flowchart LR A[Defined Metric<br/>headcount, attrition, compa-ratio] --> B[Trend and Slice<br/>over time, by unit] B --> C[Driver Explanation<br/>why the metric moves] C --> D[Prediction<br/>what is likely next] D --> E[Prescription<br/>recommended action] style A fill:#E8F0FE,stroke:#1A73E8 style B fill:#FEF7E0,stroke:#F9AB00 style C fill:#E6F4EA,stroke:#137333 style D fill:#FCE8E6,stroke:#C5221F style E fill:#F3E8FD,stroke:#8430CE
14 HR Metrics and HR Analytics
14.1 Why the Distinction Matters
A metric is what the firm measures. An analytic is what the firm asks of the measurement.
HR metrics and HR analytics are sometimes treated as the same discipline under different labels, and sometimes treated as warring functions competing for the same budget. Both readings are wrong. Metrics and analytics are two stages of the same intellectual journey. The first stage describes the workforce reliably. The second stage interrogates that description with questions, models, and counterfactuals. A function that runs both stages well is a function that has the audience’s trust at the description level and the audience’s attention at the interrogation level. A function that runs only one is a function whose dashboard tells half a story.
The maturation of e-HRM and digital workforce systems made the distinction sharper rather than blurrier. As Tanya V. Bondarouk & Huub J. M. Ruel (2009) argued in their early work on electronic human resource management, the digitisation of HR processes produced datasets large enough that interrogation became a distinct discipline rather than a side effect of reporting. The analyst who used to spend most of their week assembling a number now spends most of their week asking questions of the same number. The metric is the input; the analytic is the output of the question put to the input.
The distinction matters for a second reason: the audiences and decision rhythms differ. As Dave Ulrich & James H. Dulebohn (2015) documented in their review of where HR is headed, the line manager opens an HR-metrics page on a weekly cadence to run the team, while the executive committee opens an HR-analytics page on a quarterly cadence to make a strategic decision. The same number can serve both, but the chart, the framing, and the surrounding context have to be different. Mixing them into a single page produces a dashboard that satisfies neither audience.
The visualisation lens, as always, is what makes the distinction legible. A chart that reads as a metric — a count, a rate, a trend — looks one way. A chart that reads as an analytic — a model output, a comparison, a counterfactual — looks another way. The audience reads the function’s stage of maturity through the chart, not through the function’s slogan, and the analyst should design the chart to declare the stage honestly.
TipThe metrics-and-analytics contract
- Every dashboard makes the metrics-or-analytics framing of each chart explicit, so that the audience reads the chart at the level of confidence it has earned.
- The metrics layer feeds the analytics layer. An analytics chart that cannot be traced back to a defended metric is built on sand.
- Decisions are matched to the layer: line-manager and operational decisions are made on metric pages with weekly cadence; strategic and investment decisions are made on analytics pages with quarterly cadence.
14.2 HR Metrics in Detail
HR metrics, taken seriously, do four jobs. They count, they trend, they compare, and they categorise. Each job has its own data discipline, its own visual conventions, and its own failure modes. A metrics layer that does these four jobs reliably is the foundation on which the analytics layer can stand.
TipThe Four Jobs of HR Metrics
| Job | What the metric provides | Visual that serves it |
|---|---|---|
| Counting | A defended number for a defined population at a point in time | KPI card, headline number |
| Trending | The same number over time, with seasonality and comparison visible | Trend line with prior-year overlay |
| Comparing | The number against a benchmark, target, or peer group | Bar chart with reference line |
| Categorising | The number sliced by a workforce dimension that the audience cares about | Stacked bar, treemap, or heat map |
TipWhat the metrics layer earns
A solid metrics layer earns the audience’s trust at the description level. When the executive committee asks for headcount, attrition, or compa-ratio, the answer comes back the same regardless of which page they opened, on which day, or in which meeting. That consistency is what makes the analytics layer possible. A function that has not yet earned trust at the metrics level cannot ask the audience to take a predictive chart on faith.
14.3 HR Analytics in Detail
HR analytics, taken seriously, also does four jobs. It explains, it predicts, it prescribes, and it tests. Each is built on top of a reliable metrics layer, and each requires methods, framing, and visualisation choices that go beyond the metrics dashboard.
TipThe Four Jobs of HR Analytics
| Job | What the analytic provides | Visual that serves it |
|---|---|---|
| Explaining | A defended account of why a metric moves | Driver chart, regression-output visual |
| Predicting | A forecast of the metric or a per-employee risk | Risk distribution, forecast line with confidence band |
| Prescribing | A recommended action with cost and outcome | Scenario comparison page |
| Testing | A causal claim with a comparison group | Counterfactual chart, before-and-after with control |
TipWhat the analytics layer earns
A solid analytics layer earns the audience’s attention at the interrogation level. When the chief people officer wants to know whether a leadership programme moved revenue, whether attrition will spike next quarter, or which retention investment has the highest expected return, the answer is a chart that pairs an outcome with a defended comparison. The visual carries the cause-and-effect claim, not the speaker. The audience leaves the meeting with a decision rather than a discussion.
