The rule may be asking the wrong question
A motion sensor can be excellent at noticing someone walking into a hallway and much less useful at recognising someone reading quietly. PIR detection depends on movement within its coverage and an appropriate line of sight.
If the rule says “no motion means nobody is here”, an occupied room can go dark. Increasing the timeout may help, but it does not address every cause.
Look at placement before adding complexity
Check what the sensor actually covers after furniture and doors are in place. Test the activities that cause trouble, including sitting still. Different sensing technologies and models have different strengths; buying a “presence” label is not enough.
I separate arrival behaviour from the decision to switch off. Useful additional inputs may help, but a pile of unreliable signals just creates a more complicated guess.
Make the mistake recoverable
Give people an obvious manual control and a way to keep the room lit. A quiet study needs a different policy from a passage.
Then test again under ordinary conditions. The target is comfortable use, not the shortest possible timeout.