HR metrics and KPIs every people team should track

Summary:

HR metrics are the quantified measures HR teams use to track hiring, retention, capability, cost and employee experience. Producing them is rarely the hard part; choosing the ones that change a decision is. This guide covers which HR metrics to track, where each one comes from, and how to set targets you can defend.

HR metrics are quantified measures of how a people function performs and how a workforce behaves. They span recruiting, retention, performance, capability, cost and employee experience, and they exist to make people decisions evidence based rather than instinctive.

Most HR teams already produce metrics. Far fewer produce metrics anyone acts on. The difference usually comes down to whether the number was chosen before or after somebody asked a question.

An HR metric becomes an HR KPI the moment you attach a target to it and agree that it reflects a business goal. Turnover rate is a metric. Reducing voluntary turnover in field operations from 34% to 25% by the fourth quarter is a KPI.

Every KPI is a metric, but most metrics are not KPIs. Treating them as interchangeable is how HR dashboards end up carrying forty numbers and signaling no priorities.

A workable rule of thumb: track as many metrics as your systems produce cheaply, and elevate only a handful to KPI status in any given year.

Lagging indicators tell you what already happened. Turnover, cost per hire and absenteeism are all lagging, and by the time they move, the decision that caused the movement is months old.

Leading indicators move before the outcome does. Engagement scores, manager support ratings, intent to stay responses and internal application rates all shift ahead of turnover, which is why they belong on the same dashboard as the lagging numbers they predict.

Most HR reporting is almost entirely lagging. Pairing each lagging metric with at least one leading indicator is what turns a monthly report into an early warning system.

HR measurement matters because it is the mechanism by which people programs get funded. Leadership rarely defunds an HR program because it dislikes the idea. It defunds the program because nobody could show what changed.

The scale of that gap is documented. Gartner surveyed 1,403 HR leaders in July 2024 and found that 66% of nearly 475 respondents said their workforce planning is limited to headcount planning, and that they struggle to demonstrate return on investment for strategic workforce planning efforts.

That is a measurement problem before it is a strategy problem. Headcount is the easiest number to produce and among the least useful for arguing that a program worked. Our research on the trends and opportunities facing HR teams points the same direction.

Executive questions about people are narrower than they first appear. In practice they reduce to four:

A metric earns its place on your dashboard by answering one of those four questions. If it maps to none of them, it is interesting rather than important.

HR metrics fall into five categories, each answering a different question about the workforce. Organizing your reporting this way prevents the common failure of tracking six recruiting metrics and nothing at all about capability.

Recruiting metrics answer whether you can fill roles at acceptable speed, cost and quality. Time to fill, time to hire, cost per hire, offer acceptance rate, source of hire and quality of hire make up the standard set.

Time to fill and time to hire are not the same measure, and they are routinely confused. Time to fill counts from requisition approval to offer acceptance, while time to hire counts only from the point a candidate enters the process.

Quality of hire is the one recruiting metric no applicant tracking system can produce on its own. It depends on hiring manager and new hire feedback gathered after the fact, which makes it a survey measure rather than a system measure.

Retention metrics answer whether you are keeping the people you want to keep. The core set is turnover rate, retention rate, the voluntary and involuntary split, first year turnover and regretted attrition.

Segmentation matters more than the headline figure here. A stable company-wide turnover rate can conceal a single department losing a third of its team, which is why turnover is worth cutting by function, tenure band and manager. Our guide to employee retention strategies covers the formula and the levers that move it.

Regretted attrition, the share of leavers you wanted to keep, is the number most likely to change an executive conversation. It is also the one fewest teams actually track.

Engagement metrics answer how people feel about working at your organization and whether that is trending the right way. Employee Net Promoter Score, employee satisfaction, manager support ratings, intent to stay, participation rate and psychological safety measures sit in this category. Each one is a self reported measure, which is what separates this category from the four around it.

These are the leading indicators in the set. They move before turnover does, which makes them the early warning half of any HR dashboard, and it is why employee feedback programs tend to be where HR measurement starts.

They also share one characteristic that shapes everything in the next section. None of them exists until you ask employees a question. For the full measurement approach, see our guide on how to measure employee engagement.

Capability metrics answer whether your workforce can do the work that is coming. Internal mobility rate, promotion rate, training completion, skills gap coverage, time to proficiency and performance distribution are the usual measures.

Training completion is the weakest metric in this category and among the most reported. It records attendance rather than capability, and it needs pairing with a post training knowledge or application measure before it means anything.

Internal mobility rate deserves elevation. When it falls, employees start looking outside for their next role, and voluntary turnover tends to follow.

Cost metrics answer what the people function and the workforce cost relative to output. HR cost per employee, HR to employee ratio, revenue per employee, absenteeism rate, overtime spend and total workforce cost as a share of revenue belong here.

These are the metrics finance already tracks, often using different definitions than HR uses. Agreeing the definition with finance before you report saves an awkward meeting later.

Revenue per employee is the most commonly misused of the group. It is a business performance number that HR influences rather than owns, so present it as context, not as an HR result.

Every HR metric comes from one of a small number of systems, and knowing which one determines whether you can actually produce it. This is the step most metric lists skip, and it is why teams commit to dashboards they cannot populate.

