Camera choice comes down to one question — are you protecting the company from a false claim, or trying to change how people drive? Those need different hardware.
**Road-facing only.** One lens through the windscreen. Records what happened in front of the vehicle. This is evidence hardware: it exists to show that the other party pulled out, ran the light, or stopped without cause.
**Dual-facing.** Adds an inward lens covering the cab. This is where the safety programme lives, because most preventable incidents involve something happening inside the cab — distraction, a phone, following too close, no seat belt.
It's also where driver resistance lives. Worth handling honestly rather than installing quietly.
**AI-enabled.** Processes video on the device and flags events in real time — following distance, lane departure, distraction, sometimes in-cab audible alerts. The difference is that a plain camera gives you footage after somebody tells you to look; an AI camera tells you which clips are worth looking at out of thousands of hours.
Most camera marketing quotes percentages with no source. There is one genuinely well-credentialed data point.
The **National Transportation Safety Board**, in Safety Report SR-15/01 on commercial vehicle onboard video systems, cites a Virginia Tech analysis of **10,648 heavy truck and bus crashes**. Event-based video systems **combined with driver behaviour modification** produced estimated reductions of about **20% in fatal crashes and 35% in injury crashes**.
Two things about that figure deserve emphasis.
First, the effect comes from video **plus coaching**, not from cameras alone. A camera nobody reviews changes nothing.
Second, the report dates from 2015. It remains the best-credentialed source we know of, but it isn't current-year data and we won't present it as such.
The NTSB report makes a separate observation that installers rarely mention: video evidence "eliminated the need for the forensics work traditionally required to quantify occupant motion" in NTSB's own crash investigations. That is a federal safety board saying video changes how a crash gets reconstructed.
You'll see a specific "exoneration rate" cited widely — some percentage of drivers cleared by footage. We went looking for a credible source and every version traced back to camera vendors or law firm marketing.
We're not going to publish a number we can't stand behind. The argument works without it.
The American Transportation Research Institute studied 600 trucking cases from 2006 to 2019. Two findings:
Think about what you have three and a half years after a crash without video. A police report, a driver's memory, and whatever the other side's expert reconstructs from physical evidence.
ATRI also found average verdict size grew **51.7% annually from 2010 to 2018** against 1.7% general inflation, and that awards fell about **13%** when the defence used expert witnesses and the plaintiff did not.
The US Chamber Institute for Legal Reform puts the median nuclear verdict at **$21 million** across 2013–2022, with six states producing roughly 61% of them nationally — **Georgia among them**.
If you run trucks in Georgia, the litigation environment is measurably worse than the national average, and contemporaneous video is the cheapest evidence you will ever buy.
**Storage and retrieval.** Continuous local recording with event upload is the normal pattern — everything records to a card, flagged events go to the cloud. Ask how long local footage survives before it overwrites, because that determines your window to retrieve something after a late-reported incident.
**Windscreen placement.** There's a real tension between camera field of view and driver sightline. It also interacts with lane departure sensors and rain sensors already on the glass. Getting this wrong is the most common visible mistake we see.
**Power.** Same discipline as any telematics install: switched circuit, clean ground, fused, concealed. A camera wired to constant power that never sleeps is the classic cause of weekend battery complaints.
**Driver communication.** Not our job, but the fleets where dual-facing cameras go badly are almost always the ones where drivers found out by discovering the lens. The ones where it goes fine explained the purpose first.
If your exposure is other people's claims — you run in dense traffic, you get rear-ended, you're accused of things that didn't happen — road-facing gets you most of the protective value for the least cost and the least friction.
If you're trying to reduce incidents rather than just document them, the NTSB figure is explicit that the effect comes from video plus behaviour modification. That means dual-facing, and it means somebody reviewing footage and coaching. Without the second half you're buying the hardware and not the outcome.
If you're managing a large fleet where nobody has time to review thousands of hours, AI event detection is what makes review practical at all.
The best-credentialed evidence is NTSB Safety Report SR-15/01, citing a Virginia Tech analysis of 10,648 heavy truck and bus crashes: roughly 20% fewer fatal crashes and 35% fewer injury crashes from event-based video combined with driver behaviour modification. The critical detail is "combined with" — the effect comes from video plus coaching, not from installing cameras. Note the report dates from 2015.
Road-facing if your main exposure is defending against other parties' claims. Dual-facing if you're trying to change driving behaviour, since that's where the preventable events are visible. Dual-facing needs a conversation with drivers first — the deployments that go badly are the ones where drivers found out by noticing the lens.
We won't quote one. Every specific exoneration percentage we could find traced back to camera vendors or law firm marketing rather than to independent research. The stronger argument is ATRI's finding that mean time from crash to verdict is 1,319 days and that longer intervals correlate with substantially larger awards — evidence you already have beats evidence assembled years later.
High on the windscreen, centred behind the mirror where it has clear forward view without obstructing the driver's sightline, and clear of existing lane-departure and rain sensors. Placement is the most common visible installation error, and it's worth getting right the first time because moving it leaves adhesive marks on the glass.
We install road-facing, dual-facing and AI camera systems across Georgia, South Carolina and North Carolina, at your yard. If you're deciding between configurations, call us and describe what you're actually trying to protect against — the right answer differs a lot between a delivery fleet in Atlanta traffic and a long-haul operation.
Get a free quote from API Signal → or call/text 803-513-1009. We install across Georgia, South Carolina and North Carolina, and we come to your yard.
We install GPS, cameras and telematics across GA, SC & NC — at your yard, on your schedule.