Documents that are worth writing
Three documents that are almost always produced and rarely useful. The failure is the same each time: they record activity rather than commitment.
Out of scope is the charter's most valuable section
It is also the one most often skipped, because writing it means saying no to something specific while everyone is still enthusiastic.
Every exclusion written down is an argument you do not have in month four. A charter listing ten deliverables and nothing excluded has quietly agreed to whatever anyone assumed was included.
Measurable means numbers or dates
"Improve customer experience" is a direction, not a criterion. At the end nobody can say whether it happened, so the project cannot be judged and the next one learns nothing.
"Reduce average handling time from 6 minutes to 4 by 31 March" can be assessed by anyone, including people who were not there. That is the entire purpose of writing it down.
Never generate a governance document
A charter is a record of what people agreed. Filler text — however plausible — ends a conversation that needed to happen, and it does so invisibly: the section looks complete, so nobody asks.
A blank prompts an argument. That argument is the actual value of the exercise.
A status colour must trace to a number
Colours chosen by feel are how a project reports green for three months and then slips a quarter. Nobody lies, each judgement is individually defensible, and the aggregate is a fiction.
An indicator derived from figures, showing the threshold that produced it, can be challenged — and a report that cannot be challenged cannot be trusted either.
Overall should be the worst of its parts
The common corruption is an overall status better than its components, because the conversation was difficult. A late milestone or an open high-severity risk should push a report off green on its own, regardless of what the other indicators say.
Lessons need a cause and an owner
"The integration was three weeks late" is a fact about this project and useless to the next. "We committed to an integration date before anyone had seen the third-party API" is a pattern the next team can recognise in time.
And a recommendation nobody owns is a wish. The universal failure of lessons learned is that they are captured well and never reach anyone who could act on them.
An all-failure review says more about the room
If a retrospective finds nothing worth repeating, the usual explanation is not that nothing went well — it is that the session did not feel safe, or was run as an accountability exercise. Such reviews also tend not to be believed, which wastes the genuine findings alongside the missing ones.