- The single hottest packaging machine in this corpus is the grocery/consumer exposé format. 'We Tested {product} From Walmart, Costco, Target, Aldi & N More Stores | Only X Were Worth Buying' has a median outlier of 287.8x and every instance is recent [@everythinguncovered-i8c ox3DNkiD0_E 950k views on 21.5k subs; 8ZpHlYP0jX4 876k]. A 1,100-sub channel cloned it and got 64k views [@americaunderpressure-f6q UmgKcCASCxs, 245x]. The format works at zero subscriber base, which is exactly what a faceless operator needs.
- Second hottest: the 'store news' curiosity loop '{Brand} Just Did What No Grocery Store Would' — median 157.6x, 3/3 recent, on a 9k-sub channel [@consumertruthyt ReJh-U102t0 Whole Foods 456x, j3SwZwCiSD8 Trader Joe's, TNVVw1z-h3I Kroger]. It is infinitely repeatable across every retailer in America and requires no footage, only a news-desk thumbnail.
- Third: the regional heritage-cook persona. 'NEVER {method} Again — Try This Old {Region} Trick' median 59.7x and it is minting new channels in weeks: [@otisgrangerkitchen] 26k subs and 1.93 views-per-sub with r5cw3juqudw at 1.01M; [@backyardreggie] quHbNgSqkUA at 1.34M on 18.7k subs (71 views/sub). These are the highest views-per-sub assets in the entire dataset and they all sell cookbooks.
- Views-per-sub, not subs, is the money metric here. [@goldenhealth-channel] shows 38.6 v/s, [@michikotanaka-o5w] 1.52, [@mayatanakayt] 2.87, [@otisgrangerkitchen] 1.93, [@theribdavid] 1.24 on 573 subs. Big legacy channels are the opposite: [@wildernesscooking] 0.17, [@thebeardedbutchers] 0.03, [@curaxa] 0.02. Audience size is not moat; packaging is.
- Decay is fast and brutal in this space. [@consumerexposed-ghz] went 148k subs in three months then dropped median views from 62k to 37k; [@thehiddenpantry0405] halved (29.4k→15.7k); [@healthfixnow] collapsed from 324k median to 2.6k; [@drvitalishealth] from 495k to 5.1k. The pattern: one format, mass repetition, algorithmic fatigue. Plan channel #2 before channel #1 peaks.
- Dog psychology is a mature but still-paying machine, with the emotional-guilt cluster outperforming the trivia cluster. 'Signs your dog is thriving' (opportunity 88.8), 'Things dogs do for favorites' (87.3) and 'Things that emotionally hurt dogs' (85.1) are rising, while 'dog stretching meaning' (saturation 51.3) and 'signs a dog is dying' are decaying [@thedogspirit96 E_QV8ycuqmk 770k, 493x].
- Senior health is the highest-risk / highest-ceiling vertical. [@vitalityafter60health ne3H6-gsKwM] hit 761k views on a 10.9k-sub channel (2,533x) with a stroke warning-sign video, but three of the four senior-health channels tracked are in freefall and one openly labels itself an AI presenter [@alphareboothealth]. Treat this as a monetization vertical only with disclaimered, non-prescriptive framing.
- Title triggers that actually correlate with outliers now: urgency (26.7 median), hope (19.4), outrage (19.4), curiosity_gap (17.1), greed (15.8). Thumbnail-wise, warning_sign is the most-used pattern (n=85) but red_circle_arrow (15.9), scene_story (15.3) and list_preview (16.4) beat it per-use — the fresh edge is a labelled list-preview or a newsdesk scene, not another red DANGER triangle.
SATURATION TRAPS — do not enter these: 'Egg frying trick hacks' (100/100 saturation, recent performance down from 7.81 to 2.16 across 8 channels), 'Brands to avoid vs real' (67.4 saturation, 118 videos, 70 in the last 90 days), 'Genius car hacks' (53.3, collapsed from 7.41 to 1.29), 'Most reliable cars & SUVs' (57.0), 'Dog stretching meaning' (51.3 across 6 channels), 'Grocery store taste tests' (52.8 and climbing fast — you have maybe 90 days of runway on the pure format before it looks like the egg topic).
BURN RATE IS REAL. [@consumerexposed-ghz] built 148k subs in three months then watched median views fall 62,093→37,185 while still uploading 6.9x/week. [@thehiddenpantry0405] halved. [@healthfixnow] went 324,238→2,575. The pattern is single-format mass repetition. Cap any one title skeleton at roughly 40-60 videos, then fork the channel or fork the format.
YOUTUBE'S INAUTHENTIC / MASS-PRODUCED CONTENT POLICY IS THE CENTRAL EXISTENTIAL RISK HERE. Several tracked channels are running 6-10 uploads per week of near-identical, templated videos ([@eliasyoderamish] 8.5/wk, [@amishcooking] 9.5/wk, [@silasmercerfarmstead] 8.5/wk, [@silasmerceryt] 10.2/wk). That cadence with interchangeable scripts and stock footage is exactly what the July 2025 'inauthentic content' guidance targets. Add genuine per-video variation: unique narration, original commentary, differentiated visuals, or you will get demonetised in a sweep rather than individually.
