Let the cost of mistakes decide, not vibes
Since precision and recall trade off, you can't maximize both, which means you have to decide which type of mistake is more costly for your specific problem. This decision should come from the real-world consequences of each error type, not from whichever number happens to look better on a report. Every classification problem has two distinct ways to be wrong, a false positive, wrongly flagging something as positive, and a false negative, wrongly missing something that was actually positive, and those two mistakes are rarely equally bad.
Take cancer screening. A false negative there means telling a patient with a real tumor that they're fine, delaying treatment for a serious disease. A false positive means an unnecessary follow-up test that causes stress but is ultimately correctable. Missing a real case is far worse than a false alarm, so you'd deliberately prioritize recall, accepting more false positives in exchange for catching nearly every real case.