What Is a Selects Reel? The Shortlist Before the Edit (and the Clips)

Somewhere in an editing bay right now, a producer is scrubbing through a three hour founder interview at 2x speed, hunting for the forty seconds that make a good LinkedIn clip. We've all been there. A selects reel is the editor's shortlist of only the strongest moments pulled from raw footage, logged with timecode and set aside before any attempt to tell the story, so the question of what's good never gets tangled up with the question of where it goes. It sits between the stringout (everything, unjudged) and the rough cut (everything arranged), and it's the single step most rushed productions skip, usually to their own cost.

Definition

What separates a selects reel from everything around it

A selects reel, sometimes called a select reel or just "selects," is a pass through your footage where the only job is quality judgment. You watch or read every take, every answer, every alternate line, and you keep the ones that earn their place. Nothing about order, pacing or story logic gets decided here. Before the selects reel, there's usually a stringout: every usable clip from a shoot, laid end to end in whatever sequence makes the footage easy to scan, no creative filtering applied yet. The selects reel takes that raw inventory and strips it down to what's actually worth building with.

The distinction matters because the two tasks use different muscles. Logging is about whether a clip exists and is technically usable. Selecting is about whether it's one of the best things this person said. Conflating the two is how editors end up with a stringout they never actually trim, because they were waiting to feel inspired about structure instead of making the much simpler call of good versus not good.

The bar

How a working editor decides what makes the cut

Noam Kroll, a Los Angeles filmmaker who has cut interview-heavy documentary and promo work, puts the filter in plain terms: "If a clip doesn't immediately grab you, stand out from the others, or provide new insight, leave it on the cutting room floor." That's a genuinely useful test because it's binary and fast. You're not scoring clips on a scale of one to ten. You're asking one question per clip and moving on.

What that bar rules out is almost everything. A perfectly clear, perfectly true answer that says nothing new gets left behind. A slightly messy answer with one unexpected turn of phrase gets kept. This is uncomfortable the first few times you do it, because it feels wasteful to discard hours of coherent, on-topic footage. It isn't wasteful. A documentary built from "everything that's fine" plays like a deposition. One built from "only what grabs you" plays like a film.

Watching someone actually run this pass, cutting selects out of a long interview in real time, makes the ruthlessness concrete in a way a description can't:

Why it's its own step

What breaks when editors skip straight to a timeline

Cutting straight from raw footage to an assembly means solving two problems at once: is this clip good, and where does it belong. Doing both at the same time is slower than doing them in order, and it's worse, because every time you reconsider a clip's placement you end up relitigating whether it was ever worth including. A selects reel answers the quality question exactly once, up front. The assembly that follows only has to answer where each piece goes, working from material that's already been vetted. The assembly cut that comes after selects should be built almost entirely from a trusted shortlist, not from a re-scrub of the entire shoot.

This is also where division of labor becomes possible. On a larger production, one editor or a producer can pull selects while another builds structure, because the selects reel is a shareable, reviewable artifact in its own right. It's not a step that only makes sense on solo projects.

The scale of the problem is why this step exists at all. Documentary filmmaker Frederick Wiseman has shot in the range of 40 to 60 hours of footage for roughly every one hour that makes his final cut, and even a modest interview shoot can produce several hours of usable material for a ten-minute piece. Nobody holds that much footage in their head well enough to structure it directly. A selects reel is how you shrink the problem to a size a human editor can actually reason about before worrying about sequence.

Organization

Grouping selects by theme, not by shoot day

For a documentary built around one subject, selects usually get grouped by topic rather than recording order: everything about childhood in one bin, everything about the career turn in another, regardless of which day of shooting produced each answer. A subject will often answer the same question two or three times across separate sessions, phrasing it slightly differently each time, and the selects reel is where you decide which version actually earns a place, independent of when it happened to be said.

This theme-first grouping is also what makes finding the best soundbite for a given beat tractable. Instead of scanning three hours of footage in chronological order hoping the right line surfaces, you're comparing three or four candidate answers that are already sitting next to each other because someone already judged them worth a second look.

A worked example

Pulling selects from one founder interview

Say you've just shot a ninety minute founder interview for a company history video, and the brief wants four minutes. A straight transcript run comes in around 13,000 words. The stringout is already done the moment the transcript exists: every answer is there, timecoded, in the order it was asked. The selects pass means reading that transcript and marking maybe 25 to 35 lines that clear Kroll's test, probably totaling 1,500 to 2,000 words once you count only the lines you'd actually keep. That's roughly 10 to 15 minutes of raw audio pulled from 90, which is still far more than the four minutes you need, but it's a shortlist you can actually hold in your head and compare answer against answer, which 90 minutes of raw tape never let you do. The rough cut that follows works from those 30 lines, not from the original 90 minutes.

That same shortlist does double duty. Once the four minute video is locked, the leftover selects, the good lines that didn't make the final sequence but still clear the bar on their own, are exactly the raw material a social team needs for clips, and we'd argue there's no reason to make them re-scrub the footage to find them.