AI DISCLOSURE IS NON-OPTIONAL. If you use a synthetic presenter or synthetic voice depicting a realistic person, you must tick YouTube's altered-content disclosure and, in several jurisdictions, say so in the description. [@alphareboothealth] and [@menhealthwithdrjocelyn] both carry explicit AI-presenter disclaimers — copy that practice. The bigger exposure is the invented biography: '[@otisgrangerkitchen]'s mother died in 2004', '[@gloriaskitchentable] fed nine people on eight dollars a week', '[@titusfreemanyt] farmed forty acres in Alabama since 1919'. These are fabricated life stories attached to synthetic faces. Keep personas emotionally true but factually vague, and never fabricate verifiable specifics (real addresses, real family names, real obituaries).
MEDICAL CLAIM EXPOSURE IS YOUR BIGGEST LEGAL AND POLICY RISK. Thumbnails in this corpus literally say 'SAY GOODBYE TO DIABETES' [OYYC3GdjoRs], 'NATURAL WORM KILLER' [czKtQQ26mBs] and 'Better Than M*rphine' [GEQKrSglRAM]. That is medical-misinformation territory and, for the morphine comparison, potentially an unapproved-drug-claim problem. Never claim to cure, reverse, treat or replace medication. Never invent an MD persona with a fake name and credentials — that is both fraud-adjacent and a YouTube impersonation violation. Always: on-screen disclaimer, description disclaimer, and 'talk to your doctor'.
BRAND DISPARAGEMENT IS THE SECOND LEGAL RISK. 'NOT REAL CHICKEN', 'CANCER IN WATER', 'PLASTIC, NOT SUGAR', '90% FAKE BEEF', '15 Shoe Brands ROBBING You Blind' — these are factual assertions about named, identifiable companies. If untrue and damaging, that is trade libel. Protect yourself: cite the ingredient panel (public and defensible), cite regulator actions and recalls, use 'reportedly' and 'in our opinion', show your sources on screen, and never claim laboratory testing you did not perform. The 'We Tested' format specifically implies testing — if you did not test, say plainly what your method was.
FINANCIAL AND BENEFITS CLAIMS. Social Security, IRS and bank-fee content ([@westonretire]) is high-conversion and high-risk. Government-program misinformation is a specific YouTube policy category. Cite the actual rule and its effective date on screen, never promise a dollar increase, and carry a 'not a financial advisor' line — the model [@pantry-secrets] already uses in its channel description.
CELEBRITY IMPERSONATION. [@themediterraneandietyt] is putting invented first-person quotes into the mouths of Ringo Starr, John Goodman and Danny DeVito ('I'm 86 but Feel 50'). That is a right-of-publicity claim, a potential defamation claim, and a YouTube impersonation strike all at once — and the channel's own non-celebrity video (jB804hrTwJ4, 57.7x) proves you do not need the celebrity. Drop named living people entirely.
PERSONA CLUSTER CONTAMINATION. The [@silasmerceryt] / [@silasmercerfarmstead] / [@silasmercercooking] and [@eliasyoderamish] / [@amishgardening] / [@amishcooking] and [@pearlfreemancooking] / [@pearlfreemanfrugalliving] and [@titusfreemanyt] / [@titusfreemanhomestead] / [@biblelessonswithtitas] clusters share a single identity across channels. That is efficient for funnels, but it means one policy action or one authenticity exposé can take out the whole network. Separate the payment, email and hosting infrastructure per cluster, and note that the Bible and homestead spin-offs are already failing (60 and 402 median views) — spin-offs do not inherit audience automatically.
AUDIENCE VULNERABILITY IS AN ETHICAL AND REPUTATIONAL LINE. The highest-converting identities in this data — 'seniors living alone', 'if you have diabetes', 'over 70', 'living on one income' — are lonely, anxious and financially squeezed. Selling them a $19 PDF is fine. Selling them fear about stroke symptoms, parasites, bank fraud and food contamination in order to sell them a PDF is where the business gets ugly and where a journalist or a platform enforcement team eventually looks. Keep the product genuinely useful and the fear proportionate to the evidence.
CADENCE COLLAPSE KILLS PROVEN FORMATS. [@midnightoracie] had a 177x and a 129x video and then dropped to 0.4 uploads/week — health 23.9. [@drnorakhalil-seniorsheatlh] had 20 recent outliers and dropped to 0.4/week — health 31.3. [@drvitalishealth] had an 821k video and shipped 14 videos total. If your operation cannot sustain 3-5 uploads a week per channel, do not open the channel.
BEWARE THE VIEWS-PER-SUB MIRAGE IN REVERSE. [@goldenhealth-channel] shows 38.58 views-per-sub off a single video and zero recent uploads; [@adviseofgrandmamartha] has 67,600 subs and 834 median views. Purchased or inherited subscriber counts tell you nothing. Judge every acquisition or model channel on recent median views and recent outlier count only.