The same Gartner survey found that 55% of HR leaders think their current technology solutions do not cover current and future business needs, and 46% believe current HR technology hinders rather than improves the employee experience. Knowing what each system can and cannot produce is how you avoid buying a dashboard for data you do not have.

Your transactional systems produce every metric that is fundamentally a count, a date or a dollar. That covers most recruiting, retention, cost and completion measures:

What none of them can tell you is why a number moved. A system of record captures the transaction, not the reasoning behind it, so a rising turnover rate arrives without a cause attached.

Engagement, satisfaction, manager support, intent to stay, psychological safety, onboarding effectiveness, candidate experience and training applicability have no system of record. They exist only if you ask, which makes the employee survey the system of record for the entire leading indicator half of your dashboard.

That has a practical consequence for measurement design. Because you are generating this data rather than extracting it, question wording, cadence, anonymity and response rate all become measurement decisions rather than administrative ones.

Response quality is the risk to manage. SurveyMonkey research among 269 HR professionals found that 72% express concerns over employees not providing open and honest feedback about their experiences at work, and skewed responses produce a leading indicator that points the wrong way. Getting the questions right matters as much as asking them, which is why our library of HR survey questions exists.

YES Communities shows what the survey half of the dashboard can do when it works. Facing a 60% employee turnover rate, the company used SurveyMonkey Enterprise to run automated feedback across the employee lifecycle, reached a 93% survey completion rate, and used exit survey findings to correct misalignments in hiring expectations and expand bereavement and parental leave options. Average employee retention rose from about 90 days to a year and a half, a roughly 500% increase, as described in the YES Communities customer story.

Choose metrics by starting from a decision you need to make, not from a published list. If no decision changes based on the number, the number is reporting overhead.

Work through the selection in a fixed order, and the shortlist builds itself:

Working in that order means you never commit to a metric you cannot source, which is the failure the previous section describes. It also keeps the list short, because most decisions need three or four numbers rather than thirty.

A target without a baseline is a guess. Measure the metric unchanged for at least two full cycles before setting a target, so you know its normal range and its seasonal shape.

Then express the target as movement from your own baseline rather than as an absolute figure. Reducing first year turnover from 28% to 22% is defensible, whereas committing to an industry average is not, because you do not control the comparison set.

Set the review cadence at the same time as the target. Lagging metrics rarely need review more than quarterly, while leading indicators can move monthly, and reviewing those quarterly wastes the early warning you built them for.

There is no universal good value for most HR metrics. Turnover that would alarm a professional services firm is unremarkable in hospitality, and an HR to employee ratio that looks thin at 200 employees looks generous at 5,000.

Use external benchmarks for orientation and your own trend for judgment. Published national figures, such as the Bureau of Labor Statistics data referenced in our guide to calculating employee turnover rate, are useful mainly for establishing whether your figure is unusual at all.

Your internal trend is the more reliable signal. Comparing segments inside your own organization, department against department and manager against manager, controls for industry and company size automatically in a way external benchmarking never can.

Individual metrics need more depth than a category overview can carry. These guides and templates go deeper on the measures most HR teams report first.

Four measures account for most of what HR teams report first, and each has its own formula and interpretation rules:

The leading indicator half of your dashboard needs a collection method, and these three cover the essentials:

Most HR measurement programs fail for reasons that have little to do with which metrics were chosen. Four patterns account for the majority:

Small sample reporting deserves particular care. Breaking an engagement score down to a five person team produces a figure that swings on one person's mood, and it risks identifying individuals, which is both poor measurement and a breach of the anonymity you promised.

Agree a minimum reporting threshold with your team before the first survey goes out, and hold to it even when a manager asks for their own numbers.

The fourth pattern is the most damaging because it compounds. Each unactioned survey lowers participation in the next one, and once participation falls far enough, the leading indicators stop being reliable at exactly the moment you need them.

Lead with the decision you want, not with the dashboard. Executives need the number, its direction, the likely cause and your recommendation, in that order.

Three or four metrics per meeting, each with a named owner and a stated action, will land better than a twenty metric appendix. The appendix can still exist. It just should not lead.

Pair every lagging metric you present with the leading indicator you are watching alongside it. That single habit converts the conversation from explaining last quarter into managing next quarter, and it is the same discipline behind a well run employee engagement program.

The metrics that prove HR's impact are split across two very different sources. Your transactional systems hand you the lagging half, and the leading half, the part that tells you what is about to happen, exists only if you ask your employees directly.

That second half is what SurveyMonkey is for. HR teams use it to run continuous employee listening across the candidate and employee experience, gather candid feedback with anonymity controls, and turn open text comments into the causes behind a moving number.

woom, a global manufacturer of children's bicycles, used SurveyMonkey Enterprise to centralize employee and customer experience data across a 30 country distribution network. Its initial global engagement survey achieved a 78% response rate and an eNPS of 46, and its targeted learning surveys produced an average 86% learner satisfaction score, as set out in the Woom customer story.

NPS, Net Promoter & Net Promoter Score are registered trademarks of Satmetrix Systems, Inc., Bain & Company and Fred Reichheld.