Using people analytics to drive measurable business impact

Executive overview

Most HR teams collect data but never analyse it — they report activity instead of surfacing insight. People analytics shifts HR from gut-feel decisions to evidence-based ones, linking talent strategy directly to corporate goals.

Only 8% of organisations say they have usable data, yet the data usually already exists — it's just unrecognised or siloed. The four-stage analytics model (descriptive → diagnostic → predictive → prescriptive) gives HR leaders a practical ladder from raw facts to automated, personalised interventions.

The core insight: the talent plan must trace a clean line from global context → company purpose → goals → strategy → workforce needs — not start with a headcount request.

Aligning HR to business strategy

• HR leaders often start with headcount ("hire 10 engineers") rather than working back from strategy.
• The right sequence: global context → company purpose → goals → strategy → talent plan.
• A strategic workforce plan looks 5–7 years out: "We have 100 data scientists now; we'll need 600 in seven years."
• HR professionals should sit closer to the C-suite and treat talent as a direct input to corporate strategy.
• The goal is a "clean red thread" from world environment through to individual hiring decisions.

The four stages of analytics

Stage 1: Descriptive analytics

• Asks: what happened?
• Backwards-looking; gathers basic facts unemotionally.
• Most organisations do reporting and mistake it for analytics.
• Reporting lists activity (200 people attended, 4.5-star rating). Analytics maps data against other data to find meaning.
• Test: if you can't tell the story of what the data means or how it drives a decision, you have reporting, not analytics.

Stage 2: Diagnostic analytics

• Asks: why did it happen?
• Looks at causes behind past trends, not just the trends themselves.
• Requires triangulating multiple data points to answer "why."

Stage 3: Predictive analytics

• Asks: what will happen?
• Identifies consistent patterns among top performers, then applies them forward.
• Larger data sets increase prediction certainty; small sets still involve significant guesswork.
• Good enough accuracy is context-dependent: a retail display decision tolerates more error than a medical device decision.
• Test and calibrate algorithms as new data comes in; each new person either confirms or challenges the hypothesis.

Stage 4: Prescriptive analytics

• Asks: what should we do?
• Prescribes a course of action, potentially before the learner knows they have a gap.
• Machine learning can push personalised interventions to individuals in the moment.
• Only possible when the data set is large enough that prediction certainty is sufficiently high.

The reporting vs analytics distinction

• Reporting: 200 people, two-day class, 4.5 stars. This tells you almost nothing.
• Research shows level-one training evaluations correlate with catering quality, room temperature, and whether the instructor was funny — not learning outcomes.
• Real analytics: people who completed pre-work and post-work outperformed those who did neither — now you have something actionable.
• Ask of any data: "Tell me the story. How would I make a decision from this?"

Where to start when data is scarce

• Organisations almost always have more data than they think — it's unrecognised, not absent.
• Learning and development is typically among the highest expense lines on a Pareto chart; the waste from wrong content to wrong people at wrong times is significant.
• The "spray and pray" curriculum model (run everyone through the same content) ignores timing and individual gaps.
• Shift spend from classroom delivery toward analytics infrastructure; cost can stay flat or decrease.
• Every software platform in every function is already capturing data — pull it together before assuming you have none.

Building trust and overcoming resistance

• Avoid public blaming, shaming, or naming when introducing analytics.
• Start with external data: how is the industry changing? What skills will the sector need in 5–10 years?
• Ask the C-suite open questions: "Are our people today the people we'll need in seven years? How would we know?"
• Frame the project as collective and employee-focused, not as an audit of any department.
• One-on-one stakeholder conversations build safety before any data is shared publicly.
• Long-established leaders fear exposure; approach with palms open and humility, not a data scientist wielding a spreadsheet.
• Sell the collective vision: this is about doing right by the people of the company, not proving someone wrong.