whatactuallydecideswhoranksfirst
No candidate ever collapses into a single number. λ-CORE scores six dimensions independently and ranks each one against the pool. Drag the wrong lever and the whole shortlist reorders, which is exactly the point, made hands-on below.
Tl;dr
λ-CORE never reduces a candidate to one arbitrary score. It evaluates six behavioral dimensions independently, each with a confidence interval, and ranks every candidate Relative to the actual pool that applied. Change what a hiring team is weighting, and the shortlist reorders, because rank was never a fixed property of a person to begin with.
Why one score was always the wrong idea
Most screening tools reduce a candidate to a single number: an ATS match percentage, a fit score, a resume grade. The number feels precise. It is usually hiding a decision that was actually made across several dimensions and then flattened into one, with no way to see which dimension drove the outcome.
A candidate can be exceptional at structured reasoning and merely solid at collaboration. A single blended score erases that distinction. A hiring manager who needs a strong communicator for a customer-facing role and one who needs deep technical depth for a research role are looking for different things from the same pool, and a flat score cannot tell them apart.
The six dimensions
Every aperture interview is scored across the same six behavioral dimensions, independently, with evidence traced back to what the candidate actually said.
Try it: reorder the pool
Same eight candidates, same interviews. Pick a dimension below and watch who rises. Rank is not a fact about any one of them, it is a function of what you are asking the pool.
Try it yourself
Same eight candidates. Pick a dimension, watch the pool reorder.
- 01Amara o.86
- 02Marcus w.84
- 03Devon k.83
- 04Priya s.82
- 05Sofia r.82
- 06Hana t.82
- 07Leo m.78
- 08Jin l.77
Illustrative candidate data, not real applicants. The point stands either way: rank is relative to the pool and the dimension being weighted, never a fixed property of a person. Change what you are ranking on, and who looks strongest changes with it, which is exactly why λ-CORE reports all six dimensions plus a confidence interval instead of collapsing a candidate into one score.
What pool-relative actually means
A raw score means little on its own. An 78 out of 100 on domain knowledge could be the best in a weak pool or the weakest in an exceptional one. λ-CORE reports where each candidate stands Relative to everyone else who applied to the same role, So a hiring team is reading a real comparison, not a number floating with no reference point.
This is also why the same candidate can look different across two roles. The pool changes, so the relative position changes, even if nothing about the candidate did.
Why every score has a confidence interval
A fifteen-minute conversation is real signal, not certainty. λ-CORE reports a confidence interval alongside every dimension score, so a hiring team can tell the difference between "we are confident this candidate is strong on reasoning" and "this reads as strong, but the evidence was thin, look closer before deciding."
A flat score with no confidence attached hides exactly the cases that most need a human to look twice. This is where that gets surfaced instead of buried.
How this becomes a shortlist
A hiring team does not read six numbers per candidate across five hundred applicants. λ-CORE surfaces a ranked shortlist by default, but every dimension, every confidence interval, and the evidence behind each score stays one click away. The summary is a starting point, never the only thing a team is allowed to see.
For the full mathematical model behind pool-relative comparative scoring, and the case for why it beats a flat rubric, we wrote two deeper pieces.
Keep reading
related from aperture.
hiring
Hiring is broken. The numbers say how badly.
The structural failure nobody is fixing
hiring
Your best applicants never even reach you
The case for moving beyond keyword filters
candidates
Can an AI interview discriminate against you?
A careful answer about fairness, what gets evaluated, and what gets ignored
From the product