Attenuation for brewer's yeast generally ranges from 65-85%, and could be 60-90% if using lots of simple-sugar adjuncts or lots of unfermentables. That's a big range. For a "normal" beer, with an OG of 1.060, the range would be between 1.09-1.021, which is so large as to be practically useless.
To just pick 75% attenuation as your estimate is a sure way to be wrong almost all of the time.
If you have specs for the yeast strain you're using, it's a good start. For instance, the expected attenuation for WLP001 is 63-70%, so the beer in our example would finish between 1.018-1.022, which is a much more reasonable range, but is still a larger range than I feel comfortable trusting.
Only way to get a really good idea is to repeat recipe and same process, and even then it'll only work for that one brew and won't always be exact. I routinely get higher than any given yeast's attenuation ratings and drop lower than software's FG estimates by a few points. So I just adjust my recipes assuming that to be the case and it typically works out ok.
That said, I've found that extract/steeping grain recipes tend to be a bit more predictable than AG or PM recipes, as the grain and the mash add a bunch of extra variables to the equation. Not saying I know why or that that's always the case, just my experience.