Jonathan Sayeb
← All work
// weekend project · script analyzer

What makes a video work

A tool I built on weekends so my media team can see why a video lands, shot by shot, instead of only feeling that it does.

01the film
02its sound
03its image
04its story
05one playhead
  1. 01the film
  2. 02its sound
  3. 03its image
  4. 04its story
  5. 05one playhead
3 layersstory, image and sound, on one timeline
Evidencebehind every reading, one click away
Privateused by my media team, by invitation
role · idea, design and buildweekends, autumn 2026private beta, my media team
// where it started

For a while, Thursday evenings belonged to films. After work, my media team and I would put one on and watch it properly, which mostly meant pausing it. Someone would ask why a scene hit so hard, we’d rewind, and the discussion started: was it the cut, the music dropping out, or the actor waiting one second longer than we expected?

Those evenings taught the team more than most of my briefs did. By Monday I’d hear the same words come back in our own edits: hold this shot a beat longer, let the silence do the work.

// what I built

Movie nights don’t scale, though, and they depend on who is in the room. So over a few weekends I built the thing I wished we had on those Thursdays. You give it a video, and it lays out what happens in the story, the image and the sound, second by second, with the reason behind every reading.

It’s for the people who make our videos, not for judging them. The idea is to see what works in a way that feels familiar, like pausing a film together, and then try it in the next edit.

// what it shows

One video, three layers, one playhead.

Here’s the tool’s own test film, The quiet switch: a minute about a woman, a phone and an unfinished page. Play it, or pick a beat, and the reading follows.

The quiet switch · a 60-second test film, written to check the tool

Story
Feeling
Image
Sound
Moments
  • 0:08: Frustration becomes a decision
  • 0:18: First product appearance
  • 0:28: Decisive product action
  • 0:38: Follows through
  • 0:50: First brand cue
0:00–0:08 · beat 01 of 06

Establish the interruption

You see
Maya sits at a desk. A phone lights up beside an unfinished page.
You hear
Three notification sounds interrupt the narration.
It says
“One small interruption. Then another.”
It wants you to feel
Recognition · tension 2 of 5
An invented film with invented data, written so every reading could be checked against a story known beat by beat. The feelings are what the film intends, not what anyone measured.
// story

The beats, and how much tension and curiosity each one is meant to create. You can see where a video asks its question and where it answers it.

// image

Light, colour and framing, shot by shot. Repeated compositions show up as patterns you can point at, and a style atlas keeps the looks worth borrowing.

// sound

Voice, music, ambience and the level of the mix. Silence is a choice too, so it gets its own place on the timeline.

// how it works

4 steps.

01Upload a video

A file from the team. It stays private.

02It watches and listens

Image, speech, music and sound, from the first second to the last.

03It marks what happens

Story beats, shots and sounds, and why each one matters.

04You explore it

One timeline beside the player, with the evidence behind every reading.

// what changes in the room

From “it feels long” to “here’s why”.

Most feedback on a video is a feeling, and a feeling is hard to act on. What I wanted was a way for anyone on the team to point at the moment and say what’s happening in it.

// before
“It feels a bit long in the middle.”
// after
“From 0:28 to 0:38 the story goes quiet on purpose, and it holds because the image keeps moving: the switch, the phone face down. Trim it and we lose the relief.”

An example, on the test film above.

// why it matters

I think most teams already have the taste. What they’re often missing is a shared way to point at it.

The movie nights gave my team that language one Thursday at a time. The analyzer keeps the habit going on the other days of the week, including the busy ones.

// what it won’t pretend

Honest about what it knows.

  • It reads what a video intends you to feel. It doesn’t measure what an audience actually felt.
  • Every report says how much of the video was really looked at and listened to.
  • Lighting setups and creative intent are readings, marked as readings, with other explanations beside them.
  • It helps people learn the craft. It doesn’t grade the people who practise it.
// contact

Let's talk.

The bigger the problem, the more meaningful and exciting… usually.

Jonathan Sayeb · Luxembourg · LegalEnglish · French