A defensible decision is a technical choice with a clear, explainable rationale. It is not necessarily perfect, and defending it does not mean arguing with your team. It means explaining why the choice fits the business context and constraints—and which downsides you are willing to accept. section 1.2 section 2.2
Turn a gut feeling into a decision your team can understand:
- Identify the goal and constraints. What business outcome matters, and what limits the choice? section 2.1
- Compare realistic options. For each, name at least one benefit and one cost or risk. section 1.8.1
- Choose an option and record what you chose, why, and which consequences you accept. section 1.5.6
- Share the rationale with the team. Decide what to monitor and what change in circumstances would prompt a review. section 1.5.6
Table 1.1 A Simple Architectural Decision “Receipt”
|
Decision
|
Rationale
|
Consequences
|
|
We will implement Option A(adding
an is_favorite column).
|
The business has prioritized Speed to
Ship for this V1 feature to gather user feedback quickly. Option A is
significantly faster to implement.
|
We accept the short-term risk of lower Scalability and a potential effect on database
performance. We will monitor query times on the Articles Table after launch.
We will revisit this decision if the feature sees high adoption or if
performance degrades.
|
This decision receipt favors a faster launch while recording the scalability risk, a monitoring plan, and conditions for revisiting the choice. section 1.5.6
More files, less fragility
A refactoring accepts a few extra files in exchange for clearer responsibilities and easier isolated testing. The choice is defensible because it names both the inconvenience and the improvements it buys. section 3.5.4
A simple pro-and-con list may be enough for a small choice. Significant decisions can use an Architecture Decision Record, while more complex comparisons may call for a weighted-decision model. Afterward, measures such as incident rates or feature delivery time can help evaluate the result. section 1.5.6 section 2.12 section 2.2
