Transparency

How scores work.

HookSignals scores are directional signals based on publicly available data and packaging analysis. They are designed to help creators identify weak signals before publishing — not to predict or guarantee performance.

How scores work

What HookSignals scores — and what it doesn't.

Data sources

  • ◆YouTube Data API
  • ◆Public video metadata
  • ◆Title structure analysis
  • ◆Hook pattern analysis
  • ◆Packaging analysis

Scores are

  • ✓Directional guidance before you publish

Scores indicate which packaging elements are weak before publishing — not after.

Scores are not

  • ✗Guaranteed views
  • ✗Actual YouTube CTR
  • ✗Actual retention data

Actual CTR, retention curves and view counts are only available inside YouTube Studio.

Outlier Score — measured, not estimated

Unlike the AI-estimated packaging scores above, the Outlier Score is computed directly from public data. It answers one question: how is this video performing relative to its own channel's normal? A 50,000-view video on a channel that typically gets 10,000 views is a 5.0× outlier. The same views on a channel that averages a million is 0.05×. Most tools in this category show a number like this without saying how it's calculated — here is exactly how ours works:

  • ◆Baseline = the MEDIAN view count of the channel's recent uploads (up to 50), fetched live from the YouTube Data API. Median, not average — one earlier viral hit would otherwise inflate the baseline and make every later video look like a failure.
  • ◆The video being analyzed is excluded from its own baseline, and uploads younger than 7 days are excluded too (their views are still accruing and would drag the median down).
  • ◆Shorts are compared against the channel's Shorts and long-form against long-form, whenever at least 5 same-format uploads exist. If not, all uploads are used and the result says so.
  • ◆View velocity is reported separately: the video's views-per-day against the channel's median views-per-day. This lets a 2-week-old video be compared fairly against an older catalog.
  • ◆If a channel has fewer than 5 eligible uploads, we show nothing rather than a low-confidence number.

The Outlier Score describes how a published video has performed so far — it is a measurement, not a prediction, and it can change as views accrue. It uses only public view counts and upload dates; no channel login, no Studio data.

What the analyzer actually does

  • ◆Fetches public video metadata via the YouTube Data API — title, views, likes, duration and thumbnail URL.
  • ◆Analyzes title structure for clarity, curiosity gap, keyword placement and character length.
  • ◆Scores the hook and opening signals across 4 packaging dimensions.
  • ◆Returns improvement suggestions — alternative titles, hook rewrites, thumbnail text and a description angle.

What the analyzer does not do

  • ✗Does not access your YouTube account or YouTube Studio.
  • ✗Does not read private analytics — actual CTR, watch time or audience retention curves.
  • ✗Does not guarantee views, clicks or channel growth.
  • ✗Does not access private or unlisted video data.

How to use scores effectively

A high score does not guarantee a video will perform well. A low score does not mean a video will fail. Scores indicate which packaging elements — title clarity, hook strength, curiosity gap, keyword placement — are weak relative to known patterns. Use them as a pre-publish checklist, not a performance prediction.