Skip to content
Blogs / Predictive safety
The GreenRoad Blog
AskMila™, One Driver, and a One-Point Difference: A Quick Win from the Support Desk
See how AskMila™ AI helped one driver improve by a single point – a real support desk story showing how small wins drive lasting fleet safety improvements now.
Sometimes, the best support cases start with a smart question.
A few days ago, a customer reached out to me with something that really impressed me. They were reviewing performance across one of their Operating Units (OUs) and noticed that the overall safety score was unusually high for the period. Their hunch? One specific driver might be skewing the results.
They wanted to validate this: How much of the OU’s safety score was actually being driven by that one driver’s behavior?
Not the kind of general question we get every day—this was precise, data-driven, and operationally significant. If one driver is disproportionately affecting the score, it can influence coaching, reporting, and even internal reviews.
Here’s how I tackled it.
Turning to AskMila™
I decided to run the scenario through our AI assistant, AskMila™. With just two prompts, I got exactly what I needed:
- “Hi AskMila™ what’s the overall driver safety score for now?”
AskMila™ came back with a score of 15. - “What would be the score if we removed the safety events and driving hours of driver from the calculation?”
The new score? 14.
Just like that, I had confirmation: that one driver’s behavior contributed roughly one full point to the overall OU score for the period in question. It was a clean, simple comparison—and a great demonstration of how AskMila™ can be used not just for broad insights, but for really specific what-if analysis.
Closing the Loop
I replied to the customer with a breakdown of the formula we use— (safety events ÷ driving hours) × 10—and explained how the driver’s data influenced the outcome. I also shared AskMila™ output so they could see the process and results themselves.
Their response made my day:
“Jonathan, That’s fantastic—really helpful. I actually tried to look into it myself using AskMila™, but clearly didn’t phrase the question the right way. Once I saw your example, it made complete sense. And it’s great to see that AskMila™ confirms the same numbers I came up with manually!”
This was one of those satisfying moments where a thoughtful customer question, a little AI support, and a bit of human interpretation came together perfectly. It’s also a great reminder: the more specific your question, the more powerful AskMila™ answer.
And if you ever need help framing that question—well, that’s what we’re here for.
Keep exploring
Good to know
Frequently Asked Questions
How is the GreenRoad safety score calculated?
The GreenRoad safety score is calculated as safety events divided by driving hours, multiplied by 10. Because it’s based on events per hour, a higher score means more safety events relative to time on the road. Scores are therefore fair across drivers with different hours.
Can one driver skew a depot's safety score?
Yes. In one customer case, the operating unit’s score was 15. When a single driver’s safety events and driving hours were removed, it fell to 14. That one driver accounted for roughly a full point on the unit’s overall score for the period.
What is a what-if question in AskMila™?
A what-if question asks AskMila™ to recalculate a result under a different scenario. An example is “What would the score be if we removed this driver’s events and hours?”. It lets safety teams test a hunch in seconds instead of rebuilding the calculation by hand.
Can AskMila™ results be checked against manual calculations?
Yes. In this case, the customer had already worked out the figures manually, and AskMila™ returned the same numbers. GreenRoad support also shared the scoring formula and AskMila™’s output, so the customer could see exactly how the result was reached.
Can GreenRoad support help us use AskMila™?
Yes. GreenRoad’s support team helps customers frame questions, interpret results and understand how scores are calculated. If a question isn’t returning the answer you expect, the team can show you how to phrase it.