Most niche lists rank by how much money a niche is said to make. That ignores the one thing you control: whether you can keep producing after the first burst of ideas. Score your niche on supply first, then pick an engine that feeds it.
The five-part test#
A faceless channel rarely fails from bad videos. It fails from an empty queue. So a niche must supply material at volume, and be worth money at the other end. Score each item from 0 to 2. Commit at 8 or more. Do not start below 6.
1. Material supply#
How many videos exist now, without you creating new source material?
- 2: hundreds. There is an archive, a document set, a question queue, a dataset or a category with many comparable items.
- 1: dozens. Enough for a quarter with effort.
- 0: you would have to invent the material as well as the video.
Test it: write 30 concrete video subjects in 20 minutes. Subjects, not categories. If you cannot, the supply is not there.
2. Refill rate#
Does the source produce new material on its own?
- 2: weekly or faster: support questions, product releases, prices, new entrants in a category.
- 1: monthly, or you have to go and find it.
- 0: a fixed archive. Once used, it is gone.
This is the strongest predictor on the list. A small archive with fast refill outlives a huge archive with none.
3. Monetization depth#
If someone watches ten videos and then buys, what did they buy?
- 2: a considered purchase, a subscription, a service or a product with real margin.
- 1: a cheap product bought again and again, or a category with real affiliate commissions.
- 0: nothing obvious. The plan is platform payouts, which bets everything on volume and rules you do not control.
4. Format fit#
Can text, stills, screens or a generated presenter carry the material?
- 2: naturally. Words, numbers, comparisons, screens or objects that photograph well.
- 1: with work. Some of it needs footage you must source.
- 0: it needs a specific person doing a specific physical thing. Faceless is the wrong model.
5. Rule risk#
How likely is a platform policy, ad rule or regulator to make your content a problem?
- 2: low. Nothing needs a licence or a disclaimer.
- 1: manageable with disclosure and care.
- 0: high: health, financial or legal advice, anything age-restricted. Possible, but review becomes part of the job.
Niches that tend to score well#
These score well on the rubric, which is not the same as “make the most money”.
- Software and AI tools. Docs, changelogs, support queues and comparisons give huge supply and fast refill, and the material is already on a screen.
- Ecommerce with a visual product. The catalogue is the archive, restocks and drops are the refill, and slideshows do most of the work.
- Courses and coaching. The curriculum is the archive and student questions are the refill. Rule risk depends on the subject.
- Local services with a before and after. Every job is new material, and stills in sequence carry it.
- B2B services with an objection queue. Recruiting, accounting, insurance. Low glamour, high customer value, and sales objections are the content.
- Games. Endless comparable items, constant releases, and rankings that start arguments in the comments.
Browse by business type in Looptake for industries and by goal in use cases.
Niches that look good and score badly#
- General motivation. Infinite supply, nothing to sell, and an audience of everybody, which means nobody.
- Broad facts and trivia. Same shape: great supply, no buyer. It depends entirely on payouts.
- News commentary. Perfect refill, high rule risk, and an archive that is worthless a week later.
- Licensed advice. Finance and health convert because the stakes are high, which is why the rules are strict. Score rule risk at 0 and staff the review step.
- Broad lifestyle. No queue, no archive, no refill. Every video starts from nothing.
- Pets. Supply looks good, but the shot you need is one the animal has to agree to, and the reach often buys nothing.
Pick an engine, not a theme#
A theme tells you what episode one is about and nothing about episode forty. An engine is a repeatable input that produces the next episode without you inventing it. A good engine passes three tests: you know where episode 40 comes from, the input refills without you, and every episode is recognisably part of the same series.
A document you already own#
Docs, onboarding emails, a knowledge base, an FAQ. Each paragraph is an episode, already in order. Series ideas: “One setting, explained” or “The manual, out loud”. Format: text over footage, or screen capture.
A queue of questions#
Support tickets, sales objections, comments, DMs. The strongest engine, because questions arrive whether or not you post. Series ideas: “Asked this week” or “The objection”, where each episode answers one reason people say no. Format: a question card with a short answer, or a wall of text.
A number that changes#
Prices, rankings, stock levels, release counts. Series ideas: “This week’s number” or “Cheapest right now”. Format: a stat video, one number and nothing else. Try the stat video maker.
A public comparison#
Pairs grow faster than items: ten items give 45 pairs. Series ideas: “A versus B, one criterion”, a tier list, “Same job, three tools”. You can compare without naming competitors, and it ages better.
An archive#
Old blog posts, webinars, a podcast, a newsletter. Series ideas: “From the archive”, cut to the best 15 seconds, or “Then and now”. Long recordings can be clipped into shorts.
A recurring event#
A weekly release, a monthly report, a season. Series ideas: “Shipped this week” or a countdown to a dated event. Format: hook plus demo, or wall of text.
Wire the engine to a format#
The engine decides what you make. The format decides what it costs. Text, slideshow and meme formats are cheap enough to post daily. AI avatar videos cost more, so use them on the episodes that need a person saying the words. In Looptake you describe each episode in one sentence and pick or let it pick the format. See the AI video generator and pricing.
Name the series so it compounds: a fixed prefix (“Shipped this week: X”), the specific noun in the title, one playlist per series, no renumbering.
Mistakes to avoid#
- Picking a niche from an RPM table. Rates depend on the buyer, not the topic label, and they change without notice. A high-rate niche you cannot supply produces zero videos.
- Five series on one channel to see which sticks. The recommendation system and your subscribers learn nothing. Run one series for 30 episodes, then decide.
- Narrowing to a hundred items with no refill. That is a series, not a channel.
- Opening with “welcome back”. Every Short is somebody’s first.
Fixing a 6 or a 7#
The fix is almost always the same: attach a refill source. A fixed archive plus a question queue scores two points higher than the archive alone. If you sell something, your customers already produce the refill. Nobody has written it down yet. Once the niche scores, set up the channel with how to start a faceless YouTube channel.