Counterfactual Sandbox
Pick a baseline candidate and toggle the levers · inverse-keyword screen, persona version, interview latency, manager variance, sourcing mix. See how the predicted fit %, time-to-hire, and cost-of-inaction shift. Every output cell carries the rationale that produced it.
AI · simulatorLevers untouched. Output equals the baseline Decision Trace for Parker M.. Pull a lever to surface the counterfactual delta.
PM
Baseline candidate
Parker M.
Customer Support Specialist · Customer Support · pre-hire
Levers
Inverse-keyword screening
Off (NODES default) · On reverts to legacy keyword filter
Persona version
Last (v−1) · Current · Shadow-validated next (v+1)
Interview latency
Fast cuts ~4d off TTH · Slow adds ~7d
Manager variance
Normalise out outlier reviewers · Amplify =>show full variance
Sourcing channel mix
Default · LinkedIn-heavy · Referral-heavy
Predicted outcome · baseline vs scenario
Predicted fit—
Baseline
64–89%
Scenario
64–89%
Time to hire—
Baseline
38–64 d
Scenario
38–64 d
Cost of inaction (qtr)—
Baseline
$18K–$42K
Scenario
$18K–$42K
Demo simulator · each lever applies a deterministic linear effect · in production this runs against the live model with no change to the controls you see here