Videos that beat their own channel's normal. Unlimited scans, any language, no account.
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.
Go to console.cloud.google.com and sign in with any Google account. It is free. You will not be asked for a card.
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.
Open this page and click the blue Enable button. If it already says "Manage", it is on and you can move on.
Go to Credentials, click Create credentials at the top, then API key. A long string starting with AIza appears. Copy it.
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…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.
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.
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.
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.
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.
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.
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.
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.
Type a niche. findbangers searches YouTube for channels active in that niche, pulls the last fifty uploads from each one, works out that channel's median views per day, and then flags every video that beat its own channel's median by the multiple you chose. The result is a list of videos that broke out relative to where they started, which is the only kind of video worth copying.
Sorting by raw view count tells you which channels are already big. Sorting by overshoot tells you which ideas worked. Those are different questions, and only the second one is useful before you film.
Most tools compare a video to the channel's average views. An average is wrecked by a single old viral hit: one video from three years ago with ten million views drags the bar so high that today's genuine breakouts look normal. The median ignores that video entirely, because it only cares about the middle of the distribution.
Dividing by age matters just as much. A video published last week with 200,000 views is moving faster than one published last year with 400,000, and raw totals hide that completely. findbangers compares views per day against views per day, so a nine-day-old breakout can outrank a two-year-old hit.
Every result carries a second number next to the multiplier. A nine-times outlier on a channel with 900 subscribers is worth far more to you than a nine-times outlier on a channel with 800,000, because the small channel proves the idea carried the video rather than an existing audience. The steal score weights the multiplier against how far the view count outran the subscriber count, and against how recently it happened. Sorting by it puts the most copyable wins at the top.
| findbangers | Typical paid tool | |
|---|---|---|
| Cost | Free, unlimited | $30–$100 a month |
| Free tier | No gate at all | 1–10 searches, then pay |
| Non-English niches | Live, every language | Pre-scraped English mostly |
| Comment mining | Included, unlimited | Credit-metered or absent |
| Data freshness | Live from YouTube | Database, refreshed on their schedule |
| Setup | One API key, two minutes | Account and card |
That loop takes about fifteen minutes and costs a few hundred quota units out of your daily ten thousand.