Technical decisions that need a stakeholder
Some technical decisions should be made by engineers alone. Others need explicit stakeholder input, and getting the line wrong causes friction later.
Some technical decisions are for engineers to make. Others require a stakeholder — product, security, compliance, executive — to sign off. Getting the line wrong causes friction later.
Written October 2024 from experience leading technical decisions across teams.
What's usually engineer-only
- Code structure and internal API design.
- Library choices within a team's scope.
- Testing strategy and coverage.
- Local performance optimisation.
- Refactoring within an existing service.
What usually needs a stakeholder
- Choice of data store. Because operational cost, backup posture, and compliance implications all reach beyond engineering.
- Choice of external vendor. Because contracts, cost, and lock-in are business decisions.
- Change to authentication or authorisation model. Because security posture is not just engineering.
- Cross-team API contracts. Because the affected teams need input.
- Migration between major architectures. Because the cost is high enough to be a strategic decision.
The failure modes
- Engineers making stakeholder-scope decisions in isolation. The decision gets reversed later; the work is wasted; the stakeholder feels blindsided.
- Stakeholders making engineer-scope decisions. The stakeholder overrules the team on something they don't have context on; the team disengages.
- Ambiguity about which is which. Nobody knows if the decision is theirs; the decision stalls.
The clarity move
Every project should have a short list of decisions that need stakeholder input, agreed upfront. Everything else is engineer decision. Ambiguity resolved by conversation, not by resentment.
This is one of the higher-leverage moves an engineering leader makes. Getting it right saves months of friction. Getting it wrong is one of the most common causes of team dysfunction.