I consider the following, in association with the statistical fuel figures we get.
1. Am I likely to be taxiing for longer than the plan suggests?
2. Am I likely to be able to fly from A to B at the planned flight level without having to deviate for weather?
3. Am I likely to be able to make an approach without holding?
4. Am I likely to need fuel for more than one approach (if the wind is close to limits, for example)?
Then, based on the above, and on what has happened to me on that route in the past, I take a sensible amount of extra fuel if I think it’s necessary. An old sage once said to me, “I can choose to worry about the weather, or the fuel on board, but I’m not going to worry about both on the same flight.”