For the use case I saw, it was one of several variables fed into a statistical modeling engine that predicted various forms of trouble. Mostly safety and employ

For the use case I saw, it was one of several variables fed into a statistical modeling engine that predicted various forms of trouble. Mostly safety and employee retention. If I recall correctly, the absolute value was less predictive than watching for changes and relative performance vs. peers. It's a good signal that things aren't going smoothly behind the wheel; sometimes is may just be the time and place, in other situations you could have a driver issue. The usual intervention was to trigger someone back at dispatch to get the driver on the horn and ask them if everything was ok... "whoa cowboy, you all right?" Long haul trucking is an interesting place; everyone hates it, pretty much, since you're away from the family for weeks at a time. Turnover rates approach 100%. Bad driving metrics often can be linked to other forms of stress which prompt an unsafe attitude behind the wheel.... I think they wove sentiment analysis into the product as well, looking at what the driver said. Recommended intervention is to get the driver on the horn and ask them if they needed a little help; apparently that tended to cure a good share of the problems...