Trends and tracking
How to do YouTube Shorts research with your Assistant
By Jackson Gathondu. Published 1 November 2026. Updated .
YouTube Shorts research means finding the Shorts that do well in your niche and working out why: what the first line says, what is on screen, how the Video is built and what viewers ask for in the comments. Done by hand, that is an afternoon of scrolling and a spreadsheet.
This guide shows how to hand the work to your Assistant, such as Claude or ChatGPT, connected to ShortsIntel. You describe the niche, the Assistant runs a Research on YouTube, keeps the Shorts and reads you the patterns. It also lists where YouTube data stops: YouTube returns fewer fields than TikTok, and long-form Videos come back alongside Shorts.
What a YouTube Shorts Research gives you
A Research takes a plain-language goal, turns it into search phrases, collects matching Videos, drops the ones that are off topic and analyses the rest. Each Video comes back with its views, length, post date and an analysis: hook type, opening line, visual hook, format, on-screen text, tone and a transcript.
On top of the Videos, the Research writes a Digest. It groups the Videos into themes, names the formulas that pull above the niche's median views, lists hook templates you can reuse, and pulls audience gaps from the comments: questions viewers keep asking that no Video answers yet.
A Research costs 1,036 credits for up to 50 analysed Videos and 20 credits for each Video after that. Reading a finished Run back is free, as often as you like.
How to research YouTube Shorts in a niche
You type these as normal chat messages. The Assistant picks the tools.
Connect your Assistant
Add ShortsIntel to Claude, ChatGPT, Claude Code, Codex or Cursor and sign in. The MCP page for YouTube has the steps for each one.
Name the niche and say YouTube only
Describe the niche the way you would to a colleague, and ask for YouTube alone. For example: "Research B2B SaaS product marketing on YouTube only, 50 Videos." A Research that names several Platforms picks the strongest Videos across all of them, so YouTube can end up with few or none.
Ask for an estimate first
Ask the Assistant to quote the price before it runs. The quote is free and nothing is charged until you say go.
Keep the Shorts
YouTube search returns long-form Videos as well as Shorts, so the Research does too. Ask the Assistant to keep only Videos of 180 seconds or less, using each Video's duration_seconds field.
Read the hooks, formats and gaps
Ask for the opening line, hook type, format and on-screen text of each Short, then for the Digest's audience gaps. Each gap comes with the comment it was taken from.
Check the channels behind the best Shorts
Look up the channel of a Short that did well. A Creator lookup on YouTube returns the channel's Shorts with view counts and an outlier figure against the channel's usual views.
Data limits
YouTube data limits
- Creator data covers Shorts only. A channel lookup returns the channel's Shorts, not its long-form uploads.
- The Shorts in a channel lookup come without post dates or lengths, so you cannot sort them by recency.
- A Research and a Video search on YouTube return long-form Videos too. Filter by duration_seconds to keep the Shorts.
- Likes, comments and shares show as 0 on YouTube Videos in a Research or a channel lookup, and follower counts on searched Videos are empty. Views are the number to compare.
- A single Video lookup takes Shorts and long-form. Analysis and download work on Videos up to 10 minutes long.
- There are no Hashtags or Sounds on YouTube, and score_video is TikTok only; Scores still come back on a channel's Shorts through get_creator. Creator Trackers do not run on YouTube yet; Keyword Trackers do.
Real output from a YouTube Research
These are real responses from our own account, trimmed to fit. Your Assistant gets the same JSON and writes its answer from it.
YouTube, via REST API
Research B2B SaaS product marketing on YouTube only, then show me one of the Shorts with its hook and on-screen text.
{
"id": "7nehg0yjU70",
"url": "https://www.youtube.com/watch?v=7nehg0yjU70",
"platform": "youtube",
"rank": 32,
"duration_seconds": 59,
"posted_at": "2026-10-02T07:10:17.286Z",
"creator": { "name": "TriNet Studios", "handle": "trinetstudios" },
"snapshot": { "views": 726, "likes": 0, "comments": 0, "shares": 0 },
"intelligence": {
"hook": {
"type": "PROBLEM",
"opening_line": "Tuesday? Can't. Thursday? What time zone? Okay.",
"visual_hook": "fast-paced montage of conflicting calendar apps and chat windows",
"rationale": "It immediately highlights a relatable scheduling pain point that target professionals experience daily."
},
"visual": {
"format": "PRODUCT_DEMO",
"on_screen_text": [
"Tuesday? Can't. Thursday? What time zone?",
"Stop Playing Calendar Tetris.",
"Cause There Is Another Way..",
"One link. Everyone's Availability. No Back-And-Forth.",
...
]
},
...
