ShortsIntel

Creative and hooks

How to find outlier Videos in your niche

By . Published 1 November 2026. Updated .

An outlier Video is one that got far more views than its Creator usually gets. Two million views on an account that always gets two million says little. 50,000 views on an account that usually gets 2,000 means something in that Video worked, and it is worth finding out what.

This guide covers how to find those Videos in a niche, how to read the numbers that come back, and how to turn the outliers into briefs. It includes real output from our own account: a Video that got 1,340 times its Creator's median views.

What makes a Video an outlier

Raw views favour big accounts. Sort a niche by views and you get the same large Creators every time, whatever they posted. To find what worked, compare each Video with the Creator who posted it.

ShortsIntel does this with a Score. The outlier number is a Video's views divided by the Creator's median views. 1 means a typical Video for that Creator. 3 means three times their usual. Because it is a ratio, a 700-follower account's breakout and a large account's breakout can be compared directly.

The median comes from the Creator's last 20 Videos. Videos under 48 hours old are left out because their views are still climbing, and so is the Video being scored, so one huge post cannot raise the median it is measured against. With fewer than 5 usable Videos, the baseline falls back to a benchmark for the Creator's follower count and is marked low confidence.

How to find outlier Videos in a niche

Each step is a request you can type into your Assistant in plain language, or make through the REST API. Paid steps quote their price first if you ask for an estimate.

  1. Run a Research on the niche

    A Research searches your Platforms, drops off-topic Videos and ranks the rest. The ranking favours outliers: views against the Creator's usual reach, views per day, engagement and recency, with no more than two Videos from any one Creator. It analyses the top 50 for 1,036 credits, and reading it again later is free.

    Research budgeting app videos on TikTok and list the Videos that did best against their Creator's usual views.

  2. Look up the Creators behind the top Videos

    A Creator lookup costs 6 credits and returns the Creator's recent Videos, each with an outlier number, a percentile and views per hour against that Creator's baseline. This works on every Platform for Videos under 30 days old.

    Look up the Creators behind the top five Videos and show each one's Videos ranked by outlier.

  3. Score single TikToks

    For one TikTok Video, a Score costs 9 credits. It reads the Creator's recent Videos to build the baseline when we do not have one yet.

    Score this TikTok against its Creator's usual reach: https://www.tiktok.com/@budgetwithsam/video/7695092268104977694

  4. Analyse the outliers

    Video analysis returns the hook type, the opening line, the on-screen text, the format and the call to action for 23 credits per Video. Videos in a Research come back analysed already.

    Analyse the three biggest outliers and tell me what their openings have in common.

  5. Compare them with the Creator's usual Videos

    An outlier only teaches you something next to the Videos that did not break out. Ask what the outlier did that the same Creator's typical Videos did not: a different opening, a different format, a different topic.

    Compare the outlier with that Creator's Videos near 1x. What changed?

  6. Write briefs from the difference

    Turn each difference into a brief that names the hook, the format and the source Video, so the team can check it.

    Write two briefs that borrow the opening of the outlier, and link the source Video.

Real output: one outlier, its Creator and its analysis

All three samples come from our own account on the day this guide was written, trimmed to fit. The Video turned up in a Research on budgeting app Videos. Captions are left out where they contain emojis.

TikTok, via REST API

POST /v1/videos/tiktok/7695092268104977694/score

{
  "run": { "kind": "video_score", "status": "completed", "charge_credits": 9 },
  "result": {
    "video": {
      "id": "7695092268104977694",
      "url": "https://www.tiktok.com/@budgetwithsam/video/7695092268104977694",
      "posted_at": "2026-10-10T17:28:21.000Z",
      "duration_seconds": 6,
      "snapshot": { "views": 6500000, "likes": 585600, "comments": 1029, "shares": 20100, "saves": 15835 },
      "scores": {
        "outlier": 1340.76,
        "outlier_p90": 200.93,
        "percentile": 100,
        "velocity": { "window": "lifetime", "views_per_hour": 12296.2 },
        "baseline": {
          "median_views": 4848,
          "p90_views": 32350,
          "sample": 16,
          "source": "videos",
          "confidence": "high",
          "posts_per_week": 6.29,
          "engagement_rate": 0.02
        },
        "reason": null
      }
    },
    "creator": {
      "handle": "budgetwithsam",
      "snapshot": { "followers": 688 },
      ...
    }
  }
}
Captured 1 November 2026. Tools used: POST /v1/videos/tiktok/{id}/score, score_video. Output trimmed; live results change.

