What Is HR Analytics?

HR analytics is the practice of collecting, cleaning, and analysing workforce data to answer people-related business questions with evidence rather than intuition. Done well it tells a CHRO which hiring channels actually produce first-year retainers, which manager is losing people faster than the org average, and whether a $200,000 wellness spend moved any measurable dial. Done badly it is a monthly PDF of pie charts no one reads. This guide covers what HR analytics is, the four-level framework most teams use, the metrics that pay for themselves, the systems that generate the data, and where AI is starting to change what a small HR team can actually analyse.

Fabric builds the AI interview platform that runs Round 1 of hiring for engineering, sales, and other objective-evaluation roles, so we see the top-of-funnel side of this every day: recruiters who want to know which sources produce candidates who pass screening, which job descriptions attract the wrong applicants, and how the numbers actually compare between last quarter and this one. Those are all HR analytics questions in a different hat.

HR analytics defined

HR analytics, sometimes called people analytics or workforce analytics, is the systematic use of workforce data to inform HR and business decisions. The data comes from the HRIS, the ATS, the payroll system, engagement surveys, performance reviews, and increasingly from productivity tools like calendar and collaboration platforms. The output is a set of numbers and narratives that let managers and leadership see what is happening to their workforce, why, and what to do about it.

The discipline has grown up alongside the rise of the enterprise HRIS over the last two decades. What used to be an annual headcount audit is now a live view that, in the best-run companies, sits next to sales pipeline and finance dashboards in the same weekly leadership review.

The single most important thing to say about HR analytics is that it lives or dies on data quality. If your HRIS has three different spellings of the same department name, no analytics tool on earth will give you a clean attrition-by-team number. Cleaning the data comes first, always.

The four-level HR analytics framework

The framework almost every people analytics team uses (drawn from the wider analytics literature and popularised in HR by Josh Bersin and others) has four levels. Each answers a different question.

Descriptive analytics answers "what happened?" This is your quarterly headcount, your attrition rate, your open-to-fill days, your gender pay-gap number. It is the reporting layer, and it is where most HR analytics work still sits.

Diagnostic analytics answers "why?" You know attrition jumped to 22 percent. Diagnostic work is the segmentation that says: 34 percent of that came from one region, that region has three teams under one director, and exit interview text mentions the phrase "no clear career path" 11 times.

Predictive analytics answers "what is likely to happen?" Given the patterns in your historical data, which employees are at high flight risk over the next 90 days, and how confident are you in that prediction? Predictive HR models are useful, and they are also easy to over-trust; treat every model score as one signal among several, not a verdict.

Prescriptive analytics answers "what should we do about it?" This is the layer where analytics starts to become recommendation. Most teams do this in narrative form rather than as a system output: an analyst reads the diagnostic and predictive results and writes a one-page memo for the leadership team.

Most HR functions live at levels 1 and 2. Reaching for level 3 or 4 is worth it only on a handful of high-stakes questions, because the modelling and validation cost is real.

The HR metrics that actually matter

There are hundreds of HR metrics you could track. The list that pays for itself, for most companies, is short.

• Time-to-hire and time-to-fill. Measured in days from requisition open to offer accepted (fill) or from application to offer (hire). Both matter, and they measure different things.
• Cost-per-hire. All-in cost of hiring one person, including sourcing spend, agency fees, recruiter salary allocation, and interviewer time. Almost always higher than teams first estimate.
• Quality-of-hire. Some blend of 90-day performance rating, 12-month retention, and hiring-manager satisfaction. The definition matters more than the specific formula.
• Regretted attrition rate. Percentage of leavers you wish had stayed. This is a much sharper signal than raw attrition, because it strips out people you were going to let go anyway.
• First-year attrition. Percentage of new hires who leave in their first 12 months. High numbers here point at either bad hiring or bad onboarding, and you have to diagnose which.
• Internal mobility rate. Percentage of openings filled by internal candidates. A high number is usually good for retention and cost, and it is the metric that shows whether "we develop from within" is real or aspirational.
• Absenteeism and overtime. Blunt but useful. Sustained overtime combined with rising absenteeism is a burnout leading indicator.
• Engagement score, with response rate. Never publish the score without the response rate; the two together tell you whether the number means anything.

Pick five of these, define each one in writing, and report them consistently every quarter. That is worth more than an eighty-metric dashboard that nobody uses.

