What does the review workload calculator tell you?
A driver-monitoring (DSM) camera and a forward-facing collision (ADAS) camera raise an alert when they see something that looks like fatigue, distraction, a close follow or a lane drift. Some alerts are real and some are false. Someone has to open the clip, decide, and act.
This calculator turns your alert counts into review hours, reviewers and the days of alerts that wait. The alert volume tab shows how many alerts a day your fleet raises. The reviewer staffing tab compares the hours with your team. The review policies tab shows what changes if you open every alert, only the serious ones, or a sample.
Why the review load matters
- A queue that grows is a queue nobody trusts. An alert reviewed a week late does not change how a driver drives.
- A false alert costs the same minutes as a real one. Every false alert opened is time taken from a real event.
- Some fleets open only part of what the cameras raise. The calculator shows how many real events that choice leaves unseen.
- Lowering the sensitivity also removes real events. Filter first, then fix the causes of false alerts: camera position, calibration and glare.
Inputs you need
| Input | Where to find it |
|---|---|
| Vehicles | Vehicles whose cameras send alerts to your review queue. |
| Alerts per vehicle per day | A week or more of your own event report, divided by vehicles and days. |
| False alerts | A sample of reviewed alerts: false alerts divided by all alerts. |
| Review minutes | Time 20 real reviews, including the note or action, and take the average. |
| Serious share | Your own rule for a serious event. Tracom does not set it. |
| Shift, share on review, days | Your working rules and your review process. |
The table in the tool opens with example values. They are not industry averages and not Tracom device behavior. Replace them.
How it is calculated
Alerts and review time
Reviewers and queue
The lamp on the staffing tab compares review hours with team hours. Up to 100% is ready, up to 115% is check, above that is stop. The serious-only and sample policies assume that the false share inside the opened alerts equals the share of the alert type.
Worked example: 120 vehicles, five alert types
- Alerts: 120 vehicles x 4.1 alerts each = 492 alerts an operating day. About 72% of them are real.
- Review time: 486 minutes, which is 8.1 hours a day and 48.6 hours over 6 operating days.
- One reviewer: 8 h x 70% x 5 days = 28 hours a week. 48.6 / 28 = 1.74, so you need 2 reviewers. Two reviewers have 56 hours, so the team runs at 87%.
- With one reviewer: about 5,000 alerts (82 hours) wait after 4 weeks, which is 10 operating days of alerts.
- Serious only: 112 alerts and 2 hours a day, so 1 reviewer is enough. About 271 real events a day are never opened.
- Tuning: removing 30% of the false alerts cuts review time from 8.1 to 7.4 hours a day. Real events stay.
The calculator opens with these values, so the numbers on screen match this example.
Reading the results
- Alert volume shows alerts, false alerts and review time by type. The share spent on false alerts tells you where tuning pays back first.
- Reviewer staffing gives a lamp with a word and a percentage. Treat check as a warning: a queue grows even when the team is only a little short.
- Review policies shows what each policy costs in hours and what it leaves unseen. The matrix shows the fleet size where your team stops fitting.
Common mistakes
- Counting from one quiet day. Use a full week or more.
- Using the clip length as the review time. Review includes finding the clip, deciding, labelling and acting.
- Assuming the whole shift is review. Notes, handover, escalation and breaks take time.
- Leaving out coaching. Talking to drivers is separate work. Lower the share on review to make room for it.
- Cutting the sensitivity to shorten the queue. It removes real events along with the false ones.
- Treating serious-only as free. It is a choice to leave real events unseen. Write down who accepted it.
Assumptions and limits
Results are estimates. Alert counts depend on the cameras, their settings, the routes and the drivers, and review time depends on your process. The example values in the table are not Tracom device specifications and not industry averages. The calculator does not say what Tracom M1N will raise on your vehicles: read that from your own event report.
Further reading on the sensitivity trade-off and on planning a monitoring program (links and content checked on 3 November 2026):
- Fleet News: warning about lowering AI dashcam sensitivity
- Canadian Underwriter: the accuracy trade-off in AI dashcams
- NIOSH (CDC Stacks): who is monitored, roles and training as planning steps
Verify any internal service level or legal duty against your own policy and the current official text.