Skip to content
BuildTuber

Glossary

The terms that come up repeatedly in the guides, defined once here so the guides do not have to stop and define them again. Each entry says what the term means, gives an example, and explains why it matters to someone running a channel rather than to someone building models.

Where a definition is contested or the number behind it moves, the entry says so instead of picking a figure and stating it as settled.

YouTube Growth

Audience retention
How much of a video people actually watch, reported by YouTube Analytics as both a curve across the runtime and an average.
Average view duration
The average length of time viewers spend watching a video, reported by YouTube Studio in minutes and seconds rather than as a percentage.
Click-through rate
The share of YouTube impressions that turned into views: how often people who saw the thumbnail chose to click it.
Demonetisation
The loss, or denial, of a channel's ability to earn ad revenue, usually following a platform review of whether its content meets monetisation policy.
Faceless YouTube channel
A channel that publishes without showing a presenter on camera, carrying the video with voiceover, footage, graphics or animation instead.
Impressions
The number of times YouTube showed a video's thumbnail to a viewer, whether or not that viewer clicked it.
Outlier video
A video that performed far above the normal range for its channel, usually expressed as a multiple of that channel's recent median views.
Returning viewers
The share of a video's views that came from people who have watched the channel before, reported in YouTube Studio's Audience tab alongside new viewers.
Reused or inauthentic content
A YouTube monetisation policy category covering channels whose videos are judged to be repetitive, duplicated, or lacking meaningful original contribution.
RPM
Revenue per mille: a channel's estimated ad revenue per 1,000 monetised views, YouTube's headline earnings-rate figure.
Sleep documentary
A long-form faceless video, usually 60 to 120 minutes, narrating a true story or history over calm visuals, designed to be watched or half-watched at bedtime.
Watch time
The cumulative hours viewers have spent watching a channel's videos, the metric behind monetisation eligibility and much of YouTube's recommendation system.
YouTube packaging
The title and thumbnail treated as a single unit: the entire promise a viewer sees before deciding whether to click.

Cost & Economics

BYOK
Bring Your Own Key: running a tool against your own account with an AI provider, so the provider bills you directly for usage instead of the tool's vendor reselling it to you.
Cost per image
What a provider charges to generate a single still image, the smallest billable unit in an AI-assisted visual pipeline.
Inference cost
What a provider charges you to actually run a model, billed per unit of work done, not per month.
Token
The unit a language model reads and writes in, and the unit providers bill by. Roughly a short word or a fragment of a longer one.
Vendor lock-in
The cost of leaving a tool, measured in work you would have to redo rather than in money you would have to pay.

AI Tools & Workflow

API key
A secret string that identifies your account to an AI provider and authorises charges against it.
Apps Script
Google's built-in scripting language for automating Sheets, Docs, and other Workspace apps, the layer that wires a spreadsheet to an external API.
Context window
The maximum amount of text, measured in tokens, that a model can hold in view for a single request.
Google Sheet
A spreadsheet used as the interface and lightweight database for a DIY production stack, usually wired to an API through Apps Script.
Image generation model
An AI model that turns a text prompt into a still image, the model layer behind AI-generated visuals in a production stack.
Production stack
The full set of tools, models, prompts and connecting logic that turns an idea into a finished video.
Stock footage
Pre-shot video clips licensed for reuse, used in a faceless video as an alternative or complement to AI-generated visuals.
Video generation model
An AI model that generates a video clip directly from a text prompt or a still image, as distinct from a still-image model or licensed stock footage.

Get the build guides as they go out.

One email when a new guide ships. The full method, not a teaser. Unsubscribe whenever.