What Makes a Hiring Decision Fragile
A fragile hiring decision is one that would flip to the opposite outcome if any single factor changed: one interviewer re-evaluated their score, a reference surfaced new information, or the job spec was re-read more carefully.
Fragility is not about how 'close' a candidate scored. A candidate at 62% on a 60% threshold might be robust if all five interviewers converged on strong evidence. A candidate at 76% might be deeply fragile if three interviewers scored 4/4 without specific behavioral evidence.
Key Volatility Indicators
Score spread across interviewers: If two interviewers score Technical Depth 1/4 and two score it 4/4, the aggregate obscures a fundamental disagreement — they were likely evaluating different things.
Evidence density: An interviewer who scores Communication 4/4 with 'great speaker' as justification has provided low-evidence endorsement. An interviewer who scores 2/4 with three specific behavioral examples has given high-evidence concern.
Late-stage pivots: Did any interviewer change their score after the debrief? Pivots toward the most senior person in the room indicate social influence, not evidence updating.
The Cost of Getting It Wrong
A mis-hire at the senior individual contributor level typically costs 1.5–2x annual salary when factoring recruiting, onboarding, performance management, and productivity loss. At the management level, downstream team attrition and delayed projects compound this significantly.
The question is not 'should we analyze this decision or not' — it is 'is the cost of a 20-minute structured risk review less than the expected cost of a fragile decision materializing?' The answer is almost always yes.
Decision Stability Modeling
Formal decision stability modeling answers: if we introduced plausible variation into the inputs — score recalibration, weight adjustment, one interviewer replaced — how often does the current outcome hold?
A simple version: identify the three inputs most sensitive to reasonable re-evaluation, stress-test each against the decision gate, and record the results. If the decision flips under two of three stress scenarios, you have a fragile decision.
Vetriva automates this process, running probabilistic modeling across score distributions and reporting a Stability Score (0–100) and Flip Probability for every candidate reaching a decision gate.
When to Hold vs. Proceed
Proceed when: scores are based on behavioral evidence, dimension-level agreement is high, and stability analysis shows the decision is robust to plausible re-evaluation.
Hold when: score spread across interviewers is high, one interviewer has a strong contrary signal that wasn't discussed in the debrief, or core-dimension evidence is thin.
Decline quickly when the decision is fragile in a candidate-unfavorable direction. Prolonged processes serve no one. Risk analysis is not designed to delay good decisions — it is designed to prevent bad ones from moving forward on momentum.