Don't accept the onboarding hypothesis. Segment first, conclude second.

Segment by:

Hiring manager. Is 15% distributed evenly or concentrated under 3 managers? If the latter, the issue is hiring judgment or management style, not corporate onboarding.
Role and function. Ops roles often have higher early attrition than knowledge work for different reasons (schedule mismatch vs. culture fit).
Source. Agency hires vs. referrals vs. job-board applicants have different 90-day survival. A shift toward higher-volume, lower-quality sources can move the metric without any onboarding change.
Stated reason. Exit survey categories: compensation, manager, role mismatch, commute, family, another offer. Role mismatch and manager dominate early attrition.
Voluntary vs. involuntary. 15% blended can be mostly performance-based terminations masked as 'quit' — different problem.

Correlate, carefully:

90-day attrition vs. time-to-first-offer-accepted (panic hires are worse hires).
90-day attrition vs. onboarding completion rate (if actually causal, there should be a gradient; usually there isn't).

What to deliver:

Attrition decomposed by source, manager, function.
Top 3 stated exit reasons weighted by volume.
Named hypothesis: e.g., 'the 15% is driven by 40% attrition in two Support team managers; onboarding content is fine.'

Red flag: running a full onboarding redesign before segmenting — that's months of investment aimed at a problem that might be a single manager.

15% of new hires quit within 90 days. Diagnose.

Stuck? Peek at a hint