THE DOSSIER · GROWTH ANALYSIS
SNAPSHOT · 2026-05-17
Subagent · content-forensics

Content Forensics

Which videos beat or missed the channel's own baseline, and what variables they shared. Within-cohort z-scores, never cross-cohort.

Long-form: per-cohort view z-score
n = 24 videos · quartile cut at ±6
-2 -1 0 +1 +2 peak +2.44 floor -1.81
Top quartile Bottom quartile Middle
Source: data/videos/summary.jsonl
Shorts: per-cohort view z-score
n = 36 videos · quartile cut at ±9
-2 -1 0 +1 +2 peak +2.78 floor -1.83
Top quartile Bottom quartile Middle
Source: data/videos/summary.jsonl
Top quartile · long
n = 6 videos by per-cohort z-score
zTitleViews
2.44 Rampur Islam*c Horror Exposed: Yashoda’s Family Tragedy…
yCR38NmeCpY · 2760s · crime_investigation
258,681
1.74 Ex-Muslim Unfiltered: Why I Couldn’t Stay in Islam? The…
oYTJaHw4na8 · 2992s · religion_society
111,677
1.37 Operation Sindoor: When India Rained Hell in Pakistan: …
Gs38SCGxPGA · 484s · geopolitics
71,715
1.13 Illegal Infiltration &Smuggling, Sandeshkali, PostPoll …
efBn5kJ0snc · 4221s · politics_govt
53,388
1.01 Population Explosion का झूठ और Delimitation का सच । 14…
yLEakbf5gpI · 3212s · geopolitics
46,413
0.64 Judiciary's Dirty Secrets: Anti-Hindu Discrimination, F…
zv2HCSdsiJw · 4501s · law_judiciary
29,534
data/videos/summary.jsonl
Bottom quartile · long
n = 6 videos by per-cohort z-score
zTitleViews
-1.81 Iran War Explained: The Hidden Power Game | Why Are Chi…
6Ix6Tt4I47A · 719s · geopolitics
1,542
-1.74 Pakistani Hindu Refugees: Story. Struggle. Survival
atLAW_yl70Q · 1015s · religion_society
1,672
-0.95 6 December 1992: Babri demolition, Ram Mandir and Truth…
W4I42zE8qTA · 328s · religion_society
4,365
-0.83 Iran-Israel War and the Making of Nuclear Weapons by Ir…
DNHWfGBfUiM · 1481s · religion_society
5,039
-0.76 22 Rules for a Hindu Living in an Islamic State। Dhimmi…
1E6So2gYgP8 · 1356s · religion_society
5,507
-0.71 All about Nida Khan__Lady D of jihadi scandal
nFIGdsXAYpA · 293s · crime_investigation
5,811
data/videos/summary.jsonl
Top quartile · short
n = 6 videos by per-cohort z-score
zTitleViews
2.78 The TCS Story 😱 #shorts
rpk9_nXIvdM · 53s · geopolitics
14,905
2.33 भूरा बाल साफ करो लालू ने कभी नहीं कहा? #shorts #laluyad…
NHTNJPcvO0Q · 173s · politics_govt
10,547
1.76 Lalu Yadav के Theaterics, मूर्ख और देहाती होने का ढोंग!…
fmrAS_vkCUE · 135s · politics_govt
6,824
1.57 कर्पूरी ठाकुर को कह देना की मेरी जीप में तेल नहीं हैं �…
5wMrZDvZl4k · 125s · politics_govt
5,899
1.22 17 Years of Torture, Yet Colonel Purohit Holds No Bitte…
CDdpB7oWm0k · 53s · geopolitics
4,529
1.22 TCS Nashik Hub of Religious Conversion and Womb Jihad #…
rajRbU2z2mw · 114s · religion_society
4,528
data/videos/summary.jsonl
Bottom quartile · short
n = 6 videos by per-cohort z-score
zTitleViews
-1.83 Khamenei’s Last Day 😱 #shorts #iranvsisrael #war
6mZkBzkrRmE · 23s · geopolitics
437
-1.46 The Strategy Behind the Silence: How China & Russia Ben…
WCiY3qM_s00 · 21s · geopolitics
578
-1.10 The Day Cow Smuggling Stopped in Basirhat #shorts #west…
XywDiNYoGl0 · 32s · uncategorised
764
-0.93 इस्लाम में महिलाओं को कितना सम्मान मिलता है #shorts #is…
NFmZry3zzM0 · 105s · religion_society
871
-0.92 भगवा आतंक के झूठ के कारण हाफिज सईद ने UN में भारत को Te…
wCQ6uShULAM · 69s · religion_society
875
-0.82 'They consider the invaders as their heroes'. #shorts #…
HCSoVCTMz8A · 26s · politics_govt
948
data/videos/summary.jsonl
Long-form: variables that correlate with overperformance
Share of bucket (n=6 top, 6 bottom)
TOP QUARTILE BOTTOM QUARTILE Length ≥ 25 min 83% 0% Sunday upload 50% 0% Saturday upload 0% 50% ≥1 CTA detected 67% 33% religion_society as primary cluster 17% 67%
content-forensics report
Shorts: variables that correlate with overperformance
Share of bucket (n=9 top, 9 bottom)
TOP QUARTILE BOTTOM QUARTILE Length ≥ 50s 100% 33% Length < 30s 0% 44% politics_govt as primary cluster 44% 11% law_judiciary as primary cluster 0% 22% Wed/Sat upload 56% 11%
content-forensics report

