How to Log Interview Footage

Logging interview footage means going through every take and recording what was said, when it was said by timecode, and how good it was, so that when you edit you can find any moment in seconds instead of scrubbing through hours of video.

Logging is the step everyone wants to skip and nobody regrets doing. I have watched editors lose entire afternoons hunting for a line they swear someone said, because the footage was never logged. Do it well and the edit feels like shopping from a list. Do it badly, or not at all, and every cut becomes a search party.

What logging actually is

Logging is creating an index of your footage. For each clip or interview, you capture three things: where it is (file name and timecode), what happens in it (the content, ideally the actual words), and a quick judgment of quality (great, usable, no good). That index is the map you navigate the entire edit by.

For interviews specifically, the content layer is everything. A wide shot you can find by glancing at a thumbnail. A specific sentence buried at 00:34:12 in a ninety-minute conversation, you will never find by eye. You find it by reading.

The old way: hand-logging

Before transcription was cheap, editors logged by hand. You would play the tape, pause, type a note and a timecode, play again. This American Life producers were famous for meticulous logging of raw tape, a process Jessica Abel documents in Out on the Wire, her book on narrative radio. It works, and the deep familiarity you build with the material is real. It is also brutally slow. Hand-logging a one-hour interview can eat two to three hours.

The deeper problem is that hand-logging produces a summary, not the words. Your note says 'talks about the layoffs, emotional.' Useful, but when you need the exact sentence, you are back to scrubbing the clip. The note points you to the neighborhood, not the house.

The faster way: log from the transcript

The shift that changed interview editing is cheap, accurate transcription. Transcribe the interview first, and your log writes itself, because the transcript is already a word-for-word index with timecodes attached. Now logging is not transcribing by ear, it is reading and marking.

And reading is fast. The meta-analysis by Marc Brysbaert found adults read non-fiction at around 238 words per minute, while normal speech lands closer to 150 words per minute. You read a transcript noticeably faster than you can watch the footage in real time, and you can skim past the dead air, the throat-clearing, and the questions that went nowhere. A ninety-minute interview you would never re-watch in full, you can read in well under an hour.

What to mark as you read

• The strong soundbites. Highlight the lines that made you stop. These are your selects. If a sentence stands on its own and lands, mark it.
• The story beats. Note where the subject hits each beat your piece needs, the problem, the turn, the result.
• The problems. Flag where audio drops, where a phone rang, where they restarted a sentence. You want to know these before the edit, not during.
• The repeats. Subjects often say the same thing three ways. Mark the best version so future-you does not compare all three again.

Speaker labels and timecode matter

If your transcript has speaker labels, who said what, and word-level timecodes, the line points back to the exact frame, your log is doing the heavy lifting automatically. You read a great line, you already know who said it and where it lives in the footage. That is the difference between a log that summarizes and a log that connects straight to the timeline.

A worked example: cutting a 60-minute interview

You shot a sixty-minute interview for a five-minute brand film. The unlogged path: you open the clip in your NLE and start scrubbing, hunting for good lines, jotting timecodes on a sticky note, losing your place, re-watching sections. Two hours later you have a rough sense of the footage and a messy note. That is before you cut a single frame.

The logged path: you transcribe the hour, then read it in about forty minutes, highlighting the dozen lines that matter and flagging the two spots with audio trouble. Now you have a clean shortlist of selects, each tied to a timecode, and a heads-up on the problem moments. You walk into the edit knowing exactly what your raw material is. The cut starts immediately. For the next step, see cutting a documentary interview.

Common mistakes

• Logging everything equally. A log is a filter, not a second transcript. If you note every sentence with the same weight, you have not logged, you have re-typed. The value is in the highlights.
• No quality notes. A line can be perfect on paper and unusable on screen because the audio clipped or the subject looked away. Note the technical problems while you can still hear them.
• Trusting memory over the log. You will swear you remember where the best line is. You will be wrong by ten minutes. Write it down with the timecode.
• Logging in a vacuum. Log against the story you are trying to tell. A line that is fascinating but off-topic still goes in the no-pile for this project.

The honest tradeoff

Logging takes time up front, and on a tight turnaround it is tempting to dive straight into the timeline. But that time is not added to the project, it is moved earlier and multiplied. An hour of careful logging saves several hours of scrubbing later, and it is calmer work. You are reading and deciding, not fighting a playhead. The one real cost is that transcription has to be accurate enough to trust; a sloppy transcript sends you back to the clip and undoes the gain.

Where the pre-edit fits

Logging and the pre-edit are the same motion when your tool is built for it. In ScriptCut you transcribe the interview, read the transcript with word-level timecodes and speaker labels, and highlight your selects right there. The highlights are your log. From the same screen you remove fillers, arrange the strong moments into an order, and export a ready-to-cut timeline to DaVinci Resolve, Premiere, or Final Cut. There is no separate logging document that goes stale, the marks you make are the edit. To see how that builds into a sequence, read organizing interview footage.

Log first, and the edit stops being a search.

Sources

• Brysbaert, How many words do we read per minute? (2019)
• Errol Morris on indexing footage by timecode, Transom (2003)
• Michael Rabiger, Directing the Documentary (Routledge)