TikTok, via REST API

GET /v1/creators/tiktok/budgetwithsam?include=videos

{
  "handle": "budgetwithsam",
  "follower_count": 688,
  "video_count": 18,
  "baseline": { "median_views": 6445, "sample": 17, "confidence": "high", "posts_per_week": 6.29 },
  "videos": [
    {
      "id": "7695092268104977694",
      "posted_at": "2026-10-10T17:28:21.000Z",
      "snapshot": { "views": 6500000 },
      "scores": { "outlier": 1340.76, "percentile": 100 }
    },
    {
      "id": "7697246586857901343",
      "posted_at": "2026-10-16T12:48:27.000Z",
      "description": "i use milo to keep track of my spendings! #nyc #savingmoney #moneyhacks #genzfinance #softwareengineer ",
      "snapshot": { "views": 634500 },
      "scores": { "outlier": 130.88, "percentile": 93.75 }
    },
    {
      "id": "7696219186317561119",
      "posted_at": "2026-10-13T18:21:20.000Z",
      "snapshot": { "views": 6445 },
      "scores": { "outlier": 1.09, "percentile": 50 }
    },
    {
      "id": "7698742288851160351",
      "posted_at": "2026-10-20T13:32:39.000Z",
      "description": "i use milo to keep track of my spendings! #savingmoney #moneyhacks #genzfinance #nyc #softwareengineer ",
      "snapshot": { "views": 907 },
      "scores": { "outlier": 0.12, "percentile": 0 }
    },
    ...
  ]
}
Captured 1 November 2026. Tools used: GET /v1/creators/{platform}/{handle}, get_creator. Output trimmed; live results change.

TikTok, via REST API

POST /v1/research with intent "Budgeting app videos", platforms tiktok, then GET /v1/research/{id}/videos (the 1,340x Video)

{
  "id": "7695092268104977694",
  "intelligence": {
    "source": "captions_only",
    "hook": {
      "type": "CURIOSITY_GAP",
      "opening_line": "it only takes $27.397 a day to spend $10,000 a year. let that sink in.",
      "visual_hook": "young woman sitting in a dimly lit restaurant looking to the side",
      "rationale": "The text overlay presents a striking financial statistic that grabs the viewer's attention."
    },
    "visual": {
      "format": "VOICEOVER_BROLL",
      "face_visibility": "CREATOR_ON_CAMERA",
      "on_screen_text": ["it only takes $27.397 a day to spend $10,000 a year. let that sink in."]
    },
    "tone": { "speaking_style": "NO_SPEECH" },
    "persuasion": {
      "sponsored": true,
      "call_to_action": { "type": "DOWNLOAD", "text": "download milo rn. like RIGHT. NOW." },
      "brands_and_products": [{ "kind": "PRODUCT", "name": "milo", "prominence": "CENTRAL" }]
    },
    ...
  }
}
Captured 1 November 2026. Tools used: POST /v1/research, GET /v1/research/{id}/videos. Output trimmed; live results change.

What to take from that output

The account has 688 followers and posts about six times a week, almost every Video a pitch for the same budgeting app. Its median is under 5,000 views. One six-second Video reached 6.5 million, 1,340 times the median, and another reached 634,500. Half of its 18 recent Videos landed below the median.

The caption did not make the difference. The 130x Video and the 0.12x Video carry the same line, "i use milo to keep track of my spendings!", word for word. The analysis of the 1,340x Video shows its opening: no speech, one on-screen line with a number that sounds wrong at first ("it only takes $27.397 a day to spend $10,000 a year"), filmed in a restaurant. The on-screen line does not mention the app; the ask to download it sits in the caption.

A brief from this Video would say: "open on a spending number the viewer has to stop and check, keep the product out of the on-screen opening, and put the ask in the caption". We have analysis for this one Video only, so analyse the 130x Video and a few near 1x before you treat the opening as the cause. Then test the hook across several Creators.

Two more checks before you brief it. The outlier_p90 of 200.93 means the Video got 200 times the views of this Creator's 90th-percentile Video, far above their usual best. The baseline is marked high confidence with 16 Videos behind it, so the median is measured, not estimated.

How to read a Score

Every scored Video carries the same fields. These are the ones to check before you call a Video an outlier.

FieldWhat it meansWhat to look for
outlierViews divided by the Creator's median views.Above 1 beat their usual. Pick your own line; 2x or more is a common place to start reading.
outlier_p90Views divided by the Creator's 90th-percentile Video.Above 1 means the Video beat even the Creator's best tenth, not just a normal week.
percentileWhere the Video sits among the Creator's own Videos, 0 to 100.100 is their best recent Video.
velocityViews per hour since posting.A modest outlier with high velocity is still climbing. Velocity near zero means the Video has stopped.
baseline.confidence and sourcehigh with source videos means the median was measured from the Creator's own Videos; low with follower_tier means it was estimated from follower count.Treat a low-confidence outlier as a reason to look, not as evidence.
reasonaged_out when the Video is older than 30 days.Score Videos while they are young; older ones come back without numbers.

