex integration to external information sources, embodying complex business rules better described by and interpreted as language, containing statistical valuations, or relying on partially unreliable data).
Shying away from providing guidance because it cannot be automated would be foolish. However, there is a “total cost of guidance” to bear in mind: the more guidelines are automated, the more space there is for higher-value guidance requiring judgment.