An interviewer asks - tell me about a time you told a sponsor their ask was wrong when you were only about 60% sure you were right. What did you actually say? What is being tested, and what does a strong answer contain?

What the interviewer is testing

Whether you can act on partial information without either overstating what you know or hiding behind the uncertainty. Both failures are expensive and both are common. The architect who announces "this will not work" at 60% confidence and is wrong twice has spent the standing they will need at 95%. The architect who says nothing lets a known-plausible failure through and, when it lands, has to explain a concern they recorded nowhere.

The question is also a calibration test. Candidates who cannot say a number out loud usually cannot say one inside a project either.

The clarifying questions that change the answer

Is the ask reversible, and until when? What would it cost to find out cheaply? Who holds the missing 40% — often a single person in another team. And what is the sponsor really optimising: the mechanism, the date, or a commitment they have already made to someone else? The last one changes everything, because a date that has been promised externally is a constraint, not a preference, and the conversation becomes about scope.

A strong answer's arc

1. State the confidence as a number and the belief as a falsifiable claim. "I think there is a better-than-even chance this cannot hold January's volume — specifically, I expect writes to exceed what one primary can take at about four times current traffic."
2. Name what would change your mind before anyone asks. This is the move that makes 60% useful instead of irritating, and it is the part weak answers always omit.
3. Offer the cheapest test and its price. "Two days of load testing against a restored copy settles it."
4. Pre-commit to the outcome. If the test says you were wrong, you build their design and say so where others can hear it. That commitment is what buys the two days.
5. Separate the irreversible part and ask for only that to be deferred. Most asks are 90% reversible, and arguing about the whole thing when you only need the one-way door protected is how architects lose arguments they could have won.

Common weak answers

• Certainty theatre — presenting 60% as 95% because it persuades better. It works once.
• The hedge that commits to nothing: "there are trade-offs and risks here." The sponsor cannot act on it, and after a failure it reads as having known and not said.
• Escalating to the sponsor's manager first. It spends credibility you will need later and guarantees the next conversation is defensive.
• Filing a risk-register entry. A risk nobody will read transfers blame rather than changing an outcome.

What a strong answer adds

Confidence is a currency with a track record. Say 60% and be right six times in ten and people learn to buy at 60%; say "I'm concerned" forever and nobody can calibrate you, so your warnings carry no information even when you are certain. The second-order point is about the test: a sponsor who will not spend two days to settle a question that could cost a quarter is telling you the date is the real requirement. That is useful, and it should change the design rather than the argument.

When this is the wrong answer

At 60% on a reversible, cheap, locally scoped decision, say nothing and let the team find out — the learning is worth more than the saved rework, and intervening at this level on every design makes you the person who objects to everything. Reserve the move for one-way doors: data model, external contracts, anything with a regulator or another company on the far side.