When you shim starting from the spur, you are obscuring the path to the optimum shim solution because the spur's location is at least somewhat if not wholly dependent on the pinion-bevel mesh (depending on brand/fitment/gearbox type).
When starting with the spur, you're "winging it" to some degree, which hugely extends iteration time. If you can test a greater number of shim solutions than your competitor in a given amount of time at the work desk, then you will have a better shim solution faster.
Worst of all and most importantly, shimming from the spur will make it impossible to test your shimming solution "from the bottom up" as you go. Testing from the bottom up quickly informs you whether you've got problems.
I think we should be suggesting the method which is well-known to be more reliable across a wider range of brands, tolerances/precision, workloads, and skill levels... and is less likely to lead newbies astray.
Shimming from the pinion-bevel mesh first enlightens people as to why they're shimming, because the results of bad positioning are the most obvious at that mesh. The pinion-bevel mesh is the most difficult shimming in the system, and the remainder of the gears are dependent on the bevel being set correctly
Proper shimming is like a search algorithm which explores all the permutations of positions to find the best solution. The possibilities are arranged as a tree from most-influential to least-influential (or most-dependent on the next influencer over). Spur-shimming advocates starting several levels into the tree instead of at the root, which is wrong.
Instead, by starting at the pinion-bevel point, (the root of the tre), no further motor adjustment will be required once you've "solved" this point. Also, since you've only got one gear in your gearbox at the beginning, you will be able to quickly test whether everything is good by slapping your motor grip on there and checking if it sounds correct. A bad pinion-bevel mesh can be detected very easily at this point, especially if your shimming has left no possibility of movement. As you progress to the dependent gears, you will require less time to test each solution because everything you've passed is already "solved" and you can leave it be (assuming you don't end up with a high sector, but this can be avoided by keeping the spur as low as possible while still engaging the bevel with as much contact as possible).
Think about it: If your bevel's range of movement has already been limited by your spur and none of its available positions are optimal for the motor's pinion gear, then all you can do is worsen the situation by moving the motor downwards (a common piece of bad advice). This lowers torque, increases stress, greatly increases the likelihood of slip, which leads to the stripping of your pinion gear or too much load on your motor. You can avoid even having to make this mistake in the first place.
Pinion-bevel shimming can be done almost mechanically with no thinking. Just a simple loop of adjusting, observing, testing and re-adjusting. We should be telling everyone to do it.