Hyphens look harmless — they are the shortest, oldest and most compatible dash — which is exactly why they survive unnoticed until something downstream chokes. A name like “Jean-Pierre” breaks a username validator, “well-known” fails a keyword search that indexes the plain word, phone numbers formatted 555-0134 split a column in Excel, and text copied from a PDF arrives with soft hyphens and line-break hyphens that were never part of the words. The Remove tab opens each compound into readable words: “well-known” becomes “well known”, “state-of-the-art” becomes “state of the art”.
The Replace tab is for the destinations that want something other than a space. An underscore keeps the words as one token machines accept — “well_known” for filenames and column headers. Nothing fuses them — “wellknown” for slugs, hashtags and keys. A comma turns a list separator back into punctuation. What you want depends entirely on where the text is going, and this page applies one rule to the whole text in a single pass.
Count first when you are not sure what you are dealing with. The engine matches four hyphens, not one: the keyboard hyphen (U+002D), the Unicode hyphen (U+2010) that PDFs hand you, the non-breaking hyphen (U+2011) that holds a phone number together, and the invisible soft hyphen (U+00AD). They look identical on screen and a find-and-replace for “-” catches only the first. The Count tab reports each one separately, because a cleaned hyphen with a soft hyphen left beside it still fails an ASCII check later.