Jared Gillespie

CIS 103 / Video storyboard

Why people click off your videos

My plan for a 4-minute video that teaches new creators how to find their retention graph, read it, and fix one thing in their next upload. The video gets made in Module 5.

The plan

Topic, audience and purpose

Topic
Audience retention: why people click off your videos and how to fix it
Audience
New YouTube creators who just started posting and don't know why viewers leave.
Purpose
By the end, viewers should be able to find their retention graph in YouTube Studio, read the four key moments on it, and fix one thing in their next upload.
In one sentence
New creators should learn how to find and read their audience retention graph so they can fix one specific thing in their next video.
Total planned length
4:00
Format
Face cam for the intro, fixes and outro, plus screen recordings of YouTube Studio with voiceover.
Storyboard

Scene by scene

Lines in brackets are pauses or demos, not narration. The tags under each scene (S1, S2, S3) point to the source that backs up what I'm teaching there.

Scene 1

Hook

0:00 to 0:20 · 20 sec

What we see

Face cam, medium close-up at my desk. I'm holding my phone and flick past a few videos with my thumb. The phone screen faces away from the camera, so no one else's content shows. Title card fades in at the end.

What we hear (narration)

Be honest, how many videos did you swipe off today in the first few seconds? Probably a lot.

If you just started posting, your viewers are doing that to you too. The good news is YouTube shows you exactly where they leave. Let me show you how to read it.

Sound: Low background music starts under my voice, then dips when I talk.

On-screen text

Why people click off your videos

Assets and permissions

Face cam: recorded by me. Music: one track from the YouTube Audio Library (free, check its attribution column). Title card: made by me in Canva free.

Scene 2

What retention is

0:20 to 0:50 · 30 sec

What we see

Cut to a simple animated line graph that starts at 100% and slopes down. Then a quick text card with "2012" and "views dropped 20%". Back to face cam for the last line.

What we hear (narration)

That's called audience retention. It's basically how much of your video people actually watch before they leave. And it matters more than you'd think.

Back in 2012 YouTube started counting watch time in recommendations, because clicks alone were rewarding videos that didn't deliver on their title. Views across the site dropped 20 percent when they did it, and they kept it anyway. So if people bail early, YouTube notices.

Sound: Music stays low. Small whoosh on the text card.

On-screen text

Audience retention = how much of your video people watch2012: YouTube starts counting watch timeViews dropped 20%. They kept it.

Assets and permissions

Graph and text card: made by me in Canva free (Canva's free elements are licensed for use in videos). Face cam: me.

Supports the teaching: S3

Scene 3

Where to find it

0:50 to 1:25 · 35 sec

What we see

Screen recording of my own YouTube Studio. Cursor clicks Content, picks a video, clicks Analytics, then the Engagement tab. Editor zoom pushes in on the audience retention graph. Revenue and private info stay off screen or get blurred.

What we hear (narration)

Here's where to find it. Open YouTube Studio, hit Content, pick a video, click Analytics, then the Engagement tab.

[About 8 seconds of clicking, voice paused]

This graph is your retention. One heads up though, YouTube only highlights key moments on videos that are at least a minute long with 100 views, so use one of your older uploads if your newest one is too fresh.

Sound: Music off so the clicks are easy to follow. Voice only.

On-screen text

YouTube Studio > Content > Analytics > EngagementKey moments need 60+ sec and 100+ views

Assets and permissions

Screen recording: my own YouTube Studio, captured with OBS. Zooms and blur: done in my editor.

Supports the teaching: S2 S1

Scene 4

Reading the graph: the 4 moments

1:25 to 2:15 · 50 sec

What we see

Same screen recording, zoomed on the graph. As each moment is named, the cursor hovers it and a colored label pops up next to it: Intro, Dips, Spikes, Top moments. Short hold on each label.

What we hear (narration)

YouTube can mark four kinds of moments on this graph. First is the intro, which is the percent of people still watching after the first 30 seconds.

[2-second hold on the label]

Then dips. Those are spots where people skipped ahead or just left.

[2-second hold]

Spikes are parts people rewatched or shared. That's either a good sign or a sign that part was confusing, so you have to actually go watch it.

[2-second hold]

And top moments are stretches where almost nobody dropped off.

Sound: Soft pop sound when each label appears. Voice only otherwise.

On-screen text

Intro: % still watching at 0:30Dip: people skipped or leftSpike: rewatched or sharedTop moment: almost nobody left

Assets and permissions

Screen recording: mine. Labels: made in my editor.

Supports the teaching: S1

Scene 5

Fix 1: the intro

2:15 to 2:55 · 40 sec

What we see

Face cam. Then a demo: two 6-second clips of the same video opening, played back to back with labels. Clip A rambles before getting to the point. Clip B opens on the thing from the thumbnail. Back to face cam.

What we hear (narration)

Alright, three fixes. Fix one is your intro. YouTube says a strong intro usually means the first 30 seconds matched what the title and thumbnail promised.

Here's a bad one.

[Clip A plays, about 6 seconds]

And here's the same video, but it opens on the thing from the thumbnail.

[Clip B plays, about 6 seconds]

Night and day.

Sound: Music off during the clips so their audio is clear.

