The Science Behind the Score
How does AttentionLeak
measure distraction?
Every video is broken down into five measurable signals. Those signals are combined into a single Attention Score — a number from 0 to 100 that tells you how demanding a video is on your focus.
Score at a glance
Low fragmentation. Safe for deep work or learning.
Some stimulation. Fine for casual viewing, not ideal for focused work.
Heavy fragmentation. Likely to pull attention and increase mental fatigue.
The Process
How a video gets its score
From URL to score in five steps. Tap any milestone to see what happens under the hood.
Paste any YouTube URL
Public or unlisted — no login needed.
AI extracts five signals
Frame-by-frame analysis across video and audio.
Score computed — 0 to 100
Weighted formula, no black boxes.
Video classified and tagged
Focus-friendly, Engaging, Distracting, or Overstimulating.
Explore your results
Score, signal bars, classification — all in one view.
Paste any YouTube URL
Public or unlisted — no login needed.
AI extracts five signals
Frame-by-frame analysis across video and audio.
Score computed — 0 to 100
Weighted formula, no black boxes.
Video classified and tagged
Focus-friendly, Engaging, Distracting, or Overstimulating.
Explore your results
Score, signal bars, classification — all in one view.
Platform Philosophy
Why YouTube? And not Instagram.
We didn't pick YouTube at random. We picked it because it's the only major platform where quality is still a choice.
People go to YouTube to learn, grow, and invest real time. They expect that time to be worthwhile. We exist to prove whether it actually is — and separate the videos that earn your attention from the ones that just steal it.
Instagram is engineered for short bursts by design. Reels, auto-scroll, Stories — the format itself fragments attention before any creator gets involved. There's no meaningful baseline to score against.
The bigger picture
YouTube is where the battle for your attention is actually winnable. Creators there canchoose depth over chaos — but many don't. AttentionLeak exists to make that visible, so you only spend time on content that actually earns it.
Other platforms? We're watching. If the content culture shifts toward quality, we'll be there.
The Five Signals
These are the raw measurements extracted from every video. Each one captures a different way a video can tax your attention.
What it measures: How many times the editor switches the camera or scene each minute.
Why it matters: Every cut is a micro-interruption. High cut rates keep your brain constantly re-orienting instead of absorbing content.
What it measures: How much movement is happening on screen at any moment.
Why it matters: Constant on-screen movement competes for your visual attention and drains focus faster than static or slow-moving content.
What it measures: How loud and intense the audio track is on average.
Why it matters: Sustained high audio energy keeps the nervous system in a heightened state, making it harder to settle into deep focus.
What it measures: What fraction of the video has no speech or music.
Why it matters: Silence gives your brain space to process what it just heard. More silence usually means more room to think.
What it measures: How well the video cuts line up with the beat of the background music.
Why it matters: When edits match the music rhythmically, transitions feel smoother and less jarring to the viewer's attention.
Tags & Classifications
Special Tags
Audio Safe
The video may be visually distracting, but its audio track is calm enough to listen to in the background — like putting on a podcast while working. Safe to play with sound on even during focused work.
Classifications
Low cut rate, calm motion, balanced audio. Great for focus sessions.
Stimulating but structured. Keeps attention without overwhelming it.
Frequent cuts or high motion that pulls focus away from content.
Extreme sensory load. Rapid cuts, loud audio, and constant movement combined.
Analysis takes a moment — please bear with us
Generating an accurate score isn't instant. Our system extracts every frame and audio sample for high-precision analysis, then runs multiple passes before producing a result. This can take anywhere from 1 to 5 minutes depending on the video length.
We think the wait is worth it. A score based on real signal data — not a quick guess — is what makes the number meaningful. Thank you for your patience.
Ready to check a video?
Back to the feed