14.4 Where the Two Meet
Metrics and analytics meet at the working interface of the dashboard. The same underlying number is rendered one way for an operational audience and another way for a strategic one. The discipline is to design the interface deliberately, with the metric and the analytic visibly connected so that the audience can trace the journey from one to the other without leaving the page.
TipThe metric-to-analytic interface
Read the chain from left to right and the function climbs from describing the workforce to recommending action. Each step depends on the previous one. A prediction made without a defended trend is guessing dressed up as forecasting. A prescription made without a tested driver is opinion dressed up as analytics. The audience that sees the whole chain on one page is the audience most likely to act on the recommendation at the right end.
TipAudience and cadence at each step
Each step in the chain has a natural audience and cadence. The metric layer serves line managers and HR business partners at a daily-to-weekly cadence. The trend and slice layer serves the same audiences plus HR leadership at a weekly-to-monthly cadence. Driver explanations serve HR leadership and the executive committee at a monthly-to-quarterly cadence. Predictions and prescriptions serve the executive committee and the board at a quarterly cadence. Designing each page for its audience prevents the most common dashboard failure: serving the wrong layer to the wrong audience and watching attention drift away.
14.5 Visualising Both Together
A dashboard that holds both metrics and analytics on the same surface, without confusing them, is the most credible artefact an HR-analytics function can produce. Five design moves help the page do that work.
TipFive Design Moves That Hold Metrics and Analytics Together
| Move | What it does on the page |
|---|---|
| Layer label | Each chart names whether it is a metric or an analytic |
| Confidence rendering | Analytics charts render uncertainty visibly; metric charts do not need to |
| Comparison discipline | Every chart, metric or analytic, carries a benchmark, target, or control |
| Trace path | A click takes the curious user from analytic back to underlying metric |
| Cadence header | Each page declares the cadence its layer is designed for |
TipThe dashboard as a journey
Read end-to-end, the dashboard is a journey from a count to a recommendation. The journey starts with a metric the audience trusts and ends with an analytic the audience can act on. The function that designs the journey deliberately has built a dashboard that earns its place in the executive conversation. The function that ships only one end of the journey has built a dashboard that will be read once and forgotten.
Summary
| Concept | Description |
|---|---|
| Why the Distinction Matters | |
| Metrics and analytics as one journey | Metrics describe the workforce; analytics interrogate that description |
| Description first, interrogation second | An analytic without a defended metric underneath is built on sand |
| Audience and cadence differ | Line managers read metric pages weekly; executives read analytics pages quarterly |
| Chart declares the stage | The chart declares the function's stage of maturity to the audience |
| Trace from analytic to metric | A click takes the curious user from analytic back to underlying metric |
| HR Metrics in Detail | |
| Counting job | A defended number for a defined population at a point in time |
| Trending job | The same number over time with seasonality and comparison visible |
| Comparing job | The number against benchmark, target, or peer group |
| Categorising job | The number sliced by a workforce dimension the audience cares about |
| Trust at the description level | When the metrics layer is consistent, the analytics layer can ask for trust |
| HR Analytics in Detail | |
| Explaining job | A defended account of why a metric moves |
| Predicting job | A forecast of the metric or a per-employee risk |
| Prescribing job | A recommended action with cost and outcome |
| Testing job | A causal claim with a comparison group |
| Attention at the interrogation level | When the analytics layer is credible, decisions follow |
| Where the Two Meet | |
| Defined metric | Headcount, attrition, compa-ratio rendered as a count or rate |
| Trend and slice layer | The metric over time and across business dimensions |
| Driver explanation | Why the metric is moving, with regression or driver visuals |
| Prediction with confidence | Forecast of the metric with confidence band |
| Prescription with cost and outcome | Recommended action paired with cost and outcome on a scenario page |
| Operational cadence | Daily-to-weekly cadence for line managers and HR business partners |
| Strategic cadence | Quarterly cadence for the executive committee and board |
| Audience-cadence match | Each page is designed for its audience and cadence rather than reused across both |
| Visualising Both Together | |
| Layer label | Each chart names whether it is a metric or an analytic |
| Confidence rendering | Analytics charts render uncertainty visibly; metric charts do not need to |
| Comparison discipline | Every chart carries a benchmark, target, or control |
| Trace path | Drill from analytic chart back to the underlying metric and data |
| Cadence header | Each page declares the cadence its layer is designed for |
| Dashboard as a journey | Read end-to-end, the dashboard runs from count to recommendation |
| End-to-end design | Functions that design the journey deliberately earn the executive conversation |