There are probably as many ways to collimate RCs as there are RC owners...
Don't be afraid to adjust your primary if you have reason to believe that it's out of whack. Best way to determine that is with a star test. If a star in/near the center of the field is showing coma, that's indicative of primary misalignment. There are a couple of ways to determine that -- the simplest is to put in an eyepiece and look. There are also software aids like Metaguide (free) and SkyWave (expensive) if you have a Windows machine you can connect to the scope.
You can always un-adjust what you've adjusted; you may want to keep a log. It rarely takes very much.
Many of the collimation techniques you'll see online use mechanical alignment, e.g. the wall of the tube as a guideline. But the optical figuring and the mechanical bits do not necessarily coincide or align. For example, the donut at the center of the secondary mirror may be precisely in the middle of the circle, but the optical center of the mirror's curvature might not be! So first-order indicators like on-axis coma and off-axis astigmatism are more reliable. This is a classic, and relies solely on the optics.
Two complicating factors that drive many folks batty: Adjusting the primary may throw off your secondary (and vice versa), so the process is necessarily iterative, but hopefully it converges so that you're making tinier and tinier adjustments each time around. The other issue is that the GSO scopes that are all but ubiquitous at the lower end of the market move the primary and the focuser as one assembly when you adjust the former. So as you reduce coma, you can introduce tilt. If you haven't bent anything, and didn't have tilt problems before, you should be able to restore the former state of affairs, yah? If not, a tilt plate is not expensive (my 8" was $108), and you can get the collimation done first, then take out the tilt by eye or by using ASTAP (which I actually installed on my Pi -- slow, but it works).
Another aid of which I've become very fond is the Tri-Bahtinov mask, I use it all the time now to check up on (and touch up) the secondary on my AT8RC. That and a paper donut with one-third of its circumference cut out really does make it super-easy. Focus on one set of spikes. If the others aren't centered, cover up the first set and one of the others, then adjust the secondary until the single set showing is correct. Move to the second one that was out, tune it up. Remove the donut, voila, all three spikes centered, you're collimated. (Note: Do get rid of any tilt before relying on this though!)