100K views on a 10M subscriber channel is a flop. 100K on a 5K channel is a hit. Video Trends scores every video against what its own channel normally does, so you see the real breakouts instead of the biggest names.
Free. No credit card.
Live data from the running instance. ◆ marks values Video Trends computes.
The problem
Knowing that MrBeast pulled 100M views teaches you nothing you can act on. What matters is the 5K subscriber creator whose latest upload did 8x their usual numbers, because their topic, title or thumbnail found something you can borrow while it is still working.
Normalized per channel: every video is measured against the median of its own creator
Age adjusted: yesterday’s upload is not punished against one from last month
Shorts kept separate: two formats with different scales, never mixed into one ranking
ranked by views
The breakout is invisible, buried under sheer channel size.
ranked by outperformance
It ranks first. It is the one with something to teach you.
How it works
Paste a URL, an @handle or a channel ID. Video Trends pulls the channel history and computes its reference median, separately for Shorts and long form.
Counters are read and timestamped every two hours. The cadence follows video age: dense over the first 48 hours, then progressively wider.
Filter by 24 hours, 7 days, 30 days or a custom range. Sort by outperformance, velocity, acceleration or engagement. Narrow down to a single topic.
What you get
Is this video doing better than it should be? Everything below exists to answer that one question honestly.
Every video is compared to its own channel median, corrected for what a video normally accumulates at that age. A six hour old upload and a three week old one become comparable.
No YouTube API returns view history. Video Trends builds its own time series, which is what makes it possible to see a video picking up speed rather than simply being old and large.
"AI", "ChatGPT" and "LLM" point at the same subject without sharing a word. Channel assignment, keywords and semantic similarity stack up so a topic keeps working as vocabulary shifts.
Their recent uploads, their historical best, and what stands out against it. Hundreds of channels fit comfortably inside the free API quota.
Proof
A sample of what the instance watches today
Questions
Video Trends uses the official YouTube Data API v3 and nothing else. No scraping, no password sharing, no access to your account. Raw API values are displayed as returned, and the metrics we compute are explicitly marked as ours, which is what the Developer Policies require.
No. Video Trends watches public channels, not yours. There is no YouTube authentication, so there is nothing to put at risk. The trade off is that owner only metrics such as retention, watch time and click through rate are out of reach, here and anywhere else.
Outperformance works from the first sync. Velocity needs two spaced measurements, acceleration needs three. Expect roughly 24 hours for solid signals, and a few weeks for medians that stop moving.
A hundred channels consume around 5 percent of the free daily quota. You can watch several hundred without approaching the limit, and current consumption is displayed live inside the app.
Nothing today. Video Trends is open while it is being built with its first users. If a paid plan arrives, it will be announced ahead of time.
View history cannot be backfilled. Every day you wait is a day of measurements you will never have. The clock starts when you add your first channel.
Free. No credit card.