See what's working on rival channels, before they scale it.
Track any competitor. Scan their uploads on demand. Videos punching 2×+ above the channel's own baseline float to the top,and AI reads each outlier's transcript to explain the title hook, thumbnail motif, first-30s beat, and topic angle. Adapt the pattern, don't clone the video.
No OAuth needed for competitors · Scan on demand · Cancel anytime
Scan complete
8 channels · 12 outliers in the last 30 days
AI: contrarian title + face-forward thumbnail
Adapt the pattern into your next video
How it works
From watchlist to a decision in four steps
Track the channels you care about, scan when you want signal, and let AI explain why the outliers worked.
Build your watchlist
Get a shortlist ranked to your channel's size, language, and topic overlap, or paste any handle, channel ID, or URL. No OAuth needed for the channels you track, and they're never notified.
Scan on demand
One click fans out across your watchlist, pulls each channel's recent uploads, and computes each video's ratio against the channel's trailing-10 median.
Spot the outliers
Videos punching 2×+ above their own channel's baseline float to the top. Filter by tier (notable, strong, hot) and sort by breakout multiple, views, or recency.
Explain with AI
Open any outlier. AI reads the transcript and returns four pattern cards: title hook, thumbnail motif, first-30s beat, and topic angle. Adapt, don't copy.
Built for research, not surveillance
Signal you can act on, not a firehose
No daily digests. No noisy dashboards. Just the videos actually over-performing on their own channel, with the reason why.
Trailing-10 baseline
Every scan computes each channel's median view count across its last 10 uploads. Outliers are measured against a channel's own performance, not a global benchmark.
Three outlier tiers
Notable (2×+), Strong (3×+), and Hot (5×+). Colour-coded badges so you can eyeball the board and drill into the biggest breakouts first.
AI insight cards
Four axes for every outlier: title hook framing, thumbnail motif hypothesis, first-30s beat structure, and topic angle. Explains why the video worked, not just that it did.
Filter by tier and date
Drill into the outliers that matter. Filter by tier (Notable / Strong / Hot), date window (7d / 30d / 90d), or channel. Sort by breakout multiple, views, or recency.
Suggestions ranked to your channel
The Suggested tab ranks candidates by your subscriber size, primary language, and topic overlap with your recent titles. A 20K Hindi finance creator sees peers, not MrBeast. Empty pool? One click runs a live YouTube search for your niches.
On-demand only
No background polling, no wasted API quota. You scan when you want signal, perfect for weekly niche reviews or right before scripting a new video.
How outliers are picked
Every video, judged against its own channel
A 200K-view video is a hit on a small channel and a flop on a huge one. Competitor Radar normalises every upload against the channel's own trailing-10 median,so you spot the videos actually breaking out, not the ones riding an existing subscriber base.
- Trailing-10 median, robust against a single viral outlier skewing the number
- Excludes the freshest upload (still accruing views) from the baseline set
- Recomputed every scan so your baselines don't drift with the channel
- Colour-coded tier badges: notable (2×+), strong (3×+), hot (5×+)
Trailing-10 baseline
Uploads normalised to their own channel
AI insight
Four pattern axes per outlier
Title hook
Contrarian frame
“Everyone’s wrong about X” pattern with a 6-word length and superlative last word.
Thumbnail motif
Face-forward, 3-word overlay
Presenter’s reaction shot centre-frame, bold red text top-left, minimal secondary elements.
First-30s beat
Cold open + payoff tease
Opens on the surprising result, backs into the setup, promises the reveal at ~90s.
Topic angle
Insider positioning
Framed as advice from someone who’s done it, targeting viewers halfway to their own attempt.
AI insight cards
Why it worked, in four axes
Open any outlier. AI reads the transcript and returns four pattern cards,title hook framing, thumbnail motif hypothesis, first-30s beat structure, and topic angle. Every card ends with a takeaway you can apply to your own scripting.
- Title hook, framing pattern, curiosity gap, word count, first-word emphasis
- Thumbnail motif, composition, focal element, colour accent, overlay text style
- First-30s beat, cold-open structure, chapter markers, cut cadence
- Topic angle, positioning play (contrarian, insider, superlative, taboo)
Why Competitor Radar
Manual research vs. AI-explained outliers
You could scroll ten channels every morning, screenshot the breakouts, and guess at what made them work. Or you could scan once and read the pattern.
Use cases
Made for anyone who ships in a niche
If someone in your space is already winning, their catalog is the cheapest research you'll ever run.
New creators studying the niche
Track the top 5-10 channels in your space and reverse-engineer what's working before you're big enough to move the algorithm on your own.
Established channels hunting angles
Skip the doomscroll. Once a week, scan the watchlist, open the hot outliers, and ship videos that adapt the pattern to your voice.
Content teams and creators-with-editors
Everyone on the workspace sees the same watchlist and the same outliers. Reviews turn into decisions instead of discovery.
Agencies managing multiple channels
One watchlist per client channel, one insight modal for every breakout, one board your whole team works from.
FAQs
Questions, answered
Everything you might want to know before your first scan.
Do I need OAuth or manager access to a competitor's channel?
No. Competitor data is limited to what any viewer can see on the channel page. We never touch private analytics, and competitors are never notified you're tracking them.
How is 'outlier' defined?
Every scan pulls each tracked channel's 15 most recent uploads and computes the median view count of the last 10 (excluding the freshest one, which is still gaining views). Any upload with 2×+ that baseline is flagged as an outlier. 3×+ is Strong, 5×+ is Hot.
Why scan on-demand instead of automatically?
Auto-tracking tends to surface signal you never act on, a firehose of daily notifications about videos that aren't actually breaking out. On-demand scans put the trigger under your control, perfect for a weekly niche review or a research session before scripting.
How does insight generation affect my AI credits?
Generating a fresh insight consumes credits from your workspace's AI credit pool. The cost per insight is modest and every plan's monthly quota is sized to comfortably cover a weekly review of a normal watchlist. See the AI Credits page in the docs for the full model.
How many channels can I track?
Depends on your plan. The watchlist size cap is a plan-level limit, see the pricing page for current allowances. You can add and remove competitors freely within that cap.
Can I search YouTube for channels I don't know the handle of?
Yes, sort of. The Suggested tab is a personalised shortlist ranked to your channel. If it's empty, hit Find channels for me and we'll run a live YouTube search for your niches and add the top matches. Free-text keyword search (type any topic, get channels back) is on the roadmap. For known handles, Paste URL is the fastest path.
What happens if a competitor privates or deletes a video?
It quietly drops out of subsequent scans. If you'd already generated an insight for that video, it remains readable in your Insights history, useful for the analysis, even if the source is gone.
Want a deeper dive? Read the full guide.
Stop guessing what's working in your niche.
Add your first competitor, run a scan, and let AI explain the breakouts before you script your next video.