Pete:
Some roofs have their panels mounted on racks tilted to the optimum angle, but the far-cheaper "right on the roof" variety also exists in many, many places.
And rightly so, if I can describe the logic correctly - these numbers are for illustrative purposes only. If I need ten panels' worth of PV output, and if it is 30% more expensive to tilt-rack a set of 10 panels, and I have room on my roof to install 13 panels....then, ta da! - I'm ambivalent as to which route to go.
Now, some real-world experience. One of my three sets of panels in Alaska is ground-mounted (another is tower mounted; the third roof-mounted). As we're at 63ºN lat, our mid-summer incident angle is really steep, and our mid-winter incident angle is really, really steep! I can alter the panels' angle - either every single day or, less insanely, twice a year (say, at the equinoces). HOWEVER....what we have found by experimentation is that the day-long kWh intake, at mid-summer, is indistinguishable whether at "perfect mid-summer" or "dead mid-winter". And, since it's a pain-in-the-neck really kludgy operation to move the panels (I need use my excavator with bucket at full extension), several years ago I abandoned the process and just leave them at their very steepest mid-winter optimum all year round.
I have surmised that one of the reasons this works for me - but might not at 45º or 40º or 35º or 30º latitude is because of the crazy round-the-heavens track that our summer sun follows. That's another situation we have to contend with. Might be, maybe not (It's also partly the reason that our tower-mounted small array - just four panels - is on a tracker. But that's a different subject).
Someone with more knowledge than me is going to have to step in but I'm going to go out on a limb and suggest that maximizing effiency in Alaska is a cost-install outlier. Adding 30% to the cost of even 1 panel just to add a rack seems beyond highway robbery.