Re: small sample sizes. Unless at least 40 students (better 50+) are applying to a college every year from your school, you cannot learn very much from the stats. For colleges with acceptance rates of <10%, you likely need a few more.
So if your HS has 20 applicants per year to a college, you won’t be able to learn much about whether it is a safety, unless every single student has been accepted. If 100 apply, that will tell you a lot more.
For example, over 200 students apply every year from my kid’s HS to UIUC, so there are more than 600 data points of the Naviance scattergram for UIUC. The fact that all kids within a certain range of GPA and SAT were accepted (a solid green upper right corner) would indicate that, for kids with that range of stats from our HS, even Engineering and CS at UIUC was a safety, or close. On the other hand, there were fewer than 15 applying a year to, for example, Lafayette college, so even if all the students who were within a stats range were admitted, there were few enough that other factors besides stats + attending our HS may have determined admission for these students.