}
}
(49 Videos analysed, 6 of them 180 seconds or less; tone, transcript and safety trimmed)YouTube, via REST API
What are viewers asking for in the comments that no Video answers yet?
{
"gaps": [
{
"ask": "Viewers want to know how to market high-ticket, low-frequency B2B products.",
"angle": "A breakdown of marketing strategies specifically tailored for high-ticket, enterprise-level B2B SaaS with long sales cycles.",
"comments": [
{
"text": "this video is fantastic. thank you! curious about how this approach works for b2b products that are low frequency, high price (six figures), ...",
"video_id": "youtube:KzS5Qs7vSMg"
}
]
},
{
"ask": "Founders are afraid of the costs associated with paid marketing and want low-budget alternatives.",
"angle": "A guide on zero-budget, organic go-to-market strategies for bootstrapped SaaS founders.",
...
},
...
]
}
(5 gaps in all, each with the comment it came from)YouTube, via REST API
Look up the MrBeast channel and list its recent Shorts with how they did against its usual views.
{
"platform": "youtube",
"handle": "MrBeast",
"follower_count": 524000000,
"baseline": {
"source": "follower_tier",
"confidence": "low",
"median_views": 600000,
"p90_views": 1800000
},
"videos": [
{
"description": "Can A Robot Do My Chores?",
"duration_seconds": null,
"posted_at": null,
"snapshot": { "views": 45000000 },
"scores": { "outlier": 75, "outlier_p90": 25 }
},
{
"description": "How People Treat You Based on Subscribers",
"snapshot": { "views": 32000000 },
"scores": { "outlier": 53.33, "outlier_p90": 17.78 }
},
...
],
"next_cursor": "eyJwIjoiWU9VVFVCRSIs..."
}
(48 Shorts on the first page, no long-form uploads)How to read the output
Expect long-form Videos in the mix. In the B2B SaaS Research above, 6 of the 49 analysed Videos were 180 seconds or less. The rest were talks, demos and walkthroughs that YouTube search returned for the same phrases. In a niche like this one, ask for more Videos, up to 200, so enough Shorts are left once you filter.
Compare Shorts by views. YouTube search does not give likes, comments or follower counts for these Videos, so an engagement rate cannot be worked out. The Digest says so when it explains a formula.
Read the dropped list. The Research checks every Video against your goal and drops the ones that miss. This one dropped 16 as off topic, such as a spice company success story and a Netflix case study, and the response lists a sample of them with the reason. If too many good Videos are dropped, make the goal more specific.
Treat a channel's outlier figures as a rough guide when the baseline says low confidence. A big channel's baseline comes from its follower tier rather than from its own Videos.
Name YouTube on its own when YouTube is what you want. In two of our test Runs that named TikTok and YouTube together, on mobile games and home coffee, all 49 analysed Videos came from TikTok.
YouTube Shorts research prompts to try
Start with a quote, then run it. New accounts get 1,000 credits free, valid for 90 days. A Run that fails costs nothing, and a partial one is charged only for what finished.
- Quote a Research on meal prep Shorts on YouTube only, 100 Videos, then run it and keep the ones under three minutes.
- From that Research, list the five most viewed Shorts with their opening line, hook type and format.
- Write three hook templates from the Shorts that beat the niche median, and say which Video each came from.
- What do viewers ask for in the comments that no Short in the Research answers?
- Look up the channel behind the top Short and tell me which of its Shorts did best against its usual views.
- Analyse this Short and write out its hook, on-screen text and call to action: (paste a link)
Questions
- What is YouTube Shorts research?
- It is finding the Shorts that do well in a niche and working out why: their hooks, formats, on-screen text and the questions viewers ask in the comments. You use what you find to plan your own Shorts or briefs.
- Can I research only Shorts and leave out long-form Videos?
- Not in the search itself. YouTube search returns both, so a Research does too. Each Video carries duration_seconds, so ask your Assistant to keep Videos of 180 seconds or less. Channel lookups are different: they return a channel's Shorts only.
- Why are likes and comments 0 on YouTube Videos?
- YouTube gives view counts but not likes, comments or follower counts for these Videos, so they come back as 0 or empty. Compare Shorts by views, and read a Video's comments with a comments lookup if you need them.
- Can I track a YouTube channel's new Shorts?
- Not yet. Creator Trackers run on TikTok, Instagram and Facebook. On YouTube you can track a Keyword daily, and look a channel up again whenever you want fresh numbers.
- What does a YouTube Shorts Research cost?
- 1,036 credits for up to 50 analysed Videos and 20 credits for each Video after that, up to 4,036 credits at 200 Videos. Reading the finished Research, its Videos and its Digest again is free. A channel lookup is 6 credits.
Try it in your Assistant
Connect ShortsIntel to the Assistant you already use and ask the question this guide answers, or try the calls in the Playground first.