Content Forensics — The Dossier

Methodology

The view_z_score field in summary.jsonl is a log10-views z-score computed within cohort (24 long-form vs 36 shorts treated separately), so it already normalises for the platform-level view-count gap. Given small n, quartile cuts were used: top quartile = 6 long-form / 9 shorts; bottom quartile = same. All comparisons below are top-quartile vs bottom-quartile within the same cohort. No cross-cohort ranking is performed. Variables examined: length bucket, topic_cluster, day-of-week of upload, title pattern (interrogative / declarative / list / how-to / reactive / accusatory), numeral presence in title, title length in chars, CTA count, age in days.

Top performers (long-form)

video_idtitleview_zage_dayslength_stopic
[yCR38NmeCpY]Rampur Islamic Horror Exposed: Yashoda’s Tragedy2.4382582760crime_investigation
[oYTJaHw4na8]Ex-Muslim Unfiltered: Why I Couldn’t Stay in Islam?1.7412242992religion_society
[Gs38SCGxPGA]Operation Sindoor: When India Rained Hell in Pakistan1.374375484geopolitics
[efBn5kJ0snc]Illegal Infiltration, Sandeshkali, PostPoll Violence1.1292804221politics_govt
[yLEakbf5gpI]Population Explosion का झूठ और Delimitation का सच1.013303212geopolitics
[zv2HCSdsiJw]Judiciary’s Dirty Secrets: Anti-Hindu Discrimination0.6383014501law_judiciary

Top performers (shorts)

video_idtitleview_zage_dayslength_stopic
[rpk9_nXIvdM]The TCS Story2.7783253geopolitics
[NHTNJPcvO0Q]भूरा बाल साफ करो लालू ने कभी नहीं कहा?2.326205173politics_govt
[fmrAS_vkCUE]Lalu Yadav के Theatrics, मूर्ख और देहाती होने का ढोंग1.758207135politics_govt
[5wMrZDvZl4k]कर्पूरी ठाकुर को कह देना की मेरी जीप में तेल नहीं1.568204125politics_govt
[CDdpB7oWm0k]17 Years of Torture, Yet Colonel Purohit Holds No Bitter1.22323353geopolitics
[rajRbU2z2mw]TCS Nashik Hub of Religious Conversion and Womb Jihad1.22229114religion_society
[TnGcTkrTrEo]Milords are not against religion, against one Dharma1.058240136religion_society
[PGNi1Zw-tA0]India was never partitioned for Muslims?0.7031281politics_govt
[Flc5mIdH6Ug]मुंबई ATS कर्नल पुरोहित को कहती थी देखो0.304225155geopolitics

Underperformers (long-form)

video_idtitleview_zage_dayslength_stopic
[nFIGdsXAYpA]All about Nida Khan — Lady D of jihadi scandal-0.71032293crime_investigation
[1E6So2gYgP8]22 Rules for a Hindu Living in an Islamic State-0.7551131356religion_society
[DNHWfGBfUiM]Iran-Israel War and the Making of Nuclear Weapons-0.8283281481religion_society
[W4I42zE8qTA]6 December 1992: Babri demolition, Ram Mandir-0.947162328religion_society
[atLAW_yl70Q]Pakistani Hindu Refugees: Story. Struggle. Survival-1.7431831015religion_society
[6Ix6Tt4I47A]Iran War Explained: The Hidden Power Game-1.81066719geopolitics

Underperformers (shorts)

video_idtitleview_zage_dayslength_stopic
[Q4G_3RBO0Nk]Those masjids in Delhi from which Muslims pray-0.6225625law_judiciary
[5JF7pxPlRmE]सोनिया ने रची Mohan Bhagwat को गिरफ्तार करने की साजिश-0.62723995law_judiciary
[s-PodvUEtv8]The fundamental, fanatic Islamic fascism has to stop-0.7624735religion_society
[HCSoVCTMz8A]They consider the invaders as their heroes-0.8195526politics_govt
[wCQ6uShULAM]भगवा आतंक के झूठ के कारण हाफिज सईद ने UN में-0.92323769religion_society
[NFmZry3zzM0]इस्लाम में महिलाओं को कितना सम्मान मिलता है-0.929212105religion_society
[XywDiNYoGl0]The Day Cow Smuggling Stopped in Basirhat-1.1004432uncategorised
[WCiY3qM_s00]The Strategy Behind the Silence: China & Russia-1.4646321geopolitics
[6mZkBzkrRmE]Khamenei’s Last Day-1.8286523geopolitics

