KEY TAKEAWAYS
- Fleet safety KPIs fall into two groups: lagging indicators that count what already happened (crashes, injuries, claims) and leading indicators that show risk building (harsh events, speeding, near-misses, coaching completion).
- Leading indicators are where performance gaps can still be closed, because they change weeks or months before a crash does.
- FMCSA’s Crash Preventability Determination Program reviews 21 crash types, including any crash where video shows the sequence of events. As of June 2026, FMCSA had found 746 of the 818 crashes it reviewed under that video crash type Not Preventable.
- As of June 2026, FMCSA had found about 96% of the crashes it reviewed under the program Not Preventable, and those crashes are removed from a carrier’s Crash Indicator calculation in the Safety Measurement System.
- A commercial vehicle injury crash carries an average comprehensive cost of about $331,000 (2023 dollars), according to FMCSA, which is why even small KPI improvements matter.
- Intangles brings the data behind these KPIs into one platform: DriveIQ flags 20+ behavioral exceptions per driver, predictive health monitoring covers vehicle condition, and AI-driven video telematics adds context to events.
Two fleets each report 12 crashes last year. One had 12 drivers involved in one crash each. The other had three drivers involved in four crashes each. On a monthly report, they look the same. In practice, they have completely different safety problems.
That’s the trouble with measuring safety only by crash counts, and why fleet safety KPIs need to go further. By the time a crash shows up in a report, the opportunity to prevent it has passed, and the number alone doesn’t say where the risk came from. The stakes are high: FMCSA’s Crash Cost Methodology 2025 puts the comprehensive cost of a commercial vehicle injury crash at about $331,000 (2023 dollars).
Intangles close that gap by turning every trip into driver- and vehicle-level data, so safety can be measured while risk is building, not only after a crash.This blog covers the difference between leading and lagging fleet safety KPIs, the 12 KPIs that matter most, how to separate preventable from non-preventable crashes, why near-misses are worth tracking, and how to set targets that close the gap.
Why fleet safety performance stalls
Many fleets don’t lack safety data. Performance stalls because the data is measured too late, too broadly or without an owner:
- Only lagging measures.
Crash counts and claims arrive after the damage is done. - Fleet-wide averages.
An average hides the few drivers, routes or depots creating a large share of the risk. - No link to action.
A score that isn’t followed by coaching or a schedule change doesn’t move.
Leading vs. lagging fleet safety KPIs
| Lagging indicators | Leading indicators | |
| What they measure | Outcomes that already happened | Behaviors and conditions that come before outcomes |
| Examples | Crashes, injuries, claims, insurance cost | Harsh braking, speeding, following distance, near-misses, coaching completion |
| When they change | After an incident | Weeks or months before an incident |
| Best use | Proving results, benchmarking, insurance renewal | Spotting risk early and directing coaching |
| Weakness | Too late to prevent the event they record | Need context (route, load, traffic) to interpret |
A strong safety scorecard uses both: lagging KPIs to prove progress and leading KPIs to drive it.
12 fleet safety KPIs to track
Type | How to calculate | Why it matters |
Lagging | Crashes ÷ miles driven × 1,000,000 | Normalizes crash counts for fleet size and mileage |
Lagging | Preventable crashes ÷ miles × 1,000,000 | Isolates the crashes your fleet could have influenced |
Lagging | Injury crashes ÷ miles × 1,000,000 | Tracks severity, not just frequency |
Lagging | Total crash-related costs ÷ crashes | Shows the financial impact of severity |
Lagging | Annual premiums ÷ miles driven | Ties safety performance to cost; compare with the industry average of 10.6 cents per mile in 2025 from ATRI’s Analysis of the Operational Costs of Trucking: 2026 Update |
Leading | (Harsh braking + acceleration + cornering) ÷ miles × 100 | The most common early warning of rising risk |
Leading | Speeding events ÷ miles × 100 | Speed raises both crash likelihood and severity |
Leading | Close-following events ÷ miles × 100 | Close following leaves less time to react |
Leading | Recorded near-misses ÷ miles × 100,000 | Captures close calls that didn’t become crashes |
Leading | Coaching sessions completed on time ÷ sessions required | Shows whether data is turning into action |
Leading | Drivers with recurring high-risk events ÷ all drivers | Finds the drivers whose risk keeps recurring |
Leading | Services completed on time ÷ services due | Vehicle condition is part of crash risk |
Track each KPI per driver, per depot and per route, not just fleet-wide. For general fleet metrics such as cost per mile and utilization, see data-driven fleet management: key metrics.
