A strong safety analyst balances lagging outcomes with leading information about the health of the control system.

Leading versus lagging indicators

Lagging

Examples:

  • recordable incidents;
  • lost-time events;
  • injuries;
  • process-safety events.

They describe outcomes that already occurred.

Leading

Examples:

  • overdue critical actions;
  • audit completion;
  • control verification;
  • training/competence;
  • inspection findings;
  • safeguard impairment.

They may provide earlier signals, but only if they connect to meaningful risk.

Incident and near-miss data

Analyze:

  • event type;
  • task;
  • location;
  • time;
  • severity;
  • potential severity;
  • recurring factors.

Do not assume more near-miss reports mean worse safety. Reporting culture changes the count.

Audit and action data

Useful measures:

  • open actions;
  • overdue high-risk actions;
  • repeat findings;
  • closure quality;
  • aging.

Closing low-value actions quickly can make a dashboard look good while serious issues remain.

Exposure and activity context

Counts need context.

Examples:

  • hours worked;
  • number of lifts;
  • maintenance tasks;
  • production volume;
  • contractor exposure.

Normalization should match the risk question.

Rates and normalization

Rates can improve comparison, but they can mislead when:

  • populations are small;
  • exposure differs;
  • definitions change;
  • reporting quality differs.

Explain denominator and scope.

Dashboard design

A useful dashboard should answer:

  • Where is risk concentrated?
  • Which actions are overdue?
  • Which controls are degrading?
  • What changed?
  • Who needs to act?

Avoid visual clutter.

Data quality

Check:

  • duplicate incidents;
  • inconsistent categories;
  • missing fields;
  • late reporting;
  • denominator accuracy;
  • changing definitions.

Perverse incentives

If managers are rewarded only for low incident counts, people may feel pressure not to report.

Balance outcome measures with:

  • learning;
  • control verification;
  • action quality;
  • worker participation.

Portfolio project

Use synthetic data to build a dashboard combining:

  • incidents;
  • near misses;
  • actions;
  • audits;
  • exposure.

Include a short note on what the metrics cannot prove.

Good safety analytics makes uncertainty visible instead of hiding it behind a green KPI.

Related content

Sources