Every platform publishes a maximum, and the maximum is the number everybody reaches for. Sixty seconds, ninety seconds, three minutes. It is the wrong number to design around, because it describes what the platform will accept, not what the moment you are cutting needs.
A clip has one job: it has to make sense to somebody who has not watched the hour it came from. That is a property of the material, and the material decides how long it takes.
What a clip has to contain
Take a straightforward case. Someone asks a question, there is a pause, and the answer lands. The answer is the reason you are making the clip — but on its own it is a sentence with no subject. A viewer who joins on the answer hears an opinion arriving out of nowhere and swipes.
So the clip has to hold the question as well. Now measure it: the question is eight seconds, the pause is two, the answer runs twenty-five, and there is a reaction afterwards that is genuinely the best part. That is a forty-second clip, and there is no version of it that is fifteen seconds without throwing away the thing that makes it work.
The opposite case exists too. A single line delivered cold, with nothing before it that matters, is a twelve-second clip, and padding it to thirty seconds with the run-up makes it worse. Length is not a quality signal in either direction.
Bands, not durations
ClipSpark asks for a length band rather than a number of seconds: 15–30, 30–45, or 45–60. That is deliberate, and the reason is what happens at the edges of a cut.
If you specify thirty seconds exactly, something has to give at both ends. The cut either starts mid-word or the timing is nudged and you did not get thirty seconds anyway. A band says the same thing more honestly: the moment is placed inside a range, so the start and the end can land on sentence boundaries and the duration comes out wherever it comes out.
That is the difference between placing a cut and trimming one. Trimming takes a fixed length and slices the material to fit. Placing takes the material and finds the window it fits into. The first produces a clip of exactly the right length that begins halfway through a clause; the second produces a clip of roughly the right length that begins where a person would have begun it.
Choosing a band
A rough guide, worth more than a number pulled from a study:
- 15–30 seconds for a single statement that needs no setup — a claim, a one-liner, a reaction. If you find yourself wanting to include the question, this is the wrong band.
- 30–45 seconds for a question and its answer, which is most of what is worth clipping out of an interview.
- 45–60 seconds for an explanation, a story with a turn in it, or an exchange where two people build on each other. This band is also the one where a weak moment is most obviously weak, because there is nowhere to hide.
You are not committing to one answer for the whole upload. Asking for several candidates from one file costs a few extra credits each rather than a second full charge, and the practical way to settle a length question is to generate the same moment in two bands and watch both.
Silence changes the arithmetic
Optional silence removal tightens a clip by cutting dead air, and it can move a cut from one band into another — forty seconds of speech with eight seconds of pauses in it becomes a thirty-two-second clip. What it deliberately does not touch is the short pause between sentences. Those pauses are most of what makes recorded speech listenable, and a clip with all of them removed sounds like somebody talking without breathing.
What this article is not going to tell you
There is a genre of article that answers this question with a number and a percentage — an optimal duration, a drop-off point quoted to the second. We do not have those numbers. ClipSpark does not follow a clip after you download it, so it never learns whether anybody finished watching. Any figure here would have been borrowed without a citation or invented, and a made-up number is worse than no number because it sounds like evidence.
What we can be held to is mechanism: what a cut throws away, where a viewer loses the thread, and what the tool does with a band once you pick one.
The short version
Ask what the shortest self-contained version of this moment is, then pick the band that contains it. If the answer is not obvious, generate it twice. The platform's maximum should never be the reason a clip is the length it is.