Onboarding Metrics: The 7 KPIs That Actually Matter (2026)

Onboarding metrics answer the question most companies only feel: is our onboarding actually working? Seven numbers cover it, three that measure the outcome and four that predict it early enough to fix. This guide defines each, shows how to measure them without a data team, and explains the one metric almost nobody tracks that predicts ramp better than the rest.

Why measure onboarding at all?

Because unmeasured onboarding is always reported as fine and frequently is not. Gallup finds only about 12% of employees strongly agree their company onboards well, which means a large majority of companies are running weak onboarding while believing otherwise. Metrics replace the belief with a number, and the number tells you whether the fixes in employee onboarding best practices are landing. The financial stake, every point of early turnover paid in replacement costs, is laid out in how much it costs to onboard an employee.

The seven metrics

Time to productivity is the headline. Define "productive" per role (a closed deal, an owned feature, independent caseload), measure when each hire reaches it, and track the trend by cohort. Absolute benchmarks across companies are mostly noise; your own trend is signal.

Retention at 90 days and one year is the outcome the whole exercise serves. O.C. Tanner has reported that a large share of turnover lands in the first 45 days, so the 90-day number is where weak onboarding shows first. The full causal chain is in reducing new hire turnover.

Completion rate is the earliest warning. If hires are not finishing the flow, nothing downstream will look good, and the dropped steps tell you exactly where the flow is broken.

Satisfaction comes from the short milestone surveys covered in onboarding survey questions: under five questions, in Slack, at days 7, 30, and 90.

Time to first contribution measures momentum: the first shipped anything. Day-one or week-one first contributions correlate with confident ramps, which is why "ship something tiny on day one" appears in every good onboarding design.

The metric nobody tracks: question volume

A new hire's questions are a live feed of their ramp. Early on, high volume is healthy, it means they are engaged and unblocked enough to ask. What matters is the trend: volume should fall steadily as knowledge sticks, and the topics should mature from "where is the VPN guide" to role-specific depth. A hire still asking basic questions at week six has a gap your onboarding did not close, visible weeks before it shows up in any lagging metric. Almost nobody tracks this, because almost nobody's questions flow through anything trackable; they vanish into DMs and shoulder taps.

Leading vs lagging: where to spend attention

Lagging metrics (retention, time to productivity) are the scoreboard; you cannot act on them, only learn from them. Leading metrics (completion, satisfaction, questions, first contribution) are the steering wheel: they predict the scoreboard while there is still time to intervene with a specific hire or fix a specific step. The practical rule: report the lagging ones monthly, act on the leading ones weekly.

How Sakha gives you the metrics for free

Most of these metrics are painful to collect by hand, which is why they go untracked. Sakha generates them as a byproduct of running the onboarding: completion rates per hire and per step, satisfaction from the built-in milestone surveys, and, uniquely, the question feed, because new hire questions flow through Sakha's knowledge base instead of vanishing into DMs. You see question volume trends per hire, the topics that keep recurring, and the knowledge gaps the questions reveal, the leading indicators that exist nowhere else.

The result is an onboarding you can steer instead of hope about: the dashboard shows which step drops people, which cohort ramped fastest, and which hire needs attention this week. Measurement built into the machine, not bolted on after.