How common outliers are

Across 48 TikTok Creators we looked up on 2026-11-01, 127 of their 619 recent scored Videos (20.5%) got at least twice the Creator's median views, and those Videos carried 58.8% of the views. That leaves out the 44 Videos Research had picked, which would inflate the figures. The full breakdown by niche is on the outlier Video statistics page.

Data limits

Data limits for outlier Videos

  • Scoring a single Video on request (score_video) is TikTok only. On Instagram Reels, a Creator lookup (get_creator) returns the same Scores for each recent Video.
  • Scores cover Videos up to 30 days old. Older Videos come back with reason aged_out and no numbers.
  • Search results are never scored. Score the Videos you care about afterwards, or use a Research, which ranks by reach against the Creator's usual views.
  • YouTube Creator data covers Shorts only, not long-form uploads. In our capture the Shorts came back without posted dates, so the baseline fell back to a follower-size benchmark marked low confidence. Treat YouTube outlier numbers as estimates.
  • Video analysis takes YouTube Videos up to 10 minutes long.
  • Sounds are TikTok only. Hashtags are TikTok and Instagram only.
  • Facebook Video search is opt-in and best effort. Instagram and Facebook Keyword search are best effort too.
  • Refreshing a Research is TikTok only.
  • Creator Trackers run on TikTok, Instagram and Facebook, not YouTube.

What it costs

Each request that does work is a Run, paid from a prepaid Balance of Credits at 1,000 credits per US dollar. There is no subscription. A Research costs 1,036 credits for up to 50 analysed Videos, a Creator lookup with Scores 6 credits, a TikTok Score 9 credits and a single Video analysis 23 credits. 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. Every price is on the pricing page.

To keep watching a Creator whose Videos keep breaking out, set a Creator Tracker. Your Assistant then gets a Change Report when a new Video takes off.

Connect your Assistant

The same server works in every Assistant below. You sign in to ShortsIntel and choose which organization pays; no API key is needed.

Claude.ai

Web and desktop. No API key: Claude signs in to ShortsIntel for you.

  1. Open Claude's add connector page with ShortsIntel filled in, and choose Add. Or go to Settings, Connectors, Add custom connector, and paste the endpoint below.
    https://www.shortsintel.com/api/mcp
  2. Sign in to ShortsIntel, approve the connection and choose the organization whose Balance pays for Runs.
  3. In a chat, turn the ShortsIntel connector on and ask your question.

ChatGPT

Web and desktop, in Developer mode. No API key: ChatGPT signs in over OAuth.

  1. Open ChatGPT's connector settings and switch Developer mode on.
  2. Create a connector named ShortsIntel, paste the endpoint and choose OAuth.
    https://www.shortsintel.com/api/mcp
  3. Sign in to ShortsIntel and choose your organization.
  4. In a new chat, choose the + next to the message box, then Developer mode, and turn ShortsIntel on.

Claude Code

One command, then sign in from Claude Code.

  1. Add the server in your terminal.
    claude mcp add --transport http --scope user shortsintel https://www.shortsintel.com/api/mcp
  2. Run /mcp inside Claude Code, choose shortsintel and sign in.

Codex

One command; Codex signs in on first use.

  1. Add the server in your terminal.
    codex mcp add shortsintel --url https://www.shortsintel.com/api/mcp
  2. If no browser window opens to sign in, start it yourself.
    codex mcp login shortsintel

Cursor and VS Code

One-click install links; each asks you to confirm.

  1. In Cursor, open the Cursor install link.
  2. In VS Code, open the VS Code install link.
  3. Sign in to ShortsIntel when the editor asks.

Questions

What is an outlier video?
An outlier video is one that got far more views than its Creator usually gets. ShortsIntel measures it as the Video's views divided by the Creator's median views, so 5 means five times their usual.
How do I find viral videos in my niche?
Run a Research on the niche. It ranks Videos by how far they beat their Creator's usual reach rather than by raw views, then analyses the top ones. Look up the Creators behind the top Videos to see each one's outliers next to their typical Videos.
Why not just sort by views?
Sorting by views returns the largest accounts whatever they posted. A ratio against the Creator's own median shows which Videos did unusually well for who posted them, which is where a repeatable hook or format is most likely to be.
Which Platforms can be scored?
TikTok can be scored one Video at a time. Instagram Reels get Scores through a Creator lookup. YouTube Shorts and Facebook Videos get Scores through a Creator lookup too; in our tests the YouTube ones were estimated from follower count. All Scores stop at 30 days after posting.
What outlier number counts as viral?
There is no fixed line, and ShortsIntel does not label Videos as viral. 2x or more is a common place to start reading. Check outlier_p90 as well: above 1 means the Video beat even the Creator's best tenth.
How fresh is the data?
Each lookup reads the Creator's current Videos and dates every snapshot. Baselines are recomputed when they are more than 7 days old. The samples on this page were captured on 2026-11-01.

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.

Connect ClaudeOpen the Playground