findbangers

YouTube Shorts outlier finder

Shorts that beat their own channel's normal. Free, unlimited, any language.

What this does, in one paragraph

Every channel has a normal. Most of its videos get roughly the same views. Then once in a while one video does ten times better than the rest — not because the channel got bigger, but because that idea was better. Overshoot finds those videos. It looks at a channel's last 50 uploads, works out its normal, and shows you only the videos that smashed it. Those are the ideas worth borrowing.

It runs entirely inside your browser using a free key from Google. There is no account, no card, and no server of ours in the middle. That is also why it is free and always will be — you are spending Google's free daily allowance, not our money.

Getting your free key — about 4 minutes

Open Google Cloud Console

Go to console.cloud.google.com and sign in with any Google account. It is free. You will not be asked for a card.

Make a project

At the top of the page there is a project dropdown. Click it, then New project. Name it anything — "youtube research" is fine. Click Create and wait a few seconds for it to switch over.

A project is just a folder Google uses to group things. It does nothing on its own.

Turn on the YouTube Data API

Open this page and click the blue Enable button. If it already says "Manage", it is on and you can move on.

This is the step people skip. Without it your key exists but every request comes back refused.

Create the key

Go to Credentials, click Create credentials at the top, then API key. A long string starting with AIza appears. Copy it.

Lock the key to this website

In that same popup click Edit API key. Under Application restrictions choose Websites, click Add, and paste the line below. Then under API restrictions choose Restrict key and tick YouTube Data API v3. Save.

loading…
This makes the key useless to anyone else, even if they somehow saw it. Skippable, but do it.

Paste it into Settings

Open the Settings tab here, paste the key, and press Test key. If it says the key works, you are done forever — it is remembered in this browser.

Stuck on an error?

Press Test key in Settings first. If it fails, this page will tell you exactly which of the six steps above went wrong. If it still makes no sense, copy the question below into ChatGPT, Gemini or Claude — it contains everything they need to walk you through it.

Straight answers

Is it safe to give this site my API key?

The key is saved in your browser's own storage and sent only to googleapis.com. We have no server, no database and no analytics on your key — there is nowhere for it to go. You can verify this yourself: press F12, open the Network tab, and run a scan. Every request goes to Google. You can also delete the key at any time from Settings, or from Google Cloud, which kills it everywhere instantly.

Will this cost me money?

No. Google gives every key 10,000 free units a day. A normal scan costs about 600, so roughly sixteen scans a day. The counter on the Outliers tab shows what you have spent. If you run out, it resets at midnight US Pacific time. Google will not bill you unless you deliberately request a paid quota increase.

Does it work in my language and country?

Yes, all of them. Set the country and language on the Outliers tab. Unlike tools built on a pre-collected English database, this queries YouTube live, so Hindi, Spanish, Portuguese, Arabic, Japanese, Indonesian and everything else work exactly as well as English. Type your niche in the language you want results in.

What does the multiplier actually mean?

It compares one video against its own channel, not against YouTube as a whole. A 10× means that video is pulling ten times the views-per-day of the channel's median video. That is the useful comparison, because it strips out channel size — a 10× on a 5,000-subscriber channel is a better signal than a million views on a channel that always gets a million.

Why do some channels get flagged "erratic"?

If more than a quarter of a channel's catalogue beats its own median by your threshold, that channel's median is broken — usually it uploads a lot of near-dead videos and a handful of hits. The multipliers are technically true but they exaggerate. We cap each channel at three results and mark it so you know.

Nothing came back. What now?

Drop the threshold to 3×, lower minimum views to 2,000, and set maximum subscribers to 100,000 — small channels are where copyable wins hide. Also try turning off "Title must mention my niche", and check you picked the right language for the words you typed.

Scan settings
These are appended to your niche. They switch automatically with the language you pick — edit them freely.
0 units
today: 0 / 10,000
1 unit per 100 comments. Mining 5 videos at 300 each costs 15 units. This finds the comments where viewers are literally asking for a video — the cheapest content ideas on the internet.

Idea bank

Stored only in this browser, sent only to Google. New here? The Start here tab walks you through getting one in about four minutes.
Quota
Google gives 10,000 units per key per day, resetting at midnight US Pacific. This page counts what it spends. The count is per browser, so scans from your phone and laptop are tracked separately.

Shorts need their own baseline

The mistake almost every tool makes is comparing a Short against the channel's overall average. A channel posting both formats will have wildly different view patterns for each, and mixing them produces a median that describes nothing. This page presets the format filter to Shorts only, which means each channel's baseline is computed from its Shorts alone, and a Short is judged against the standard its own channel set for Shorts.

The 2024 rule change, handled correctly

YouTube extended the maximum Shorts length from 60 seconds to 3 minutes in October 2024. Any tool applying a flat 60-second cutoff now misclassifies a large slice of recent uploads — treating real Shorts as long-form and wrecking both baselines at once. findbangers checks each video's publish date and applies the limit that was actually in force at that moment.

What to look for

  • Recency above all. A Shorts format that worked eight months ago may be exhausted. Sort by newest and work backwards.
  • Small channels with huge multipliers. On Shorts the feed distributes almost independently of subscriber count, which makes a 900-subscriber channel with a 40× Short the purest possible signal that the hook itself worked.
  • The first two seconds. Open the winners and watch only the opening. The thumbnail barely matters on Shorts; the opening frame and first line carry everything.

Then mine the comments

Shorts comment sections are shorter and blunter than long-form ones, which makes the requests easier to spot. A breakout Short with two hundred people asking the same follow-up question is a long-form video waiting to be made.