How can a candidate tell which onsite round carries calibration weight, and adapt when it is unclear?

## The detection problem Candidates want to know which slot carries the most weight, because effort feels like a budget. In practice the identification is unreliable, and building a strategy on it is worse than not trying. The honest signal list is short and every entry is probabilistic: - **Organisational distance.** The strongest hint. If the interviewer's team has nothing to do with the role — a different product area, a different discipline — independence from the hiring team is at least plausible, and independence is the defining feature of a calibration role. - **Question range.** Peer slots stay inside their declared topic. A calibrating round often walks the whole arc of your history and then drills into one piece of it, because the write-up has to speak to a durable standard rather than to one skill. - **Standard-relative probing.** Repeated *compared to what*, *what would have happened otherwise*, *how would you know it worked*. The calibrator is building a comparison against people already at the rung. - **Note-taking of your words.** Verbatim-ish capture, sometimes with a pause to catch up, because the scorecard is expected to quote what you actually said rather than summarise an impression. - **Slot position and length.** Nearly worthless. Late slots and slightly longer slots are usually calendar artefacts. And there are two things that are *not* signals, though candidates read them as such: seniority (plenty of very senior people sit ordinary panel slots) and friendliness (a calibrating interviewer often has the warmest room of the day, because they are not defending a design and have no stake in the outcome). ## Why the detection barely matters Suppose you guess right. What would you actually do differently — try harder? Candidates who allocate effort unevenly across a day do not raise their ceiling in the round they picked; they lower their floor in the ones they wrote off. And the rounds they write off still file scorecards. Suppose you guess wrong. Now the weightiest write-up in the packet was produced by the interviewer you decided did not matter. The asymmetry is the whole argument. A wrong guess is expensive and a right guess buys almost nothing, so the correct posture is to treat every round as though it might be the one whose scorecard can block the hire. ## The adaptation that actually pays **Make every round self-contained.** Each interviewer writes their own scorecard, and it is read by people who were not in either room. Evidence you gave in round two does not appear in round five's write-up. So restate the frame every time: what the system was, who else was involved, over what span, what you owned versus influenced. It feels repetitive to you; the write-up is empty without it. The tell that a candidate has not internalised this is the phrase *as I mentioned earlier*. To an interviewer from another organisation it points at something they never heard, and it leaves a gap on the scorecard exactly where the evidence was supposed to be. Say the thing again, briefly, and move on. **Budget material across the day.** Bring three or four genuinely distinct pieces of work, each rehearsed to stand alone. If you have one story and six rounds, the debrief sees one data point regardless of how well you tell it — and the calibrating write-up, with no team context to cushion it, is the one that says so most plainly. **Ask the recruiter the right question.** What each slot covers, and in what format, is normal pre-brief information and most recruiters share it happily. Which interviewer can block the hire is a different question, and asking it signals that you intend to spend effort unevenly. **Assume no shared context anywhere.** Even inside a team, panel members often have not spoken before the debrief. Treating every round as a fresh audience costs you a sentence or two per answer and removes an entire class of failure. ## A worked illustration A platform-infrastructure candidate has a six-round day and one strong migration story. Rounds two and three both get it. By round five — an engineer from an unrelated organisation, asking broad history questions and taking careful notes — the candidate opens with *I think I covered this with your colleague earlier, but briefly...* and gives a compressed version, ten minutes shorter than the original, missing the constraints and the alternatives rejected. That compressed version is what lands on the scorecard with the most weight attached to it. Nobody in the debrief hears the good version; they read the thin one, next to two other write-ups describing the same project. The candidate's failure was not the guess about which round mattered. It was that they had prepared a performance instead of a portfolio.