Preventable vs. non-preventable crashes
Not every crash involving a fleet vehicle is the driver’s fault, and treating them all the same distorts safety KPIs.
In the US, FMCSA’s Crash Preventability Determination Program (CPDP) lets carriers and drivers submit eligible crashes for review through the DataQs system. For crashes on or after December 1, 2024, 21 crash types are eligible, including indirect strikes and any crash where video demonstrates the sequence of events. Crashes that FMCSA determines to be Not Preventable are removed from the carrier’s Crash Indicator calculation in the Safety Measurement System and noted in the Pre-Employment Screening Program.
The program is widely used. As of June 26, 2026, carriers had submitted 109,644 requests, and FMCSA had found 71,446 of the 74,513 crashes it reviewed Not Preventable, about 96%, according to FMCSA’s CPDP quarterly statistics. Of the crashes reviewed under the video crash type, 746 of 818 were found Not Preventable.
What this means for KPIs:
- Record preventability for every crash, so the preventable crash rate reflects what drivers and the fleet can actually influence.
- Keep evidence. Video and telematics data are what make a Not Preventable determination possible. Intangles’ AI-driven video telematics and trip data give fleets that record for every event.
- Coach on preventable crashes, and review non-preventable ones for patterns such as specific routes or times.
Why near-misses are worth tracking
A near-miss is an event that could have caused a crash but didn’t: an emergency stop, a swerve to avoid a vehicle, or a close-following event that ended without contact. Near-misses happen far more often than crashes, so they give a fleet many more data points to learn from.
Telematics and video make near-misses measurable. A hard-braking event with forward-facing video shows what happened and why. Reviewing near-misses alongside crashes shows which drivers, routes and situations are producing close calls before one of them becomes a crash. In Intangles, harsh-braking and other DriveIQ events are logged per driver and trip, with location tracking showing exactly where each one happened, so close calls can be grouped by route and time of day. For one common cause of near-misses, see safe following distance for commercial trucks.
How to set targets and close the gap
- Set a baseline.
Use at least three months of data for each KPI, per driver and per depot. - Set targets from your own baseline
Don’t use industry averages. Every fleet’s routes, loads and vehicle mix are different. - Prioritize leading KPIs.
Aim to move harsh events, speeding and coaching completion first; crash rates follow. - Review monthly.
With each KPI owned by a named person. - Act on the gap.
Use the table below.
| Gap | Likely cause | Action |
| Harsh events rising on one route | Traffic, schedule pressure or road conditions | Review the route and schedule; coach drivers on that route |
| A few drivers with repeat events | Habits, fatigue or training gaps | Targeted coaching; check hours and schedules |
| Coaching completion low | No owner or no time set aside | Assign supervisors; set a response time |
| Near-misses rising but crashes flat | Risk building before it shows in outcomes | Act now, before crashes follow |
| Preventive maintenance overdue | Scheduling or parts delays | Prioritize safety-critical components |
| Engine fault code detection | N/A | Intangles |
| Predictive maintenance alerts | N/A | Intangles |
| Driver behavior scoring | N/A | Intangles (DriverIQ) |
| Fuel monitoring and theft | N/A | Intangles |
| Integration method | API | Open API |
For how to structure a complete program around these KPIs, see how to build a fleet safety program.
Common fleet safety measurement mistakes
- Counting crashes without normalizing for miles. A growing fleet will always have more crashes.
- Treating every crash as preventable. It distorts KPIs and demotivates drivers.
