An architect faces conflicting requirements: sales wants a feature in six weeks, security requires a control that takes twelve, and the platform team wants it built on the paved road that does not yet support it. How should this be handled?

What not to do

Do not resolve it technically. This is not an architecture problem wearing a stakeholder costume; it is a genuine conflict between legitimate interests, and an architect who quietly picks one has made a business decision without the authority to make it.

Do not average it — a compromise satisfying nobody, delivering a weakened control late on an unsupported path.

What to do

1. Separate the negotiable from the non-negotiable. Some requirements are constraints and some are preferences. A security control required by a customer contract is a constraint; a preference for the paved road is strong and negotiable. Naming which is which usually collapses half the conflict immediately.

2. Make the trade-offs explicit and costed. Not "we cannot do all three" but three concrete options with what each costs:

• Ship in six weeks off the paved road with the control, accepting migration debt with a named owner and date.
• Ship in twelve weeks on the paved road with the control.
• Ship in six weeks with a compensating control and a defined window to implement the full one.

3. Identify who decides. Not the architect. Someone accountable for the revenue and the risk — usually one level above all three parties. The architect's job is to make the decision makeable, not to make it.

4. Record it, including the option chosen, the rationale and the conditions, so a pattern of similar decisions is visible.

The architect's actual value here

Producing the third option nobody had thought of. In this case usually a compensating control: something cheaper that reduces the risk enough to ship, with the full control committed and dated.

That is where architectural judgement pays — not in adjudicating between stated positions, but in finding the option that satisfies the underlying interests rather than the stated demands.

The pattern to watch

If this conflict recurs monthly, the problem is not the individual decisions. It is that the paved road lags the product's needs, or that security requirements arrive too late to shape design. A recurring conflict is a systemic signal, and treating each instance as a one-off negotiation guarantees it recurs.