You should focus on the score, which takes into account the difficulty of the test.
No school is going to care (or even know) that you only got 6 wrong for your 1400 when another student got 8 wrong and a 1500 on another test date.
It only means your test was a much easier one. The equating process accounts for this and provides a score/percentile that is raw score-independent. “Number wrong” is rather meaningless.
OP, be sure you are using the 10 official College Board tests. (You mention “official”, so it sounds like you are.) Score improvements on those should translate to the capacity for better performance on actual tests.
But actual performance on a single test date has a lot of variance, so it’s no guarantee.