- Tracking events without context. A harsh brake that avoided a crash isn’t the same as one caused by tailgating.
- Publishing scores without coaching. Scores change behavior only when someone acts on them.
- Ignoring vehicle conditions. Brakes, tires and lights are part of safety performance.
How Intangles turns safety data into KPIs
Many fleets struggle to measure safety because the data sits in different places: driver events in one tool, vehicle faults in another, video somewhere else. Intangles brings them together in InRoute, one fleet intelligence platform, so every KPI in this blog can be tracked per driver, per vehicle, per route and per depot.
| Intangles capability | What it captures | KPIs it feeds |
| Driver behavior monitoring (DriveIQ) | 20+ behavioral exceptions per driver, such as harsh braking, harsh acceleration and overspeeding | Harsh events per 100 miles, speeding events, repeat-event driver share, driver score trends |
| Location tracking | Trips, routes, mileage and geofences | Normalizing every KPI by miles; route- and depot-level comparisons |
| AI-driven video telematics | Visual context for driving events | Near-miss review, coaching evidence, preventability documentation |
| Predictive health monitoring | Developing vehicle faults before they cause a breakdown | Act now, before crashes follow |
| Operations automation | Alerts turned into scheduled service tasks | Preventive maintenance compliance |
Intangles is a digital twin company operating in 18 countries, with 500,000+ vehicles on its platform and 96% predictive AI accuracy. Fleets using Intangles have seen up to a 75% reduction in powertrain breakdowns. It works with safety-focused fleets in trucking, transit and construction.
Discover how Intangles’ predictive analytics platform can turn driver and vehicle data into safety KPIs your team can act on every month.
KNOW MORE
Frequently Asked Questions
What are the most important fleet safety KPIs?
The most important fleet safety KPIs are crashes per million miles, preventable crash rate, harsh events per 100 miles, speeding events, following-distance events, near-misses, coaching completion and preventive maintenance compliance. Combine lagging KPIs, which prove results, with leading KPIs, which show risk building before a crash.
What is a preventable accident in fleet safety?
A preventable accident is one that a driver exercising normal judgment and foresight could reasonably have avoided. In the US, FMCSA’s Crash Preventability Determination Program reviews eligible crashes, and those found Not Preventable are removed from a carrier’s Crash Indicator calculation in the Safety Measurement System.
What is a near-miss event and why do fleet safety programs track it?
A near-miss is an event that could have caused a crash but didn’t, such as an emergency stop or a swerve to avoid another vehicle. Fleets track near-misses because they happen far more often than crashes and show where risk is building, giving managers a chance to coach drivers or change routes before a crash happens.
How do you calculate crashes per million miles?
Divide the number of crashes in a period by the total miles driven in the same period, then multiply by 1,000,000. For example, 6 crashes over 4 million miles is 1.5 crashes per million miles. Normalizing by miles lets fleets compare performance across periods, depots and fleet sizes.
What is a realistic accident reduction target for the first year of a safety program?
There’s no single benchmark that applies to every fleet, because routes, vehicles and starting points differ. The practical approach is to set targets from your own baseline, focus first on leading indicators such as harsh events, speeding and coaching completion, and review crash rates per million miles quarterly as those indicators improve.
Which driving behaviors should fleet safety KPIs track?
Most fleet safety programs track harsh braking, rapid acceleration, harsh cornering, speeding, close following, distraction and fatigue indicators, and seat belt use where available. Tracking them per 100 miles makes drivers and routes comparable, and pairing events with video gives the context needed for fair coaching.
How does Intangles help fleets track safety KPIs?
Intangles brings driver behavior, vehicle health, location and video data into one platform. DriveIQ flags 20+ behavioral exceptions per driver, location data normalizes events by miles and route, video telematics adds context for near-miss review and preventability evidence, and predictive health monitoring covers the vehicle side of risk. Together, they let fleets track safety KPIs per driver, vehicle, route and depot.
We’re looking forward to meeting you