Onboarding metrics that actually predict retention

Q: Should the manager see individual survey results?

Manager should see their own aggregate trend; HRBP should see individual results and decide what to escalate. The asymmetry preserves honesty.

Q: How long should the survey be?

Five questions max, under 3 minutes. Longer surveys produce lower response rates and acquiescence bias on the questions late in the list.

Q: What if response rates are low?

<60% response rate means new hires don't trust the loop. Fix the action loop first (visible intervention on their feedback), response rates follow.

60-Second Summary

Training completion correlates near-zero with 90-day retention. The metrics that predict retention are behavioral: time-to-first-contribution, manager 1:1 attendance rate, peer-network breadth, and self-reported clarity.
Bauer & SHRM Foundation research: new hires who report clarity on goals, role, and expectations at day 30 are 2.6x less likely to leave in year one.
The 45-day check-in is the highest-leverage survey moment — early enough to fix, late enough to have real signal.
Most regrettable 90-day exits are visible at day 30 if you look at the right signals. The companies that get blindsided are the ones measuring activity, not connection.

Most onboarding dashboards measure what's easy: did the new hire complete training, sign the handbook, attend orientation. None of those predict whether they'll still be here in 12 months.

What not to measure

Training completion percentage — easily gamed, predicts compliance, not engagement
Days-to-badge or days-to-laptop — measures IT, not ramp
NPS at day 7 — too early, captures novelty effect, regression-to-mean within 3 weeks
Number of 1:1s scheduled — calendar theatre; attendance and quality are what matters

The four behavioral metrics

What actually predicts retention

1
Time to first meaningful contribution
First PR merged, first customer call solo, first proposal written. Tracked by manager, not self-report. Median targets vary by role: 10 days for support, 25 days for engineering, 60 days for senior IC.
2
Manager 1:1 attendance rate (first 60 days)
Below 80% is a flashing red light — either manager isn't investing or hire isn't engaging. Either way, intervene at week 3.
3
Peer network breadth
Count of distinct colleagues the new hire has met 1:1. Strong predictor of belonging — below 8 by day 30 signals isolation risk.
4
Self-reported clarity
'I understand what I'm expected to deliver in the next 90 days' — 5-point scale. <4 at day 30 is the single strongest predictor of voluntary departure in year one.

The 45-day survey

Day 30 is too early for honest feedback (new hires haven't seen enough to know what's wrong). Day 90 is too late to retain the at-risk ones. Day 45 is the sweet spot. Five questions, anonymous-with-identifier (people analytics knows, manager doesn't), with a 24-hour follow-up loop for any score below 7.

Question	Score threshold
Likelihood to recommend this company as a place to work (0–10)	<7 = at risk
I understand what success looks like in my role (1–5)	<4 = mandate ambiguity
My manager is helping me succeed (1–5)	<4 = manager intervention
I have the relationships I need to do my job (1–5)	<4 = network gap
What's the single thing that would most improve your experience?	(qualitative — read every response)

What to do with the signal

A 45-day survey that goes into a dashboard nobody acts on is worse than no survey — the new hire shared honest signal and watched it disappear. Build the action loop first: any score below threshold triggers a manager 1:1 within 5 business days, with optional skip-level if the issue is the manager.

Frequently asked questions

Should the manager see individual survey results?

Manager should see their own aggregate trend; HRBP should see individual results and decide what to escalate. The asymmetry preserves honesty.

How long should the survey be?