Designing an interview process that measures the right thing

When a senior loop grows to 6 or 7 stages, it's usually accretion. A round added after a bad hire, a stakeholder who wanted a look, a panel that never got retired. Each addition felt prudent in isolation. Together they build a process that mostly measures endurance: whether a candidate will tolerate weeks of repetitive conversation. That's not a trait you're hiring for, and the people most willing to endure it are often the ones with the fewest other options.

The cost lands on exactly the candidates you want most. A sought-after passive VP will sit through 2 or 3 well-run stages. They won't sit through 7, and they shouldn't have to. Every superfluous round is a filter that removes your strongest talent while adding almost nothing to the decision. So you tend to end up hiring the candidate who had the patience to stay to the end, which is rarely the same as the best one the search could have surfaced.

The instinct behind a long loop is defensible: a senior hire is expensive to get wrong, so more looks feel like more safety. But rounds and rigour aren't the same thing, and past a point they pull in opposite directions. Rigour comes from what you measure and how sharply. Rounds are just how many calendars you ask a candidate to clear. You can pile on rounds without sharpening the measurement at all, and most sprawling loops have done exactly that. They stack conversations that circle the same few impressions instead of testing anything new.

Meanwhile the drop-off is real and it compounds. Every additional stage is another window in which a competing offer lands, a counter is raised, or a candidate simply decides the juice isn't worth the squeeze. The attrition selects hardest against the people with the most options, who are the people you were running the search to find in the first place. A loop that loses its top candidate to fatigue was slow, and slowness got mistaken for care.

The way to redesign a loop is to force every stage to justify itself against a single test: what signal does this produce that no other stage produces? A technical screen measures capability. A working session shows how they think in real time. A leadership conversation gives you a read on how they'd operate with the team and the parent org. If two stages are chasing the same signal, one of them is just another chance for the candidate to walk.

Keep the stages that measure distinct things and sequence them so the highest-signal, lowest-cost checks come first. What survives is usually a tighter loop that reads a candidate more sharply than the sprawling one did, because every hour in it is doing work, and the process is closing the assessment instead of exhausting the person. The discipline here is subtractive. The question to keep asking is what you can retire without losing a stage the decision actually turns on.

Once you accept that the loop's job is signal and not endurance, the design question narrows sharply. A CV already tells you where someone has worked, what they shipped, and how their career has compounded. Re-litigating that across 4 rounds is theatre. The loop's real work is the handful of things the paper can't carry: how a person reasons under ambiguity, how they behave when a problem has no clean answer, whether their account of a past win survives a specific, probing question, and how they'd actually operate inside your particular org and its politics.

You get these signals from depth. A single well-run working session, built around a real problem drawn from the actual role with the candidate thinking aloud, tells you more than 3 structured interviews that each skim the surface. And a reference conversation run properly, with someone who has watched the candidate under pressure, surfaces things a candidate will never volunteer about themselves. Design the loop around those moments and protect the time they need. The stages worth keeping are the ones that could actually change your mind. The rest are usually there to confirm a decision you've half-made already.

There are two kinds of time in a loop, and organisations routinely confuse them. The first is process time: the days lost to scheduling, the week a piece of feedback sits in someone's inbox, the fortnight a decision waits for a calendar to clear. That time is pure waste. It costs nothing to remove and every day of it bleeds candidates, because a senior professional reads a slow process as a signal about how the organisation runs. Speed here is free: same-day scheduling, feedback within a day, calibration done before the loop opens rather than argued after it closes.

The second is judgment time: the hours it takes to actually assess the things a CV won't show, to run the back-channel, to sit with a reservation instead of rationalising it away. That time is the product. Yet under deadline pressure it's the first thing compressed, precisely because it's the hardest to schedule and the easiest to skip. The move is to reallocate. Collapse the process time aggressively and hand the hours you reclaim to the parts of the assessment that reward patience. The candidate experiences a crisp, decisive process and never sees the deliberation behind it, which is exactly right. Speed bought by skipping judgment just puts an earlier date on the same expensive mistake.

Put the pieces together and a method falls out. Start from the decision. Name the 3 or 4 things that would actually change your mind about this hire, most of them things the CV can't show. Assign each one to exactly one stage. Retire every stage that maps to nothing new. Then run the whole thing fast on the process clock, scheduling, feedback, calibration, the final call, and slow on the judgment clock, giving the working session and the reference the time they need to reveal something.

A loop built this way is usually shorter than the one it replaces, and it reads candidates more sharply, because every hour in it is doing assessment rather than manufacturing false confidence. It also closes faster, which matters most for exactly the people you most want: the ones who won't wait through a 7th round and were never going to. The goal was always a process that measures the right thing, and gets out of its own way everywhere else.