An ADR that nobody understands is as good as none. And an ADR that collapses at the first objection may not have been one. This lesson practices what lies between writing and implementing a decision: presenting it in a short time, taking in objections in a way that makes the decision better, and settling beforehand who decides in the end. It is also the rehearsal for the architecture review in the final project (lesson 4.9.7).

Ten minutes

Ten minutes are enough for a decision if it is well structured. This division has proven itself:

Minutes Content The question that must be answered
1 Problem Why do we have to decide now, and what happens if we don't?
2 Options Which ways are there really, including “do nothing”?
3 Trade-off What do we measure the options by, and how do they score?
2 Recommendation What do you propose, and what do you give up for it?
2 Risks and the way back What can go wrong, how will we notice, and how do we get back?

The trade-off gets the most time, because that is where the deciding happens. Microsoft names the yardsticks by which an architect should measure her decisions: constraints, trade-offs, effort, reversibility and risk. Reversibility is often forgotten and is the most reassuring piece of information for your audience. “Can be rolled back in two weeks” calms a discussion more than any number. And conversely, a decision that cannot be rolled back deserves more time and more people.

Three rules for the talk itself:

  • Show the ADR, not slides with bullet points. The document is the talk; whoever reads along sees that nothing important was only said. A diagram may be added if it answers a question faster than text.
  • Start with what changes for your audience (lesson 4.14.1). In front of a mixed audience that means: first the consequence for the business, then the technology that explains it.
  • Keep to the time. Whoever is at the recommendation after ten minutes has no time left for objections, and those are what make the decision good.