Change in advances per site · new + priorities vs old screener
The result, site by site
fewer advanced than old
more advanced than old
label shows old→new
Need 1 · highest urgency — takes Levels 1–3 +13 advance
Need 2 · takes Levels 1–2 -10 advance
Need 3 · pickiest — Level 1 only -48 advance
Need-1 sites barely move (+13 overall); the cut lands almost entirely on the pickiest Need-3 sites (−48), which take Level 1 only — exactly where the old screener was over-advancing.
569 NC applications · 6 weeks
What the new screener does with each old call
new level:
Level 1
Level 2
Level 3
Disqualify
Unable
Old: Not recommendreject
136
Old recommends split into real levels — and 40 get disqualified (caught). Old rejects stay rejected (132 disqualify). The 77-strong review pile gets resolved into clear levels.
The two questions · NC · 6 weeks
The answers
Q1 · recruiter time saved
39
interviews avoided — rejected candidates who'd been interviewed for ~0 hires
≈ 5 recruiter-hrs / week
Q2 · hires lost
1
hire lost, out of 46 real hires (2%)
one Level 2 at a pickiest site
Tested against everyone who actually signed an offer: the new screen would have advanced 45 of 46. Still needs the no-show + "no open case" gates — those are separate from screening.