You opened the "what niche" field on signup, stared at it for two minutes, and almost closed the tab. You weren't the first. Two of the first three people we surveyed during ShortsFast's launch week named that exact field as the moment they nearly left.
Picking a niche is hard because the choice locks in the next 6-18 months of work and most "how to pick a niche" guides give you a list of niches instead of a way to choose. This guide gives you the way to choose. Five questions, run in order, applied to any candidate niche you can think of. A "no" on any question means cut the candidate and move to the next one.
The whole tree takes 15 minutes. By the end you have either two or three viable candidates or — more usefully — clarity on why nothing you've been considering would have worked.
How this decision tree works
Start with a list of 5-10 candidate niches. They can be anything — topics you find interesting, niches you saw recommended on YouTube, sub-niches of a broader space you're curious about. For each candidate, walk through the five questions below in order. The first "no" cuts the candidate and you move to the next one. Don't skip ahead; the questions are ordered by cost of failure.
If you finish the tree with zero survivors, that's also useful information. It usually means you've been considering niches in a single category (often entertainment or hobby) and need to widen the search into purchase-decision-adjacent topics.
Question 1: Can you sustain 100 videos here without running out of ideas?
The first filter is cheapest to fail. If your candidate niche has fewer than 100 video angles, your channel tops out at 100 uploads and the algorithm never gets the post-frequency signal it needs to push you. Most niches that fail here are too narrow — "best wireless earbuds under $50" is a video, not a niche.
The test: write down 20 video titles for the candidate niche without searching anything. If you struggle past 10, the niche is too narrow. If you hit 20 easily and could clearly write another 80, it passes. Don't worry about whether the 20 titles are good — worry about whether there are 20 of them.
Example pass: Personal finance for first-job-out- of-college earners. 20 titles in five minutes: budgeting on entry- level salary, first 401(k) match, paying off student loans vs investing, building credit at 22, etc. Topic well is functionally infinite.
Example fail: Tesla Cybertruck reviews. Three clear titles, maybe seven if you stretch. Not a niche; a topic.
Question 2: Is there real buyer intent in the audience?
The single largest income lever. Niches where viewers are actively researching purchases pay $15-30 RPM. Niches where viewers are entertained pay $1-4 RPM. The 50× gap between top and bottom faceless niches is almost entirely a buyer-intent gap. See our companion guide on best faceless YouTube niches in 2026 for the ranked RPM data across niches.
The test: name three things the audience for this niche actively buys related to the topic. Not "they buy stuff in general" — things they buy because of the topic itself. If you can name three within 60 seconds, the niche has buyer intent. If you can't, the audience is in entertainment mode and RPM will be low.
Example pass: Personal finance — viewers buy brokerage accounts, tax software, financial advice subscriptions, credit cards, online courses. Five purchase categories in 30 seconds. Strong buyer intent.
Example fail: Generic memes. Viewers don't buy anything because of memes. They might buy unrelated things while watching, but advertisers don't pay extra for that attention. Low buyer intent.
The audience that buys things related to your topic is worth 10-30× the audience that's just there for entertainment. This question is the difference between a channel that pays your rent and a channel that doesn't pay for itself.
Question 3: Can the content ship without you on camera?
Faceless-fit check. Some niches lose most of their value when the creator's face is removed — anything that depends on personality, physical demonstration, or trust-by-recognition (fitness, dating advice, hands-on cooking, on-camera reviews). Other niches lose nothing because the visual is footage, charts, animation, or on-screen text and the voice is the host.
The test: imagine the top 5 videos in your candidate niche. How many of them work as voiceover-over-footage with no on-camera presence? If 4 or 5 do, the niche is faceless-fit. If 2 or fewer do, you'll be fighting the format for every upload.
Example pass: Geography and history explainers. Maps, footage, motion graphics, and a voice. Channels like RealLifeLore prove the format works at scale. No face required.
Example fail: Home gym workout tutorials. The face — and the body — are the demonstration. Voiceover over stock fitness clips reads as a cheap knock-off of the face-led version of the same channel.
Question 4: Will YouTube monetize it without policy headaches?
The silent killer. Several otherwise-attractive niches sit in YouTube's limited-ads tier, where ad revenue is capped well below normal rates regardless of audience size. The classic offenders: true crime (especially recent cases or with graphic detail), political commentary, conspiracy topics, fringe medical advice, firearms, and anything in the "deceptive or shocking content" boundary.
The test: search YouTube for the top channel in your candidate niche and check whether their recent videos show ads. If most of their uploads run pre-roll and mid-roll ads, the niche is monetizable. If most of their videos show no ads or only sponsor-read ads, the niche is in a limited tier and your channel will be too. Also check YouTube's advertiser-friendly content guidelines and the synthetic-content disclosure rules before committing.
Example pass: Personal finance. Ads everywhere. Premium advertiser pool. Niche is wide open monetization-wise.
Example fail: Unsolved true-crime cases involving named victims. Top channels in the space frequently lose monetization on individual videos and family-of-victim takedowns happen. The niche works for some — but the policy risk has to be priced in and most beginners haven't.