Variables that correlate with overperformance — long-form

  • Length 25+ minutes: 5/6 (83%) of top vs 0/6 (0%) of bottom. +83pp. The single non-25+ top video [Gs38SCGxPGA] is an 8-min reactive geopolitics piece; every bottom video is under 25 min.
  • Sub-15-min duration is a graveyard: bottom contains 3/6 videos under 15 min ([nFIGdsXAYpA] ~5m, [W4I42zE8qTA] ~5.5m, [6Ix6Tt4I47A] ~12m); top has zero under 8 min besides [Gs38SCGxPGA].
  • CTA presence: top avg 1.00 CTAs/video, bottom 0.33 (4/6 top have >=1 CTA, 2/6 bottom). 3x rate. Confound: long-form videos with CTAs are also the long-form videos that are long.
  • Topic mix shifts away from religion_society: top is 1/6 (17%) religion_society; bottom is 4/6 (67%). -50pp. Bottom is dominated by religion-coded explainers; top diversifies into crime_investigation, law_judiciary, politics_govt, geopolitics.
  • Title length: top avg 92 chars (range 82-100), bottom avg 79 (range 45-100). Top titles cluster near the 100-char ceiling; bottom contains the only short titles.
  • Day-of-week: Sunday uploads 3/6 (50%) of top, 0/6 of bottom. Saturday is the inverse: 0/6 top, 3/6 bottom. Saturday is also the modal upload day overall (9/24), so the channel’s default cadence is its worst day.
  • Age: top avg 245 days, bottom avg 147 days. Older videos dominate the top — a survivorship effect (long-form continues accumulating views), not a quality signal in isolation.

Variables that correlate with overperformance — shorts

  • Length >=50s: 9/9 (100%) of top vs 3/9 (33%) of bottom. +67pp. Every winning short uses the full duration window.
  • Sub-30s shorts collapse: 4/9 (44%) of bottom are under 30s; 0/9 of top are. The four worst shorts by z-score include three under 30s ([WCiY3qM_s00] 21s, [6mZkBzkrRmE] 23s, [HCSoVCTMz8A] 26s).
  • Topic politics_govt outperforms: 4/9 (44%) of top vs 1/9 (11%) of bottom. +33pp. Three of those four are the Lalu Yadav cluster ([NHTNJPcvO0Q], [fmrAS_vkCUE], [5wMrZDvZl4k]) — a single narrative arc that travelled.
  • Topic law_judiciary underperforms in shorts: 0/9 top, 2/9 (22%) bottom. Law content compresses badly into clip format.
  • Title pattern interrogative or accusatory: 3/9 of top use one of these vs 0/9 of bottom, which is 89% declarative. Small absolute count but a clean directional split.
  • Day-of-week: Wed+Sat 5/9 (56%) of top vs 1/9 (11%) of bottom. Sun+Mon 0/9 of top vs 4/9 (44%) of bottom.

Variables that don’t correlate

  • Numerals in titles (long: 50% top vs 33% bottom; shorts: 11% vs 0% — directions inconsistent at this n).
  • Title length for shorts (top 66.6 chars vs bottom 67.8 — flat).
  • CTA count for shorts (avg 0.00 top vs 0.03 bottom — channel doesn’t use shorts CTAs).
  • Topic geopolitics on either side (split roughly evenly between top and bottom in both cohorts).
  • Title pattern in long-form (accusatory and interrogative appear in both top and bottom in similar ratios).

Takeaways

  • Long-form below 15 minutes is the channel’s structural loss-maker. 5 of 6 top long-form videos are 25+ minutes; 3 of 6 bottom are under 15. Consequence: every long-form upload below the 25-min mark forfeits roughly a z-unit of expected performance unless it is a reactive news hit like [Gs38SCGxPGA].
  • Shorts under 30 seconds and over-indexed on religion_society / law_judiciary are dead weight. 100% of top shorts hit >=50s; 44% of bottom shorts are under 30s, and 5/9 bottom shorts sit in religion_society or law_judiciary. Consequence: the current shorts feed produces a lot of inventory in two formats (sub-30s + religion/law explainer clips) that consistently lands in the bottom quartile.
  • Saturday is the channel’s default and the channel’s worst day for long-form; Wednesday and Saturday are the best days for shorts. Long-form Sat = 9/24 of all uploads but 0/6 of top and 3/6 of bottom. Consequence: the upload calendar is misaligned with where the cohort-z evidence points, costing measurable z on the most expensive (long-form) inventory.
Source data acquired via yt-dlp. Channel under analysis: @thedossier_in · Snapshot date: 2026-05-17 · 60 of 80 videos sampled (YouTube bot challenge blocked the remainder). Every metric traces back to a file in data/.