The best hint removes half the remaining options instead of naming one of them.
A good hint removes half the remaining options. A bad hint removes all of them. The difference is structural, not a matter of how specific the wording feels.
Consider two hints for the same song. The first says it was a top-ten hit in the mid-eighties. The second says it was by a British duo. Both are one sentence, but the first is a filter and the second is nearly an address. A filter leaves you with work to do; an address leaves you with a search engine.
When you read a hint, ask which of your open questions it closes. If it closes none, the hint was decorative and you should not count it as progress. If it closes two or three, you can update your guess honestly. If it closes everything, you have effectively read the answer early, which is allowed but should be recorded as such.
This is why the order of hints matters more than the number taken. A decade hint followed by a family hint usually leaves a shortlist. A family hint followed by a decade hint behaves the same way, but a family hint followed by an artist hint leaves almost nothing, and the round is over before you have listened again.
If you want to feel the difference, spend a week taking only the smallest hints on give it a try and note how often you still have to guess. The rounds get longer and, annoyingly, better. The skill of reading hints is really the skill of ranking information. The same sentence can be a filter or an address depending on what you already know, so the player who thinks about the ladder before taking a rung gets more from fewer hints, and usually finishes the round with a clearer memory of why. That is the whole trick, and it costs nothing but a pause.