A senior stakeholder insists on an approach you believe is wrong. How should this be handled so the organisation reaches a good outcome regardless of who is right?

First, check who is actually right

The stakeholder may be correct, and may be optimising for something you cannot see — a commercial commitment, a partnership, a regulatory conversation, an organisational constraint. Disagreements about technical approach are frequently disagreements about objectives, and identifying which one you are having is the first move.

The productive question: "what are you optimising for that I might not be weighting properly?" It is not rhetorical, and the answer often resolves the disagreement entirely — usually by revealing a constraint that makes their position correct.

Separate the reversible from the irreversible

This is the most useful distinction available. If the decision is cheap to reverse, the correct action is usually to try it their way and gather evidence. Arguing about a reversible decision costs more than testing it, and being wrong costs little.

If it is expensive to reverse — a data model, a public contract, a platform commitment, a vendor lock-in — then the disagreement is worth escalating and the escalation should say so explicitly: "I would normally defer here; I am pressing because this is hard to undo."

Make the disagreement concrete

• State your position as a prediction with a condition: "I expect this to fail when we reach X users, because Y." A prediction can be tested; an objection cannot.
• Propose the cheapest experiment that would settle it — a load test, a prototype, a spike, a proof-of-concept on the risky part. Converting an argument into an experiment is the single most effective move available, because it removes the need for either party to be persuaded.
• Identify the specific evidence that would change your mind, and say what it is. This is credible, it is reciprocated, and it demonstrates that the position is analytical rather than positional.
• Quantify the consequence of being wrong in each direction, because the asymmetry frequently decides it.

When it remains unresolved

Disagree and commit, in writing, without hedging. Record the decision, the reasoning, your dissent and the specific conditions under which it should be revisited — stated neutrally, as a monitorable trigger, not as a prediction of vindication.

Then commit genuinely. A half-executed approach fails for reasons unrelated to the decision, which proves nothing, damages the organisation and destroys your credibility for the next disagreement.

Escalate only when the decision is both irreversible and materially harmful, and escalate the decision, not the person — presenting both positions fairly, including the strongest version of theirs. An escalation that misrepresents the other side is resolved against you regardless of the technical merits.

The long game

Credibility is the currency, and it is spent on disagreements. An architect who is right about the things they escalate, who defers gracefully on the things they do not, and who reports outcomes honestly afterwards — including when they were wrong — accumulates the standing that makes the next disagreement easier. One who objects to everything has none, and their genuine concerns are discounted with the rest.