Question 5: Can you produce this faceless-at-scale with current AI?
Last filter. By now the candidate niche has passed sustainability, buyer intent, faceless-fit, and monetization. The only question left is whether the current generation of AI scripting, voice, and video-assembly tools can ship the format at the cadence the niche needs.
Some niches are structurally easy for AI pipelines (list and countdown formats, fact compilations, quiz/puzzle channels, evergreen psychology explainers). Some are workable but voice- dependent (narrated storytelling, atmospheric formats). Some require research depth or animation budgets that current AI cannot honestly reproduce — see our 13 faceless channel examples guide for the per-channel verdict on this exact question.
The test: find one existing faceless channel in your candidate niche that uses voiceover + footage as its primary production stack (no original animation team, no signature human narrator whose voice is the brand). If you can name one, the format is AI-producible. If every channel in the niche is animated or signature-narrator-led, the niche is gated by production budget you don't have.
Example pass: Personal-finance explainers in the Coffee Break / list-channel style. Voiceover over animated charts and footage. AI-friendly script structure. Niche has multiple AI-feasible producers shipping weekly.
Example fail: Geography explainers in the RealLifeLore mold. Pass on every other filter, fail here — the synthesis depth and the host voice are the moat and AI can't match either at the quality the audience expects.
Putting it together — three example walk-throughs
Three real candidate niches, run through the tree end to end. Watch where each one drops.
Walk-through 1: Personal finance for first-job earners
- Sustainability: ✅ 20 titles in five minutes. Topic well is infinite.
- Buyer intent: ✅ Brokerages, tax software, courses, credit cards. Five purchase categories in 30 seconds.
- Faceless-fit: ✅ Charts, on-screen text, voiceover. Multiple existing channels prove it.
- Monetization: ✅ Premium advertiser pool, no policy issues for general personal finance.
- AI producibility: ⚠️ Doable but fact-check load is real. Plan for a human review pass on every script.
Verdict: Strong candidate. The fact-check load on question 5 is the operational constraint, not a blocker.
Walk-through 2: Unsolved true-crime cases
- Sustainability: ✅ Effectively infinite case backlog.
- Buyer intent: ⚠️ Audience converts on books and streaming subscriptions but not at the rate finance audiences buy financial products.
- Faceless-fit: ✅ Voiceover over moody visuals is the dominant format.
- Monetization: ❌ Limited-ads tier risk, family-of-victim takedowns, sensitive-content flags.
Verdict: Drop. The monetization risk on question 4 is the killer. Skilled creators do make this niche work, but beginners pay the policy tax for 12-18 months before they understand which case angles are safe.
Walk-through 3: General gaming highlights
- Sustainability: ✅ Infinite gameplay footage.
- Buyer intent: ❌ Audience watches for entertainment. Advertisers pay $1-4 RPM.
Verdict: Drop at question 2. Doesn't matter that the format passes the other filters — the RPM ceiling is too low for ad revenue alone to be the business model. Gaming works for sponsorships and merch, not for faceless channels relying on AdSense.
Common mistakes when running the tree
Four ways beginners make this tree produce wrong answers:
- Running questions out of order. The order is deliberate. Sustainability is first because it's cheapest to fail. Buyer intent is second because it's the largest income lever. AI producibility is last because it's the one that requires tooling investment to test. Skipping ahead wastes evaluation time on niches that fail earlier filters.
- Rescuing niches that fail. If a candidate fails one question, drop it. Don't try to invent a sub-niche that might save it unless the sub-niche is genuinely different. "What if I niche down to celebrity true crime" is still the same monetization failure.
- Picking based on personal interest only. You will lose interest in any niche by month three to six. The channel still has to post. Pick a niche you can tolerate for 100 videos in topic well, not one that you find personally fascinating today.
- Skipping the monetization check because you "saw a channel doing well" in that space. The channel you saw is either grandfathered into ads from before policy changes, monetizing via sponsorship and merch (not AdSense), or sitting on the wrong side of the limited-ads line and losing revenue you don't see. Run question 4 anyway.
How ShortsFast fits once you've picked a niche
The tree above gives you a niche. The next problem is shipping 100 videos in it before the algorithm decides you're real. That gap — from "I have a niche" to "I have a back catalog YouTube can rank" — is the gap most faceless channels fail in. ShortsFast is built to close it: pick a niche, the system writes the scripts in that niche's tone, picks the B-roll, dubs the voice, renders the cuts, and uploads to YouTube Shorts and TikTok on the schedule you set. See how it stacks up against the other faceless tools in the faceless YouTube tool comparison.
Related guides
- 13 best faceless YouTube channel examples in 2026 — once you have a niche, pick a format to model.
- Best faceless YouTube niches in 2026 — ranked by real RPM data — the niche-by-niche RPM data behind question 2 above.
- How to start a faceless YouTube channel in 2026 — what to ship after the niche is chosen.
- Faceless YouTube channel pre-flight checklist (coming next).