An architect must explain to a product leader why a requested feature will cost 150ms on the trading path. How should the trade-off be framed?

The framing that works

Present options with consequences, not an objection. "We cannot do that" invites a negotiation about whether engineering is being obstructive. "Here are three ways to do it and what each costs" invites a business decision.

For this case:

• Option A: on the critical path. Ships in two weeks, adds 150ms to every order placement. State what that means concretely — orders acknowledged more slowly during the open, and a measurable competitive disadvantage.
• Option B: asynchronous, off the critical path. Ships in five weeks, no latency impact, the feature's result appears a second later. State what that costs the user experience.
• Option C: precomputed. Ships in eight weeks, no latency impact, no experience compromise, higher operational complexity.

Why this framing changes the outcome

It moves the decision to the person accountable for the business consequence, with the information they need. It also makes the constraint concrete rather than a matter of engineering preference — 150ms on the order path is a fact about the system, not an opinion about priorities.

And it demonstrates that the answer is not "no", which preserves the credibility needed for the cases where the answer genuinely must be.

The vocabulary to use

Consequences the listener already cares about. Not "it adds a synchronous call to a service with a p99 of 150ms" but "orders will be acknowledged 150ms more slowly, and during the market open that is when customers notice most".

The translation is the architect's job and it is where most of these conversations fail — not because the analysis was wrong but because it was delivered in a vocabulary the audience could not act on.

The thing to avoid

Anchoring on the worst option to make the good one look reasonable, which is manipulative and is detected. And presenting a false binary, because a binary against a revenue deadline always resolves the same way, and engineering's credibility is spent doing it.