How to Know When You Have a Keeper
Somewhere around generation six or seven, everything starts sounding decent. Not because the tracks are actually improving, but because critical listening is exhausting and your ear starts giving things a pass it wouldn't have given an hour ago. Knowing when you actually have a keeper means separating that fatigue from an actual quality judgment, and that requires a method, not a feeling.
The actual criteria for a keeper aren't "do I like this." They're closer to: does this have at least one element strong enough to build a finished track around, and does nothing in it disqualify it outright. A keeper doesn't need to be perfect. It needs a foundation worth the additional work of mastering and finishing. Something can be a keeper with a mediocre bridge if the vocal performance and the chorus energy are genuinely strong, because those are fixable or workable-around. Something with a great bridge and a vocal that's fundamentally wrong for the song is not a keeper, because vocal character isn't something you patch after the fact.
Listening evaluatively instead of emotionally means asking specific questions instead of waiting for a gut reaction. Does the chorus actually lift, or does it just get louder. Is the vocal performance selling the lyric, or technically correct but flat. Is there a moment, anywhere in the track, that would make a stranger stop scrolling. Emotional listening asks "do I like this." Evaluative listening asks "is there evidence here that this is working," which is a completely different, and much more reliable, question.
A thirty-second triage handles most of the sorting before you commit to a full listen. Jump straight to the chorus, since that's where a track's core hook and energy live, and if the chorus doesn't grab you in the first ten seconds, the rest of the track rarely rescues it. If the chorus does grab you, listen to the opening ten seconds next, since that's what determines whether a listener stays past the first moment at all. If both pass, you've earned a full listen. If either fails clearly, you likely already have your answer, and a full listen is optional rather than mandatory.
"Worth finishing" is a specific standard, not a synonym for "good." It means: if I put real mastering and finishing work into this, would the result represent me at the level I want to be represented at. That's a higher bar than "this doesn't have obvious problems" and a lower bar than "this is my best possible work." A track can be worth finishing without being the strongest thing you'll ever make. The standard is whether the finishing work you're about to invest will pay off, not whether the raw generation is already flawless.
Know when to stop generating even before you find perfect. If you're deep into a session and you've got something that clears the worth-finishing bar, take it. The instinct to keep generating in case something slightly better shows up is a trap, because "slightly better" rarely materializes and the actual cost is real time and real credits spent chasing a marginal improvement over something that already works. Once something clears the bar, the better use of the next hour is starting the finishing work, not gambling for an upgrade.
One more thing that helps more than any of the criteria above: never make the final call in the same sitting you generated in. Come back the next day, on different speakers, with fresh ears. What survives that second listen is telling you something true. What doesn't was probably session fatigue talking, not the track.
Josh, Founder, JG BeatsLab
