I teach a course on genetic genealogy. One of the first things that I tell the students is to avoid the trap of viewing the ethnicity percentages as your "results". This is not only inaccurate, but you miss out on the incredible explanatory power of the test.
Just to keep it simple, the ethnicity percentage prediction that you get on the results page is typically relevant within the past 500 to 1,000 years. That is a huge window of time, especially given the fact that there were so many population shifts throughout Europe just within the past few centuries. I have always liked this video as an illustration of this:
https://www.youtube.com/watch?v=Iha3OS8ShYs
The video is just over 3 minutes long and shows all of the political border changes in Europe over the past 1,000 years. We have to remember that each one of these changes was typically accompanied by massive population shifts. If you had European ancestors, they were almost certainly caught up in these.
As such - you really have to watch when trying to interpret the specific percentages and the specific parts of Europe that the test predicts. The test is highly accurate at the continent level (i.e. if you had European ancestors, you will have European percentages) and is fairly accurate at the regional level (i.e. if your ancestors came from the Mediterranean, you will typically get Italy, Greece, perhaps some Northern Africa and Western Asian). The percentages are not reliable though at the country level. There have been too many changes and population shifts over time.
And - the entire point of my course is that if you really want to use the test results to learn something about your family history, you should focus on the past five generations of your family tree - as you have a 99% chance of matching second cousins and still a 90% chance of matching third cousins (shared great-great-grandparents). Think about it - we each have 16 great-great-grandparents, each of whom was probably born 130-150+ years ago.