Intelligence for YouTube creators

Understand what your channel is telling you.

Algorithm Whisperer uses your authorized read-only YouTube data to compare ideas with your own history, build creator-specific forecasts, lock experiments before the outcome is known, and keep independent public-web algorithm research in a separate evidence layer.

Read-only YouTube accessCreator-private analysisIndependent public research

Built around creator decisions

Turn channel history into better next decisions.

AW is designed to turn a creator's own history into questions that can be checked before and after publishing.

Before you make it

Does this idea resemble videos your audience has responded to before—and what historical evidence is actually comparable?

Before you publish

What does your own channel history suggest about the likely range, confidence, and biggest uncertainty?

After it goes live

How is the release tracking against your creator baseline and the forecast that existed before the result?

After the experiment

What did the outcome teach you, and which claims deserve another test instead of a confident story?

Three separate layers

A clear boundary between YouTube data, AW analysis, and public research.

01 · Your YouTube data

Authorized and read-only

AW reads the channel, video metadata, and private Analytics that you authorize through Google. It does not edit videos, titles, thumbnails, or channel settings.

02 · AW analysis

Creator-private calculations

AW builds baselines, historical comparisons, forecasts, and experiment evidence from your authorized data. These outputs are labeled as AW-created analysis, not YouTube-provided metrics.

03 · Algorithm Intelligence

Independent public research

AW separately evaluates publicly available sources and reported evidence about YouTube distribution. Private creator Analytics are not converted into platform-wide claims.

Clear provenance

Algorithm Whisperer does not have access to YouTube's private recommendation systems or hidden recommender state. AW-created forecasts, scores, comparisons, and experiment conclusions are third-party analysis based on authorized creator data; they are not metrics, judgments, or recommendations supplied or endorsed by YouTube.

Product

From an idea to evidence you can inspect.

Preflight

Compare a proposed video with your own historical analogues before publishing.

Your Algorithm

See patterns observed in your channel history while keeping retrospective evidence separate from prospective tests.

Experiment Lab

Freeze a creator question before the result exists, then track the outcome without rewriting the prediction afterward.

Algorithm Intelligence

Follow independently researched public evidence, reported changes, myths, and unresolved questions about YouTube distribution.

Data & permissions

AW requests only the Google access needed for its creator analytics workflow

youtube.readonly

Identifies the authorized YouTube channel and reads channel/video metadata required for the creator catalog and historical comparisons.

yt-analytics.readonly

Reads the authorized creator's private YouTube Analytics for channel history, performance evidence, forecasts, checkpoints, and experiments.

openid

When withdrawal controls are active, AW uses OpenID identity only to derive a one-way account-level grant key so the correct Google authorization can be safely grouped and revoked. AW does not persist the raw Google account identifier.

Third-party processing boundary

AW does not send your private YouTube Analytics reports or performance metrics to OpenAI. OpenAI is used separately for public-web Algorithm Intelligence research and, only after an explicit in-product opt-in, for semantic Preflight text processing using creator-entered proposal text plus current public title/description/tag metadata. See the Privacy Policy for the exact data flow.

You remain in control. You can revoke Google access through your Google Account permissions and can request deletion of the YouTube channel data stored by AW from Settings. Revoking authorization and deleting AW-stored data are related but distinct controls.

Make the next decision with more evidence than the last.

Start with your own evidence, keep uncertainty visible, and turn each prediction and experiment into something the next decision can learn from.