A working human resource analytics framework

If you are building an HR analytics practice from scratch, a workable framework has four moving parts, not four levels.

1. A data layer. One canonical source for each fact. Employee data lives in the HRIS. Applicant data lives in the ATS. Payroll data lives in payroll. If two systems store the same field, you decide once which one is authoritative.
2. A definitions layer. A short written document that defines every metric you report, in one sentence. What is "attrition"? Is a contract-to-hire conversion counted as a hire? What is a "regretted" leaver? This document prevents the meeting where three people compare three different numbers.
3. A cadence. Weekly operational reporting for the recruiting and HR ops teams. Monthly management view for functional leaders. Quarterly deep-dive for the CHRO and CEO. Skip the ad-hoc report if the same question shows up more than twice: add it to the cadence.
4. A narrative layer. Every report is accompanied by two or three sentences of analyst commentary that says what changed and why. Dashboards without narrative are furniture.

That is the minimum. Everything else is enrichment.

Where AI changes the analytics story

The two AI shifts that matter for HR analytics right now are on the intake side and on the narrative side.

On the intake side, AI-led screening and interviewing produce far more structured data about candidates than a manual process ever did. Fabric runs Round 1 interviews (resume screening, eligibility checks, and a conversational AI interview with role-specific formats like pair programming for engineers and cold-call simulation for sales) and every step of that produces analysable data. Recruiters can then answer questions like "which source produces candidates who actually pass the technical section" instead of only "which source produces the most applications." Cheating detection is built into Round 1 as a core part of the product, not an add-on, which means the accept rate is not being inflated by candidates who cleared the interview using an AI assistant.

*Fabric's eligibility and technical-depth scoring are designed to flag mismatches and surface them to your recruiter. It's a signal for your team to weigh, not an automatic reject.*

On the narrative side, generative AI is useful for turning dashboards into first-draft analyst commentary. It is not useful as a replacement for the analyst; the model does not know which of your three attrition definitions is the one to use this quarter. Treat it as a drafting tool for the human who signs off.

Getting started, week by week

You do not need a data science team to start. A realistic first-90-days plan for a small HR function looks like this.

Weeks 1 to 2, pick five metrics and define each one in a shared document. Weeks 3 to 4, clean the underlying data (department names, level codes, manager IDs) so the metrics can be computed cleanly. Weeks 5 to 8, build the first version of the report in whatever tool you already have — the HRIS reporting module is fine to start. Weeks 9 to 12, present the report to your leadership team, capture the questions they ask, and iterate.

If after 90 days you find yourself answering the same five questions from leadership each week, the reporting is doing its job. If nobody asks anything, the metrics are wrong, not the tool.

Related posts

• Capacity planning and workforce analytics
• Workforce planning fundamentals
• Attrition vs. retention rate: which one to track
• Human capital management explained
• HR technology systems: an overview

FAQ

What is HR analytics in simple terms?

HR analytics is the practice of collecting, cleaning, and analysing workforce data to answer people-related business questions, such as why attrition is spiking or whether a hiring channel actually pays off. It turns HR gut-feel into evidence a CFO or CHRO will accept.

What is the difference between HR analytics and people analytics?

The terms are used interchangeably in most companies. Where a distinction is drawn, HR analytics tends to focus on HR-owned processes (hiring, comp, attrition), while people analytics is a broader label that includes engagement, performance, and network data across the business.

What are the four levels of HR analytics?

The common framework is descriptive (what happened), diagnostic (why), predictive (what is likely to happen), and prescriptive (what to do). Most HR teams live in the first two and reach for the last two on a handful of high-stakes questions.

What are the most important HR metrics to track?

Time-to-hire, cost-per-hire, quality-of-hire, first-year attrition, regretted attrition, absenteeism, and internal mobility rate cover most of what a CHRO is asked about. Each should be defined once in writing so the same number means the same thing across quarters.

What tools are used for HR analytics?

Most teams start with the reporting layer inside their HRIS (Workday, BambooHR, SAP SuccessFactors), then add a spreadsheet or BI tool (Excel, Google Sheets, Power BI, Tableau) for anything the HRIS cannot show natively. Dedicated people analytics platforms like Visier come in once the volume justifies them.

Is HR analytics the same as data science?

No. HR analytics uses statistical techniques, but the job is to answer HR questions clearly, not to build models for their own sake. A good HR analyst is a domain expert who knows enough SQL and stats to be useful, not a data scientist who happens to work in HR.