How to Cut Down a Long Interview
You cut down a long interview by choosing the strongest self-contained moments and throwing out everything else, not by trimming the whole thing evenly. Write a one-sentence purpose for the piece, read the transcript instead of scrubbing footage, keep only what earns its runtime, tighten what remains, then reorder for flow. Do that before you open an editing timeline and the edit becomes assembly, not discovery.
Start here
One sentence decides everything after it
Before you touch a single clip, write down what this piece is and how long it needs to be. A ninety-second social cutdown and a five-minute case study pull entirely different moments from the same forty-minute conversation, even when the subject is the same person talking about the same event. Skip this step and every decision downstream turns into a debate, because almost everything in a good interview is at least a little interesting. The one-sentence brief is what lets you say no fast.
Read first, watch second
Work from the transcript, not the scrubber
Documentary editors have called this a paper edit for decades: read the whole interview as text before you touch footage. Reading moves faster than real-time playback and lets you hold the entire conversation in view at once, instead of a ten-second window on a scrub bar. Editor and radio host Ira Glass has described tape-cutting as a skill he built by knowing "the tape he needed" the moment he heard it in an interview, a instinct that comes from tracking the whole conversation, not one clip at a time.
The catch is that a transcript is only as useful as its timecode. If highlighting a line in the document doesn't hand you an exact in and out point on the actual audio, you're back to hunting for the moment by ear. A transcript built for text-based editing ties every word to a frame, so a highlighted sentence becomes a clip automatically.
What earns a place
Four questions for every candidate line
- Does it stand alone? A viewer who's heard nothing else should still understand it.
- Does it earn its seconds? Every second on screen is a second of someone's attention; filler doesn't get a pass because it's near a good line.
- Does it serve the one-sentence brief? Great material that's off-topic belongs in a different piece, not this one.
- Is it the cleanest version? Subjects often repeat the same point three different ways across an hour; keep the sharpest delivery and cut the rest.
Reading for these four questions across a full transcript surfaces the real candidates faster than any amount of scrubbing, because you're scanning for ideas instead of waiting for them to arrive in real time.
Tighten what stays
Cut inside the line, not just around it
Selecting the right moment is only half the job. Most people don't speak in clean sentences, and a strong idea often arrives wrapped in false starts: "So, you know, I guess what we, what we really wanted to do was, basically, build something simple" is really just "What we really wanted to build was something simple." Removing the hedges and restarts doesn't change what someone meant, it just removes the noise around it, and it makes subjects sound like the sharpest version of themselves.
Reorder for momentum
Interviews are never told in the right order
People answer questions in the order they're asked, not in the order that makes the best piece. Your strongest material usually belongs at the open, as a hook, or at the close, as a payoff, regardless of when it was actually said. Reordering complete answers and tightening language inside them are both normal, honest editing. There's a hard line just past that: stitching words spoken at different times into a sentence the subject never said. That's a frankenbite, and it's fabrication, not editing.
The real math
The Fog of War: twenty hours down to a war memoir
For The Fog of War (2003), Errol Morris recorded more than twenty hours of interviews with former U.S. Secretary of Defense Robert McNamara, using his signature Interrotron rig to keep McNamara looking directly into the lens. The finished film runs 107 minutes and won the Academy Award for Best Documentary Feature. That's a real interview-to-runtime ratio: roughly twenty-plus hours of raw conversation reduced to well under two hours of finished film, built entirely around the strongest self-contained passages of McNamara reflecting on the biggest decisions of his career.
You don't need twenty hours of source to learn from the ratio. The same discipline applies at any scale: a sixty-minute founder interview headed for a five-minute piece follows the same shape, self-contained moments pulled first, repetition and tangents cut hard, then the survivors tightened and reordered, all before the timeline build starts.
Where people go wrong
Four mistakes that keep showing up
- Trimming evenly. Cutting ten percent off everything just makes the same sprawling structure ten percent shorter. It's not the same as choosing what matters.
- Selecting on the timeline. Scrubbing through raw footage in an NLE trades the wide view of a transcript for a narrow, slow one.
- Keeping tangents because they're good. A brilliant aside that doesn't serve the brief belongs in a different piece, not a compromise on this one.
- Over-tightening. Stripping every pause and breath makes a subject sound rushed instead of confident. Leave room to breathe.
The tradeoff, honestly
A tight five beats a flabby twelve
Cutting hard means throwing away good material, and that never feels great in the moment. It's worth it anyway: attention is the budget you're spending, and a tight five-minute piece holds it far better than a rambling twelve-minute one, even if the twelve-minute version has more good lines in it somewhere. The discipline here also depends on your transcript being accurate. If the transcription itself is wrong, you'll misjudge lines on the page and need to double back to the footage anyway.
Where this fits in your workflow
The pre-edit, before the NLE
Defining the piece, reading for selects, tightening lines, and ordering for flow together make up the pre-edit, the work that happens before an editing application opens at all. ScriptCut is built around exactly that sequence: transcripts with word-level timecode, a place to highlight and reorder selects, and an export to XML, EDL, or subtitles once the story is locked, so the NLE stage becomes assembly work instead of the place where the story gets figured out. For finding the strongest individual lines inside your selects, see how to find the best soundbites in an interview.
Sources
frequently asked questions
Start with a one-sentence brief for the piece's purpose and target length. That sentence becomes the filter for every moment you consider keeping.
From the transcript. Reading lets you scan the whole conversation at once; scrubbing footage forces you through it in real time and buries quiet moments.
It should stand alone without context, earn its runtime, serve the piece's one-sentence purpose, and be the cleanest version if the subject repeated the point elsewhere.
Yes. Reordering complete answers and tightening language inside them is normal editing. Splicing fragments from different moments into a sentence someone never said is a frankenbite, and that's fabrication.
For The Fog of War, Errol Morris recorded more than twenty hours of interviews with Robert McNamara and released a 107-minute film, an extreme but real example of how much gets cut for the strongest material to survive.
An even cut keeps the same sprawling structure, just shorter. Cutting well means choosing the strongest moments and removing the rest, not shaving a percentage off everything.
You might also like...
