We considered advanced driver training but what is the point of investing in drivers who don't follow the basics. Instead we opted for driver assessment, which is like the driving test but without pass or fail. it's also cheaper and only takes about an hour. No classroom involved. The "tester" gives feedback and provides a report on the drivers ability, followed with a discussion with a line manager.
It provided good feedback, we had one real failure when one of the team didn't pass and gave the I've been driving for 40 years", we replied 40 years is great but tomorrow is where our concerns rest. He resigned in the end.
We also offered a prize draw with some driver related gifts and a top prize of an advanced drivers course. Pleased with the outcomes.

The question is really whether online training is a substitute for in-person training. In our research we found that online training is better for theory than in-class (better retention and understanding), and even better when people have English as a second language or learning difficulties because they can take it at their own pace and use language/literacy tools. That's generic across whatever topics, not just specific to driver training.
We use www.drivingtests.co.nz for things like road rules, fatigue management, logbooks, load security, etc. Then we put selected people with TR Driver Training for in-cab or in-car assessments, plus any other specific machine-related assessments. Online also means that we can maximise the time in the vehicle because DT's courses are fairly cost-effective while hiring a trainer is expensive - we don't want a trainer standing in front of a whiteboard explaining stuff at $180/hr for 4 trainees when the equivalent online course works out at a third that cost, and they can go through it time and time again if required.
What we found didn't seem to work was the video-based hazard perception stuff that's out there. It's just not realistic enough and the drivers found it very frustrating. That kind of thing seems to be better with a trainer. One of the team is a member of the Institute of Advanced Motorists so he helped with hazard perception - using Roadcraft (what the UK police use - you can buy it from Amazon) we started identifying what things drivers were missing, and that was supported by some of the online courses. We encouraged drivers to do 'commentary drives' where they described what they were seeing and why they were making specific decisions; that seemed to make a difference because it made them mindful of their driving. There are plenty of YouTube videos about how to do that, or a trainer should be able to help.
In the end, the blended approach worked for us - theory online, practical with a trainer, then put selected people back through the theory (mainly because you forget 90% of what you learned within a couple of weeks).