Selects vs. the cuts around it

Where it sits in the paper-edit chain

A selects reel has no structure, no pacing, and no sense of beginning or end. It's the good material, isolated and nothing more. A rough cut is the first attempt at ordering those selects into something that plays like a story, and the weakest selects often get cut a second time at that stage once they're sitting next to stronger neighbors. Zoom out further and the whole chain reads: stringout, selects, rough cut, fine cut, picture lock, each pass narrowing and sharpening the one before it. A full paper edit usually folds the selects step in as its second stage, right after transcription and before the written arrangement goes into a draft script.

Editors sometimes work both passes on screen at once, a technique often called pancake editing: the selects reel on one timeline, the developing master on another, dragging clips down from selects into the master as the story takes shape. It keeps the vetted material visible and separate from the structure being tested, so a clip that doesn't work in sequence two can simply go back to the selects pile instead of being lost or second-guessed.

Selects become social content

Can the same selects reel feed a clips or social strategy?

Yes, and it's one of the more underused things a good selects reel gives you. Every line that cleared the bar but lost out to a stronger neighbor in the final sequence is still a clip worth posting on its own; it was never weak, it just lost a competition for airtime inside one specific video. A podcast team that pulls selects from a full episode usually has eight or ten strong moments and room for maybe three in the finished edit, which leaves five or six sitting in the selects pile doing nothing. ScriptCut Social picks up right there: it finds the moments worth posting from the long recording or from a ScriptCut edit, frames them on the speaker, writes captions and a platform-specific description for each one, and schedules them out, with one approval link per client. It posts directly to LinkedIn today; everywhere else, you're exporting the framed, captioned clip and uploading it yourself.

A transcript, not a scrub bar

Why reading selects beats watching them

Pulling selects from hours of interview footage by ear means playing most of it, even at 1.5x speed, just to judge whether each answer clears the bar. A transcript lets you scan for strong answers the way you'd scan a document for the good paragraph: quickly, and without losing your place in a two hour recording. This is the entire idea behind transcript-based editing: you're not abandoning the footage, you're reading it first, then verifying by ear only on the lines that already passed the read-through.

ScriptCut's AI Select step works this way. You read the transcript, mark the lines that meet Kroll's test of grabbing you, standing out, or adding insight, and every marked line carries its exact word-level timecode with it automatically. Nothing gets re-timed by hand, because the timecode was captured at the word level when the transcript was generated. Play any selected line back in place to check tone before you commit, since a transcript can tell you a line is sharp but can't tell you whether the delivery landed. The selects pass that used to eat an afternoon of scrubbing becomes a focused read-through, and it works the same way whether the source is a single-camera interview or a panel with several speakers talking over each other, since AI Select, AI Story and AI Arrange all read speakers apart automatically. It also reads transcripts in 54 languages, so a selects pass isn't limited to English-language shoots.

If you need a clean transcript to hand off alongside the selects, this free SRT to VTT converter also exports a plain text transcript and cleans up caption files without any signup.

Common mistakes

Where selects passes go wrong in practice

The most common failure is scoring clips on usefulness instead of strength, keeping anything that's merely on-topic because it might be needed later. That's a stringout wearing a selects reel's name. A second mistake is selecting in shoot order and never regrouping by theme, which makes every later comparison between similar answers slower than it needs to be. A third is treating the selects reel as final once it's built. It's a hypothesis about what's strong, not a verdict, and a line that reads brilliantly on the page can still fall flat on playback because of pacing, breath, or a flat delivery that text can't capture. Keep the organizing system loose enough to revise as you discover, mid-build, that a clip cut on day one was actually the strongest thing anyone said.

After selects

Turning a shortlist into a real timeline

Once selects are locked, the next move is arranging them into story order and exporting to whichever NLE finishes the job. Because the timecode has been accurate since the transcript stage, the handoff to DaVinci Resolve, Premiere Pro, Final Cut Pro or Avid lands as a real editable timeline with cut points intact, ready to build out rather than something you have to re-cut from scratch. The Resolve and Premiere Pro panel sends footage out and brings the cut back linked to the media; Final Cut Pro and Avid take it in through the XML in the Export menu. If your pipeline already runs through a transcript studio workflow in Premiere Pro, the selects step slots in before that handoff, not instead of it.

If you want to try different selects for different cuts of the same material, build multiple edits from one master transcript without touching the original transcription, so a festival cut and a broadcast cut can draw from the same source without duplicating the transcribing work.

If your footage is sitting in a stringout and you've been putting off the selects pass because scrubbing is slow, try reading it instead. Here's how we'd start: drop the raw recording into ScriptCut, let AI Select surface the strongest lines, and build your selects reel with word-accurate timecode attached to every one you keep.

Sources

• Noam Kroll, best editing workflow for cutting interview footage
• Frame.io, video post-production workflow guide: stringouts
• Documentary video editing workflow, organizing 100+ hours
• Documentary Editing, Chapter 5: Making Select Reels (O'Reilly)
• Film editing, Wikipedia