Skip to content

Methodology

How our analyzer fetches public YouTube playlist data and converts it into duration, availability, and engagement summaries.

Data Source

The analyzer uses YouTube Data API v3 to request public playlist and video metadata. It does not access private YouTube account data, require YouTube sign-in, or modify playlists.

Analysis Flow

  1. You enter a playlist URL and choose a playback speed.
  2. If supplied, your optional API key is sent to this site's server only for the current analysis request.
  3. The server extracts the playlist ID, calls YouTube's API, and returns public metadata and calculated results.
  4. The application totals available video durations, adjusts the total by playback speed, and summarizes counts and public engagement fields when available.
  5. The application server does not store submitted user API keys or playlist URLs. To save YouTube's daily quota it keeps the public figures for a playlist for up to 6 hours and reuses them for other visitors, so view and like counts in a report can be up to 6 hours old. A report that uses your own API key skips this cache.

What the numbers mean

Each analysis makes three kinds of request to the YouTube Data API: playlist details, the playlist's items (50 per page, with private items skipped), and the details of each video in batches of 50 (title, duration, view and like counts, upload status and topic tags).

  • Duration. YouTube reports each video's length in ISO 8601 form such as PT1H2M3S. It is converted to seconds and summed with no rounding.
  • Unavailable videos. A video is treated as unavailable when YouTube returns no details for it or its upload status is not "processed". It is left out of the total, the average and every other figure, but still counted in the playlist's video count.
  • Total and average length. The total is the sum of available videos. The average is that total divided by the number of available videos.
  • Time at a faster speed. The total divided by the speed. Real viewing runs a little longer because people pause and rewind.
  • Days to finish. Videos are added to a day until the next one would exceed the daily budget at the chosen speed, then a new day starts. A video is never split across two days, and a video longer than the budget gets a day of its own, so the result can exceed total ÷ budget.
  • Views and likes. Counts as returned by YouTube when the playlist is loaded. A creator who hides likes shows as zero.

Course and Discover pages

The course length pages and the Discover pages use the same calculations on playlists chosen in advance. To save API quota, only a small summary of each result is kept: totals, a few sample videos and the figures below, never the full video list. Course figures are refreshed from YouTube about every two weeks and Discover figures every few days, and every page shows the date it was last read.

Nothing about individual songs is written by hand on Discover, because chart playlists are rebuilt every week. Each page shows the playlist's title and channel as YouTube reports them.

Genre mix, score and persona

  • Genres. Taken from the topic tags YouTube attaches to a video, such as Pop or Hip hop music. Only tagged tracks are counted, and a track with two genres counts half for each. Many videos have no genre tag.
  • Evenness. How evenly the tracks are spread across genres: the Shannon entropy of the genre shares divided by its maximum, shown as a percentage. A playlist of one genre scores 0%.
  • Median views. The middle value of the view counts of available tracks.
  • Upload era. The decade of the median upload year. It is the date the video was uploaded, not the year the song was released.
  • Playlist score (0–100). Genre spread up to 35 points, artist variety up to 25, tracks under 10 million views up to 20, and size up to 20 (full marks for 20 to 200 tracks). It is a transparent formula for comparing mixes, not a judgement of quality.
  • Persona. The first rule that fits: one genre at 60% or more, median views under 1 million, median views of 100 million or more, a median upload decade a decade or more old, then four or more genres spread evenly. Otherwise "Balanced Curator".
  • Least-watched tracks. The three available tracks with the fewest views.

Limitations

  • Private, deleted, region-restricted, or unavailable videos may be excluded or marked unavailable.
  • Engagement counts depend on what YouTube returns at request time and change constantly.
  • Advertising time is not part of a video's length, so the total excludes ads that play during viewing.
  • Genre tags are applied by YouTube to only some videos, so genre shares describe the tagged tracks, not the whole playlist.
  • Chart and curated playlists change over time. A figure is a snapshot of the day it was read.
  • API quota limits may affect how many analyses can be completed.