Paste two or three channels. Each one is measured on videos from the same window, old enough to have collected their views, so no channel is judged on last week’s uploads against another’s from last spring. Every number comes straight from the YouTube Data API.
Putting channels on the same window
Paste up to three channels in whatever form you have at hand: an @handle, a link to the channel page, a channel ID that starts with UC, or a link to any video the channel published. Old /c/ links are the one format that can’t be used, because the YouTube API has no way to look them up, and a link to one of the channel’s videos is the quickest way around that.
The window chips decide which uploads count. With the default setting a video qualifies only if it went up between 14 and 90 days ago, so a channel that posted yesterday isn’t judged on a video still collecting its first views, and a channel that posts weekly isn’t measured over a different stretch of time than one that posts monthly. The conditions block under the toolbar puts those two dates in large type and lists how many videos each channel brought into the comparison. Three is the minimum for a median, and anything under five is marked as a small sample. If a channel falls short, one button widens the window to 180 days for every channel at once, so nobody ends up compared on a longer stretch than the others.
Subscriber counts on the channel cards appear as a range. YouTube publishes them rounded down to three significant figures, so 1.24M stands for anything from 1,240,000 to 1,249,999, and the line under the cards gives the gap between two channels as a range too. When two channels show the same rounded number, the tool says their order is unknown instead of picking one.
The plot
Every point is one video, placed by its views on a log scale, so the step from 10K to 100K takes the same space as the step from 100K to 1M. Circles are videos up to 60 seconds, diamonds run 61 to 180 seconds, and squares are longer than 180 seconds. The horizontal line in each column is that channel’s median. Points packed close around the line mean steady results from one upload to the next, while one point far above a loose cluster means a single video is doing most of the work. The ringed point is each channel’s top video, and tapping it shows how much of the window total it holds and what the total would be without it. Any point also tells you whether it sits above or below the other channels’ medians.
Exclude top video sets that point aside for every channel and recalculates everything below it, which is the quickest way to see whether a lead survives without its best upload. Blind compare swaps names, avatars and video titles for letters, for when you want to judge the shapes before the brands. The length chips narrow the comparison to videos up to 60 seconds, up to 180 seconds, or over 180 seconds. They are named by length rather than as Shorts and long-form on purpose, for the reason given further down.
Under the plot
Typical upload is the median views per comparable upload. With five or more uploads it is followed by the middle half, the range that holds the central half of a channel’s videos: a narrow middle half means most uploads land near the median, a wide one means the median hides a large spread. The next figure is the median recalculated without the top video, with the change in percent, so you can see how much the typical result leans on one upload without switching anything. Window total adds every comparable upload together, and the bar shows how much of it came from the top video. A large total next to a modest median usually points to one breakout rather than consistently strong uploads.
The next block counts how many of one channel’s videos land above the other channel’s median and how many below, in both directions, with any video on exactly the same number counted separately. It is a plain count you can check against the dots on the plot, not a significance test. When each median falls inside the other channel’s range of views, the tool notes that the gap between them is smaller than the spread within either channel.
Upload rhythm uses every upload in the window, including the newest ones the views comparison leaves out. The typical gap is the median number of days between uploads, the small chips list every gap in order, and the longest one is highlighted, so a channel that posts every few days and then goes quiet for a month doesn’t look as regular as an average would make it seem.
The length breakdown keeps 61 to 180 seconds as its own group. Since October 15, 2024, square and vertical videos up to three minutes count as Shorts, but the API reports a video’s length and not its shape, so a two-minute upload could be either one.
What differs turns the main contrasts into sentences: the higher typical upload, the more frequent uploader, the heavier reliance on one video. Read with care lists whatever weakens the comparison, such as very different sample sizes, a channel younger than the window, a hidden subscriber count, or a channel that uploads so often that only its latest 200 videos could be loaded. Copy comparison puts the whole result into plain text with the window, the time the data was fetched and every note, so whoever reads it later sees the same conditions you did.