YouTube Shorts Viral Trend Tracker

See what is actually going viral on YouTube Shorts, by hashtag, with the numbers you need to rank it.

Give it hashtags. It returns YouTube's own ranked grid of Shorts for each one, then opens each clip for the figures the grid does not carry: exact views, likes, the like-to-view ratio, duration, sound attribution and hashtags.

Shorts only — a regular upload that sneaks into a hashtag grid is flagged (isShortFormat: false) rather than quietly mixed in.

No login, no cookies, no API key.

What you get

{
  "hashtag": "cooking",
  "gridPosition": 1,
  "title": "Frog Eggs Time Challenge! ⏱️🐸🧪",
  "channelName": "Meow Nhi Nhi",
  "viewCount": 3397251,
  "likeCount": 36218,
  "likeToViewPercent": 1.0661,
  "likeCountIsApprox": false,
  "lengthSeconds": 52,
  "soundAttribution": "Original Sound",
  "hashtags": ["Mukbang", "buldak", "buldakchallenge", "cooking", "foodchallenge"],
  "url": "https://www.youtube.com/shorts/FX4RT1cp7Rs"
}

gridPosition is YouTube's own ranking for that hashtag — it is the trend signal, so it rides along on every record and enrichment never reorders anything.

Why the per-Short fetch is on by default

The hashtag grid alone cannot rank anything. Measured 2026-09-20:

	Hashtag grid	Short's own page
Views	rounded — "3.3 million"	exact — 3,397,251
Likes	none at all	yes
Like-to-view ratio	impossible	yes
Sound	no	yes
Duration	no	yes
Cost	1 request per hashtag	1 request per Short

Set detail: false if you only want the ranked list and can live with 3.3M-sized buckets.

Two kinds of like count

Surface	Like figure	Sound?
/shorts/<id> (default)	rounded — "36 thousand" → 36000	yes
/watch?v=<id> (exactLikes: true)	exact — 36,218	no

Every record carries likeCountIsApprox so a rounded figure is never presented as an exact one. If you ask for exactLikes and the Short's page has no sound attribution, the Actor spends a second request to go get it — that is the only path that costs two requests per Short.

Tracking growth over time

Run the same hashtags on a schedule. Each run stamps _scrapedAt, so two runs of #cooking give you the delta in viewCount, likeCount and gridPosition per videoId — which is the growth-rate signal. The Actor does not keep state between runs; comparing datasets is yours to do.

Known limits, stated plainly

• 36 Shorts per hashtag. That is one grid page, and it is what YouTube serves. The page carries a single continuation token; this Actor does not follow it. Ranked top-36 is the useful part of a trend anyway.
• No historical data. YouTube exposes the grid as it is right now. Growth requires scheduling (above).
• Sound attribution is often just "Original Sound". That is what YouTube reports for most Shorts; a licensed track shows its title instead.

Reliability

The hashtag grid measured unusually stable — nine fetches across three TLS profiles on 2026-09-20 returned 37 lockups and 36 view figures every single time. Individual Shorts pages behave like watch pages and occasionally arrive without their payload, so the Actor retries up to five times, rotating TLS profile, before emitting recordType: "REDUCED". A REDUCED record still carries its grid figures — you lose the exact numbers, not the row.

Input example

{
  "mode": "hashtag",
  "hashtags": ["cooking", "fyp", "asmr"],
  "shortsPerHashtag": 36,
  "detail": true,
  "exactLikes": true,
  "concurrency": 4
}