If I'm bushwhacking and I come across a road or path, that is within 90 degrees of my bearing I will almost always follow it for a short while. The exception would be wide open woods with easy footing.
During the time on the road, I'm evaluating the road vs. the off trail route. I track, on the back burner of my mind, the amount of time on the road and the rough average of the deviation off of my bearing. You can get the deviation by looking at the angle from the compass needle vs. the "box" arrow aka "shed" on the bezel of the compass. So I might think, for example, 10 minutes at 30 degrees left of bearing. When traveling close to a bearing or on a well defined off trail route (such as following along a ridge) I won't look at the compass often, but playing the game of following a road off bearing I keep the compass in my hand and look at it every 20 sec of so.
So how long I stay on the road depends on how far I've deviated from the route vs. the relative difficulty of traveling in the woods. It is a constant decision making process - how far out of the way vs the improved progress.
Then eventually when I leave this road I know the correction needed to get back on track.
This is the same way I navigate even without the road. I look at my bearing and immediately find the easiest route to go in that direction. It might mean that I'm wandering far to the left or right but always tracking my deviation. I once had someone ask me if I knew I was off bearing by 30 degrees for 20 minutes, Of course, but it was the better and faster route.
Evaluating the difficulty of traveling in the woods is based on the visable route and estimated non-visable route speculated upon by years of expereince and subtle changes in the forest lighting / shadows, grades, etc. One insight here, the information about route difficulty is dependent upon finding little holes of open woods in a forest (that may seem like a sea of inpenetrable spruce) Seeing those holes depends on your vantage point which changes every few feet. So as I walk, I constantly make note of the holes - left, right, near, far, infront, on the back azmith, on the route I'm on, and on the ridge next to mine, etc - and then use that "data" for route decision making. Over the years I have developed a swaying motion when moving or even waiting during off trail travel. I've noticed that in one spot you might not see a particular point of interest such as an old blaze or an opening in the woods but with your head 1ft to the left it becomes visible. The sway moves my head to a different vantage point and improves the things I can see.
If the woods are really crappy then I'll make an exception to the 90 degree rule and follow a new found road, away from my bearing, on the chance that it curves around more toward where I'm heading. This has happened more than once. It usually happens when the road switches back or follows a contour
