i posted a while back a comparison of the same loop route done dozens of times with two bikes that were similar in many ways (position, tires, etc) but with a 15lb weight difference. the average speed over those dozens of ride for each bike was almost perfectly proportional to the weight difference, because the majority of the time on the course was spent climbing. it's also a loop, so wind tends to equalize, although not fully. the difference of those 15lb averaged about 2.5 minutes over a roughly 30 minute ride. within the set of rides for each bike, changes like tires and wheels and clothing and temperature and even wind had a maximum effect of a minute or so.
i've since continued to ride the exact same loop, and the variation in my own output and the wind is about a minute.
so i would say, if a majority of the time of a route is spent climbing - and especially if the descent velocity is skill limited - then the biggest modifiable factor is actually weight. which makes sense, because the biggest force you're fighting is gravity, not friction. this is a 6.85 mile loop with 917 feet of climbing.
