Lume: script patterns by public-view group
@lume-channel
Lume's higher-viewed and lower-viewed uploads open differently, and this page shows the difference in their own words. Its widest-travelling upload here reached about 12.4× the channel median. We compared 3 recurring patterns from the higher-view group against 3 from the lower-view group, quoted verbatim. This is an observational read of the scripts, not a measurement of retention.
What this page is
A risk map derived from the words in Lume's public scripts. The uploads are split into groups by public view count against the channel's own median, and the analyzer reports the script patterns that recur in each group, quoted verbatim. It is not a measurement of this channel's audience retention, which is private to the channel owner, and it does not establish that any pattern caused any view count.
Script difference to review
Lower performing videos often begin with broad organizational collapses, meta channel drama, or passive crime scene retrospectives rather than framing the scene around an individual consumer or underdog exploiting a precise, relatable oversight.
Lume has 316K subscribers and a median of 126.1K views across its recent long-form uploads. Its highest-viewed video in this sample did about 12.4× that median. This breakdown compares recurring transcript patterns in the channel-relative public-view groups, each backed by a verbatim quote. The comparison is observational and does not reveal private retention or establish what caused the view counts. The lower-view group has a cumulative gap of 348.1K public views below the channel median. That is descriptive arithmetic, not an estimate of views lost because of a script pattern.
Patterns in the higher-public-view group
More specific titles tend to get more views on this channel. Specific titles tell the viewer what the video covers before they commit. Across 23 videos it's a clear link, and the odds it's just luck are about 1 in 67 (rho +0.44, p 0.01).
12.4× channel median“Pepsi Spent $2 Million to Stop One Kid”
Titles with a number in them tend to get more views on this channel. A number makes the title's promise checkable before the click. Across 23 videos it's a clear link, and the odds it's just luck are about 1 in 13 (rho +0.36, p 0.07).
2.4× channel median“He Turned One Piece of Junk Mail Into $95,000”
Establishes precise financial stakes upfront
12.4× channel median“Inside, a cashier's check for $700,008.50 made out to PepsiCo Incorporated. Stapled to it, a single order form from the official Pepsi Stuff catalog.”
Patterns in the lower-public-view group
Focuses on passive background narrative details
0.6× channel median“Behind him, 300 officers, snipers on the rooftops, 6 hours of live television, a block sealed since lunchtime.”
Leads with abstract corporate systemic collapse
0.5× channel median“In minutes, the doors are changed shut and 1.4 million people from London shopkeepers [music] to villagers in Bangladesh wake up to find their money is simply gone.”
Starts with meta YouTube drama questions
0.5× channel median“Who owns Fern? Who owns their channels? And who owns their name? You're curious, too, right?”
A writing hypothesis to test
Try this in your next video: make the title more specific about what the video covers. On this channel, more specific titles tend to get more views. Across 23 videos it's a clear link, and the odds it's just luck are about 1 in 67 (rho +0.44, p 0.01).
What this page cannot tell you
It cannot tell you whether these patterns would do anything for your channel, because it never saw your script. The patterns above are what recurs in one channel's published uploads; whether your own opening holds attention is a question about your draft, and it is answerable before you record.
Questions about this breakdown
What does this Lume breakdown actually measure?
It compares the transcripts of Lume's higher-viewed and lower-viewed recent long-form uploads, grouped against the channel's own median of 126.1K views, and reports the script patterns that recur in each group with a verbatim quote for each. It is a read of the scripts. It does not use audience retention data, which is private to the channel owner, and it does not establish that any pattern caused any view count.
Is this a prediction of how the videos performed?
No. The analyzer reads the words in a script and flags where attention is most at risk on the page. Published view counts are used only to split the uploads into groups to compare. Nothing here is a measurement or a forecast of what real viewers did, and the same script can travel very differently depending on topic, packaging, timing, and distribution.
How current is this analysis?
This page was analyzed on October 11, 2026 and the result is frozen at that point, with the analyzer version stamped on the page. YouTube changes, channels change, and a scan run today could group different uploads. Re-read the quotes against the channel's current videos before relying on any pattern.
Can I get this for my own script?
Yes, and that is the more useful direction. This page shows what the analyzer found in someone else's published scripts. Running your own draft through the free check gives you the same read on the script you are about to record, while you can still change it: hook, pacing, and structure scores, and the passages most at risk of losing attention.
This page reads someone else's script. Read your own.
Paste the draft you are about to record and get the same kind of read on it: hook, pacing, and structure scores, and the passages most at risk of losing attention, while you can still change them. Free, no login. The line-by-line rewrites come with a Creator plan if you want them.
Check my script freeRelated reading
- All channel breakdowns : the same read on other channels.
- Hook examples that open on a claim : the pattern most often separating the two groups above.
- The retention guide : what the curve shapes mean and which are script problems.
- Our 349-hook study : first-party research with the row-level data published.
- Run this scan on your own channel : free, needs captioned long-form uploads.