Gazenest Methodology
How Gazenest measures YouTube watch behaviour and why each metric was designed the way it was.
Honest Watch Tracking
Every score and metric in Gazenest starts from Honest Watch Tracking: a measurement standard that counts only seconds where the YouTube tab is active, the browser window has focus, and the video is not in hover-preview mode. All three conditions must be true simultaneously.
This differs from YouTube's own measurement, which counts all time the video is open regardless of whether the window is focused or the user is present. The result: Gazenest numbers are typically lower than YouTube's own estimates. The gap represents passive exposure -- the video was running but you were not actually watching.
All three scores below are derived from Honest Watch Time, not total open-tab time. This matters: a score computed from inflated data is a score about your proximity to a screen, not your actual attention.
Self-Control Score / 0-100, weekly
The Self-Control Score measures how well your YouTube sessions this week matched what you intended when you started them.
It is powered by Intent Mode: before each session you choose a mode (Learning, Relax, Research, Specific video, or Free scroll), and Gazenest evaluates how the session actually unfolded relative to that declaration. The score is updated weekly.
Behavioural science basis
Gollwitzer (1999) showed that "implementation intentions" -- specifying when, where, and how you will pursue a goal -- substantially increase follow-through compared to vague goals. Intent Mode is a direct application of this: the act of naming your intent before opening YouTube is the intervention, not just a data collection step. The Self-Control Score is the feedback loop that closes the cycle.
Free scroll sessions are not penalised. The score measures the gap between declared intent and actual behaviour -- not whether your choice was productive.
Clarity Score / 0-100, weekly
The Clarity Score measures the coherence and focus of your watching sessions. A session where you watched a tightly connected set of videos scores higher than a session that pinballed across 20 unrelated recommendations.
Clarity is a session-level metric, distinct from the Diversity Score which operates at the week level. You can have high Diversity (varied topics across the week) while maintaining high Clarity (each session was internally coherent).
Behavioural science basis
Passive YouTube usage is characterised by low clarity: sessions that start with a stated purpose and end in an unrelated recommendation chain. YouTube's autoplay and sidebar are optimised for session length, not for topic coherence. The Clarity Score makes this drift visible after every session.
Diversity Score / 0-100, weekly
The Diversity Score measures how varied your YouTube consumption is across topics and channels over the past seven days. It operates at the week level, not the session level.
There is a healthy band for each user, computed from their own baseline. A score below the band signals algorithmic narrowing -- your consumption has collapsed into a small set of channels and topics, likely driven by recommendation feedback loops. A score above the band signals scattered, unfocused browsing. The healthy band is personal, not universal.
Behavioural science basis
Pariser (2011) described the filter bubble: recommendation systems show you more of what you have already watched, progressively narrowing your information diet. YouTube's own 2016 paper (Covington et al.) explicitly names session watch time as the optimisation target -- the algorithm is designed to extend sessions, not broaden exposure. The Diversity Score gives users visibility into whether that narrowing is happening in their own watch history.
Weekly cadence
All three scores cover the rolling seven-day window ending at the time of viewing. The dashboard shows the current week's scores and a trend chart for previous weeks so you can track whether your habits are improving, stable, or drifting.
A weekly cadence was chosen deliberately over real-time scores. Day-to-day variation is high and not meaningful -- one unusual day does not indicate a pattern. A week is long enough to smooth noise while being short enough to be actionable. The weekly email report summarises the same period.
Last updated: 15 June 2026