How would you change the technical direction of a team that does not report to you and disagrees with you?

What is being tested

Whether you can produce agreement rather than compliance, and whether you treat influence as a finite resource.

The sequence

1. Understand why they disagree, genuinely. Frequently they know something you do not: a constraint in the code, a previous attempt that failed, an operational reality. Asking first is how you avoid being confidently wrong in front of the people who will implement it.

2. If they are right, change position visibly. This is one of the fastest routes to credibility, because it demonstrates your positions are held for reasons rather than for status. An architect who never concedes is one whose agreement means nothing.

3. Find the shared objective. Almost always there is one. Reframing the disagreement as two routes to the same goal removes most of the adversarial energy.

4. Share reasoning, not conclusions. Someone who understands why applies the principle to the next case; someone who receives a directive applies it literally and stops.

5. Make the trade explicit. "This buys X and costs Y — is that acceptable?" turns an argument someone must win into a decision someone can take.

6. Offer a bounded experiment. "Try it on one service; if the concern materialises we change it." Lowers the stakes and produces evidence, and it is available far more often than people assume.

7. Build the coalition before the meeting. A position agreed with the key people beforehand is confirmed in the room; one raised cold is debated.

What spends credibility

Every strong position, every escalation, every mandate. It is finite.

Which means: decide which battles matter. An architect who objects to everything is ignored on the things that count. Concede readily on the reversible and the stylistic; hold on the one-way doors — and being visibly relaxed about small things is what makes it credible when you are not.

What builds it

Having built things and being willing to read the code. Being specific rather than general — "the payment client shares a pool with the catalogue client" beats "add a bulkhead". Doing the unglamorous work occasionally. Making other people successful. Reliability.

The failure to avoid

Winning the argument and losing the implementation. A decision the team resents is implemented badly, worked around, or quietly abandoned. Agreement is the goal; compliance is the consolation prize.