Why it matters when you are writing
Statistical testing works with a pair. The null hypothesis states there is no effect or no difference; the alternative states there is. A test estimates how surprising your data would be if the null were true, which is why a non-significant result means you failed to reject the null, not that you proved it. Writing “the results proved there is no difference” is a claim the machinery cannot support, and reviewers strike it.
A directional hypothesis predicts which way the effect goes and buys statistical power in exchange for committing you in advance. A non-directional one predicts only that something differs. Choose before you look at the data, and say which you chose — switching after seeing results is the practice pre-registration exists to prevent.
The bigger honesty issue is HARKing: hypothesizing after the results are known, and presenting a pattern you discovered as something you predicted. It makes an exploratory finding look confirmatory and is one reason findings fail to replicate. The alternative costs nothing: report the analysis as exploratory and say the hypothesis it generates deserves a confirmatory test.