On-screen text

Fix 1: Match your thumbnail in the first 30 secondsClip labels: "Before" and "After"

Assets and permissions

Face cam: me. Clips A and B: two openings I record myself for the demo.

Supports the teaching: S1

Scene 6

Fix 2 and 3: dips and top moments

2:55 to 3:35 · 40 sec

What we see

Back to the screen recording. Cursor clicks the biggest dip and the video preview plays that exact spot for a few seconds. Then the cursor jumps to a top moment late in the graph, and an arrow animates it moving toward the start of the timeline.

What we hear (narration)

Fix two, go to your biggest dip and watch that exact part. Most of the time it's a slow setup or something you could've said in half the time, so cut it or speed it up.

[Dip moment plays, about 4 seconds]

Fix three, check your top moments. If your best part shows up late, move it earlier, because there's always fewer people left at the end. Then make more videos like it.

Sound: Voice only, quick whoosh on the arrow.

On-screen text

Fix 2: Cut or speed up your biggest dipFix 3: Move your best part earlier

Assets and permissions

Screen recording and video clip: my own channel. Arrow: made in my editor.

Supports the teaching: S1

Scene 7

Wrap up

3:35 to 4:00 · 25 sec

What we see

Face cam, same framing as the hook. End card with the three steps stays up for the last 4 seconds.

What we hear (narration)

So here's your homework. Open your last video, check the intro number, find the biggest dip, and see where your top moment lands.

You don't have to fix everything at once. Just fix one thing in your next upload and compare the graphs. That's it, go get people to actually stay.

Sound: Music comes back in under the end card and fades out.

On-screen text

1. Check your intro 2. Find your biggest dip 3. Move your best part up

Assets and permissions

Face cam: me. End card: made by me in Canva free. Music: same Audio Library track.

Timing

Scene durations

#SceneTimeLength
1Hook0:00 to 0:2020 sec
2What retention is0:20 to 0:5030 sec
3Where to find it0:50 to 1:2535 sec
4Reading the graph: the 4 moments1:25 to 2:1550 sec
5Fix 1: the intro2:15 to 2:5540 sec
6Fix 2 and 3: dips and top moments2:55 to 3:3540 sec
7Wrap up3:35 to 4:0025 sec
Total4:00

The script is about 419 spoken words, roughly 168 seconds at a normal pace. The rest of the 240 seconds is the clicking in Scene 3, the label holds in Scene 4, and the demo clips in Scenes 5 and 6.

Research

Sources

  1. S1

    YouTube Help. "Measure Key Moments for Audience Retention." YouTube Help, Google, support.google.com/youtube/answer/9314415. Accessed 18 Oct. 2026.

    Defines the four key moments (intro, top moments, spikes, dips), the 30-second intro measure, the fixes YouTube suggests, and the 60-second / 100-view minimum.

  2. S2

    YouTube Help. "Understand Your YouTube Engagement." YouTube Help, Google, support.google.com/youtube/answer/9313698. Accessed 18 Oct. 2026.

    The click path to the Engagement tab and where the audience retention report lives.

  3. S3

    Goodrow, Cristos. "On YouTube's Recommendation System." YouTube Official Blog, 15 Sept. 2021, blog.youtube/inside-youtube/on-youtubes-recommendation-system/.

    Why YouTube started counting watch time in 2012, and the 20% drop in views it accepted to do it.

Media plan

Assets and permissions

Face camRecorded by me on my webcam. Original.
YouTube Studio screen recordingsRecorded by me with OBS on my own channel. Original. Revenue and private info blurred.
Demo clips (Scenes 5 and 6)Clips from my own videos, plus two openings I record just for the demo. Original.
Graphs, text cards, end cardMade by me in Canva free. Canva's free content license covers using its free elements in videos.
Labels, zooms, arrowsMade in my video editor. Original.
MusicOne track from the YouTube Audio Library, which is free for creators. If the track lists an attribution requirement, I'll credit it in the description.
Sound effectsWhoosh and pop from the YouTube Audio Library sound effects. Free, no attribution needed.
Other people's contentNone. My phone screen faces away in the hook so no one else's videos show up.

No interviews. Every scene is narrated by me, so there are no answers to plan around.

Accessibility

Accessibility notes

  • Captions on the whole video. I'll fix YouTube's auto captions by hand and upload them, not leave the raw auto version.
  • The narration says every click out loud (Content, Analytics, Engagement), so someone who can't see the screen recording can still follow it.
  • On-screen text is large, high contrast, and stays up at least 3 seconds. Nothing flashes.
  • Music stays well under my voice and cuts out completely during the screen recording and the demo clips.
Reflection

One decision, one challenge

The biggest planning decision was putting the where-to-find-it walkthrough before any of the fixes. Most retention tips jump straight to "make a better hook," but a new creator can't fix a dip they don't know how to find. So the screen recording comes first, and every fix after that points back to the same graph they just learned to open.

The challenge is that key moments only show up on videos that are at least a minute long with 100 views. A brand new creator might not have that yet. So I'm recording the walkthrough on one of my older uploads that qualifies, and I put a heads up in the script so viewers know why their graph might look empty. The screen recording parts can also drag, so I'll cut the dead time between clicks and zoom in on the graph instead of showing the whole Studio screen.