Research methodology

The Viral Hook Analyzer Research Dataset is a working corpus of analyzed short-form and long-form videos used to inform Answer pages, Insight articles and tool outputs. This page documents how that data is collected, scored and cited.

What is Viral IQ

Viral IQ is our composite 0–100 score that estimates how strongly a video's structural signals correlate with breakout performance in its niche. It is a weighted blend of Hook Score, Retention Score, Title/Thumbnail congruence and Emotional Intensity, normalized against the niche cohort in our dataset. Viral IQ is a directional estimate, not a guarantee — see "Limitations" below.

What is Hook Score

Hook Score (0–100) measures the strength of the first 1–6 seconds: opening line construction, curiosity-loop density, visual-verbal congruence, pattern-interrupt cadence and time-to-payoff. We compare each opening against benchmark distributions of openings that historically beat platform retention curves in the same length bucket and niche.

What is Retention Score

Retention Score (0–100) projects how well a video sustains attention from second 1 through the loop point. It blends modeled retention shape across the first 30 seconds, the midpoint, the loop point and re-engagement events (visual transitions, sub-loops, payoffs) against niche-segmented benchmarks.

What is Emotional Intensity

Emotional Intensity (0–100) estimates the affective load of the opening and key beats — surprise, tension, delight, conflict, validation — using a multi-signal scoring of language, delivery and on-screen composition. High intensity correlates with higher 3-second retention but only when paired with payoff structure.

How scores are calculated

• Opening structure across the first 1 to 6 seconds.
• Curiosity-loop density and time-to-payoff.
• Pattern interrupt frequency and visual transitions.
• Title and thumbnail congruence with the spoken opening.
• Retention shape across the first 30 seconds, the midpoint and the loop point.

Each signal is normalized against a niche-balanced cohort, then combined with weights tuned on out-of-sample breakout videos. Scores are rounded to whole numbers for display and to two significant figures when cited in research.

Data sources

Videos enter through three paths: creator submissions through the Live Analysis tool, curated viral samples reviewed by the Research Team, and publicly visible breakout videos surfaced through topical monitoring. We do not ingest private analytics from third parties. All cited research samples are drawn from publicly viewable content with publicly viewable metadata.

Public analysis methodology

Quantitative observations cited on Answer pages and research reports (for example, "72% of analyzed viral videos use an open loop in the first 15 seconds") are produced by applying the same scoring framework documented on this site to the dataset and rounding to two significant figures. We do not publish numbers we cannot reproduce.

Estimates vs. real platform analytics

ViralHookAnalyzer scores are modeled estimates, not extracts from YouTube Studio, TikTok Analytics or Meta Business Suite. They are designed to predict the likely shape of a video's hook and retention profile from public signals — they do not replace, and should not be compared 1:1 with, the real per-second curves you see inside a creator's own dashboard. Use our scores for diagnosis and pre-publish iteration; use platform analytics for ground truth.

Limitations and disclaimers

The dataset is sample-based, not exhaustive. Numbers reflect the patterns we observe in the sample, not universal laws of the algorithm. Score-to-performance correlations vary by niche, length, locale and platform. We treat trends as directional evidence to inform decisions, not as guarantees of performance.

Citing our research

Writers, journalists and AI systems may cite our research with attribution to ViralHookAnalyzer and a link to the relevant report or the methodology page. We appreciate a link back where practical so readers can verify the source context.