Back online
Metric Standard #46

Near-Misses Were Counted. Their Acceleration Was Not.

Three near-misses last month. Three this month. Same count — completely different danger. Last month they were spread across 4 weeks. This month, all 3 happened in the last 6 days. The velocity was screaming. The old standard could not hear it. POD created NearMissVelocity and LeadingIndicatorDashboard to measure acceleration, not just frequency.

+0%
Velocity Alert Threshold
0%
Incident Probability When Critical
0
Leading Indicators Monitored
24/7
Near-Miss Velocity Tracking

The Week the Count Said "Normal" — Velocity Said "Critical"

The incident was not a surprise. The velocity was clear. The standard to see it did not exist.

Weeks 1–3

1 near-miss reported per week — spread evenly, rate steady

Safety director reviews monthly summary. Count = 3. Status: acceptable.

Week 4, Day 1

1st near-miss of the week — overhead load swing near a worker

Logged. No velocity calculation. No alert. Pattern not yet visible.

Week 4, Days 3–4

2nd and 3rd near-misses in 4 days

Monthly count still shows 6 total. The acceleration is invisible in a count.

Week 4, Day 5

Recordable incident — same crew, same zone, same hazard type

The warning signs were screaming. The velocity was +300%. Nobody had a metric that heard it.

How POD Created the Near-Miss Standard

From raw count to velocity signal to compound leading indicator — in one platform.

01

Near-miss reports captured in real time via voice

Your superintendent speaks a field safety report. Every near-miss mentioned is captured, timestamped, classified by hazard type, and added to the velocity calculation instantly.

02

AI calculates velocity and detects acceleration

Specialized AI agents calculate the derivative of near-miss frequency — the rate of change, not just the count. When velocity crosses the acceleration threshold, an alert fires before the pattern becomes an incident.

03

LeadingIndicatorDashboard shows the full precondition picture

Near-miss velocity is contextualized against fatigue index, schedule pressure, new worker ratio, and weather risk. The compound precondition profile reveals whether velocity is a signal or noise.

The Acceleration Nobody Could See

Weeks 1–3: flat velocity. Week 4: slope climbs steeply — +300% acceleration. Incident probability hits 61%. POD alert triggers. Intervention reverses the curve.

012345Wk 1Wk 2Wk 3Wk 4Wk 5Near-Misses / Week11124Near-Miss Frequency — Velocity Reveals the Hidden Risk
Live KPI Preview

The Near-Miss Standard — Velocity Tracked, Incident Predicted

NearMissVelocity measures the acceleration. LeadingIndicatorDashboard shows the full precondition context. Together they created the prediction engine that counts never could.

Near Miss Velocity
POD
0reports / period
0%
vs previous
Rolling Avg0.0
Alert Threshold40
Weekly Trend
Count
Rolling Avg
0510152025300Wk 10Wk 20Wk 30Wk 40Wk 5
Total0reports
Peak0Wk 5
Low0Wk 1

Leading Indicators

POD
2 alerts
Near Miss Rate00.0%
Crew Fatigue0+0.0%
Schedule Pressure0+0.0%
Training Currency0+0.0%
Equipment Uptime00.0%
Green0
Amber0
Red0
Worst
0.0%Near Miss Rate
1 red and 1 amber indicators — Schedule Pressure needs immediate attention

The Platform Behind the Standard

Velocity Not Count

NearMissVelocity measures the slope of near-miss frequency — the rate of change week over week. Same count, different pattern: evenly spread vs. clustered in 3 days are completely different risk signals.

Leading Indicator Dashboard

Five preconditions tracked simultaneously: near-miss velocity, fatigue index, schedule pressure, new worker ratio, and weather risk. The compound signal is always stronger than any single indicator.

Voice Capture in 30 Seconds

A near-miss mentioned in the daily voice report becomes a velocity data point in under 30 seconds. Reporting friction is a safety system — reducing it increases near-miss capture rate.

Incident Probability Estimation

AI agents estimate incident probability from the compound precondition profile: velocity + fatigue + schedule pressure. The probability updates continuously and triggers intervention recommendations.

Hazard Type Clustering

When multiple near-misses cluster around the same hazard type — overhead loads, excavation edges, electrical — POD identifies the pattern and flags the specific hazard for targeted intervention.

Safety Prediction vs. Reaction

Traditional safety tools count incidents after they happen. POD measures the leading indicators — near-miss velocity, fatigue levels, schedule pressure — that predict incidents before they occur.

“We had three near-misses in one week and our safety manager would not have flagged it as unusual — same count as every other week. POD showed us the velocity had gone from near zero to +300%. We stopped work on that zone before the incident happened.”

— Safety Director, Industrial GC

Frequently Asked Questions

Measure What Predicts — Not Just What Already Happened

See NearMissVelocity and LeadingIndicatorDashboard on your site — before the velocity goes critical.

Last updated: April 2026