Understand the data behind the result.
Where our data comes from, what the metrics mean, and the limits to keep in mind when using TubeSaku.
Sources & coverage
TubeSaku combines selected public observations with its own aggregation and analysis. Sources have different populations, time windows, and units. A ranking on one platform is not directly comparable to a ranking on another.
| Dataset or feature | Source & interpretation |
|---|---|
| Google Trends | Japan-focused top and rising terms from Google Trends public datasets in BigQuery. Rankings describe the source’s selected terms, not all searches or exact search counts. |
| YouTube suggestions | Suggestions collected using single-character queries with Japanese language and Japan region settings. Frequency in this collection is not search volume. |
| Upcoming & live streams | Stored YouTube stream and video observations. Schedules and live status can change after collection; searches do not cover every channel. |
| VTuber discovery & game comparisons | TubeSaku’s stored channel records and weekly game aggregates for the covered population. Missing activity records do not establish that a creator has never played a game. |
| Cover-song candidates | Japanese YouTube music charts combined with TubeSaku’s observed cover uploads. Original-song popularity and recent cover supply are separate measures. These records do not establish music-use rights. |
| Other trend dashboard sources | Bluesky public trends, TikTok Creative Center hashtag observations, and Steam rankings. Check the displayed dates, region, and metric for each source. |
Inspect the data quality catalog (JSON) · Read the public dataset index
TubeSaku’s aggregation, rankings, and game classifications are its own analysis. The named platforms are data sources; listing them does not imply their endorsement of TubeSaku.
Freshness & update timing
- Observation time or period describes when the underlying activity was observed.
- Generation time describes when an export or response was produced. It does not prove the source was collected at that time.
- Freshness and missing-data notes describe whether observations are recent, stale, absent, or unknown.
Sources update at different intervals and may publish with a delay. A newly generated file can contain older observations. Estimated next-update times are estimates, not guarantees. Where exact collection timing is unavailable, the catalog records that limitation.
MCP searches generally use TubeSaku’s saved datasets, public APIs, and generated documents. They do not trigger a new search of Google or YouTube for every request. Live-stream matches reflect the latest stored observation within the permitted window.
Measurements, analysis & examples
Source observations, calculated measures, and interpretations serve different purposes. A view count is an observation; a ratio such as views divided by subscribers is a derived measure; an explanation of why a topic may suit a creator is an interpretation.
Our landing-page dashboard illustrations are labeled as examples. Their charts, percentages, and fit scores are illustrative values and are not current measurements or validated predictions. Consult the linked datasets or research reports for actual observations.
AI-assisted tools can propose, summarize, or classify information. They can make mistakes. Research findings should be assessed using the report’s methods and supporting data. Do not assume that every API row has been manually reviewed.
How to read our research
Each study defines its own sample, observation period, exclusions, calculations, and limitations. The methods for one report should not be assumed to apply to every other dataset.
- Check the number of channels, videos, or observations actually included in each comparison.
- Distinguish changes within the same channels from differences between groups.
- Account for platform measurement changes, missing records, and selection bias.
- Treat observational associations as associations; they do not establish cause and effect.
For example, our VTuber hiatus report documents revised comparisons that exclude windows spanning a YouTube view-counting change. Its correction notes and downloadable tables explain the revised analysis.
Corrections, attribution & contact
To report a suspected error, send the page URL, the relevant statement or metric, and any supporting evidence through our contact form. Existing report-specific correction notes remain part of those reports; consult their published and updated dates when citing them.
Cite the relevant TubeSaku report or dataset and its observation period. Public access does not imply unrestricted redistribution; follow the terms of use and any report-specific instructions.
Our About page identifies the operator and research team. Our privacy policy explains usage logging and contact submissions.