The faceless documentary workflow, end to end, with the time each stage takes

A faceless documentary channel is nine stages wearing the disguise of one decision. People quit at stage six, and they quit because nobody wrote the stages down before they started.

Here they are, in order, with what each one actually demands. Then the same list again, marked with which stages a single pipeline removes and which ones it only appears to.

The nine stages

1. Subject selection. Not "a topic" — a specific claim or story with a shape. "The 1976 Olympics" is a category; "the routine that broke the scoring system" is a subject. This stage is cheap in minutes and expensive in consequences, because everything downstream inherits it.

2. Research. Finding what actually happened, in enough detail to write sixty seconds that are true. The trap is that sixty seconds of narration needs far more research than it contains, and the ratio surprises everyone the first time.

3. Script. Roughly 130–160 words for a 40–60 second piece. The hard part is not writing them, it is cutting the fourth interesting fact because the piece can carry three.

4. Narration. Recording or generating the voice. Fast now, and no longer the bottleneck it was two years ago.

5. Footage sourcing. The stage that breaks schedules. You need material that is relevant, usable, and licensable, and those three conditions rarely apply to the same clip. This is where "I'll post daily" dies.

6. The edit. Cutting footage to narration. Not merely stitching: shot lengths have to follow the sentences, and the difference between an edit and an assembly is entirely in this stage.

7. Captions. Table stakes for vertical. Word-level timing, legible on a phone in sunlight.

8. Metadata. Title, description, tags, thumbnail. Cheap in time, disproportionate in outcome.

9. Publishing and review. Uploading, checking it survived the encoder, and watching it once as a viewer rather than as its maker.

Where the hours actually go

The distribution is lopsided in a way that is not obvious before you have done it.

Stages 1, 3 and 8 are quick and high-leverage. Stage 4 stopped being a bottleneck. Stages 2 and 5 — research and sourcing — consume most of the wall-clock time on any honest accounting, and stage 6 consumes most of the skill.

That is the shape people are actually buying their way out of when they buy a tool, and it explains a common disappointment: a tool that automates stages 3, 4, 6 and 7 while leaving 2 and 5 to you has not made daily posting possible. It has made the fast parts faster.

What one pipeline changes

Ours runs 1 through 8 from a topic, and the honest description is that stage 1 stays yours and stage 9 always does.

Measured across 23 completed renders, the median delivery time was 697.5 seconds — eleven minutes and thirty-eight seconds — from submitted topic to finished file. That is machine time and runs unattended.

One render, fully documented: job b2e4b3eb, a 53.47-second piece on Nadia Comăneci at the 1976 Olympics. Fourteen cuts, a median shot of 3.2 seconds, five separate archives, zero black frames, zero silence, shipped with a media credits file naming each source.

Cost, from the same period's ledger: a median of $0.185 per finished Short across 20 jobs.

Stage	Hand-built	Single pipeline
1. Subject	Yours	Still yours
2. Research	Hours	Automated, verified against sources
3. Script	20–40 min	Automated
4. Narration	Minutes	Automated
5. Footage sourcing	The bottleneck	Automated, sources named
6. Edit	Requires skill	Automated to a measured band
7. Captions	10–20 min	Automated
8. Metadata	10 min	Automated
9. Publish and review	Yours	Still yours

The hand-built column is deliberately vague where we have no measurement of our own. We can time our pipeline; we cannot time yours, and inventing a "saves you 6 hours per video" figure would be the kind of claim this article exists to avoid.

The two stages no tool removes

Subject selection. A pipeline will make whatever you ask for, competently, including the wrong thing. The judgement about what is worth sixty seconds is the part that stays a person's job, and it is also the part that separates channels that grow from channels that merely publish.

Watching it once as a viewer. Not reviewing it as its maker, checking whether the thing you asked for arrived. Watching it the way a stranger scrolling at speed would, which is the only test that catches a video that is technically correct and dead on arrival.

Everything between those two is mechanisable. Neither of those two is, and a workflow that pretends otherwise produces the volume-without-variation pattern that both audiences and platforms have learned to recognise.

A realistic first month

Not thirty videos. A first month that works looks like this:

Week 1: three videos, one subject family. You are calibrating what a good subject looks like in your niche, and three is enough to see a pattern.

Week 2: five videos. Introduce the second subject family. Compare retention between families, not between individual videos.

Week 3: seven videos, dropping whichever family performed worse. This is the week where "daily" starts being physically plausible.

Week 4: daily, if and only if stage 1 has become fast for you. If choosing subjects still takes an hour each, daily posting will produce weak subjects at speed, which is worse than three good ones a week.

The failure mode is starting at daily. The volume is achievable from day one with any decent pipeline; the judgement is not, and volume applied to poor judgement is precisely the pattern that gets a channel classed as mass-produced.

<!-- faq -->

Questions people ask

What tools do I need to start a faceless documentary channel?

Structurally: research, script, narration, footage sourcing, edit, captions and metadata. That can be one pipeline or five separate tools. The decisive question is whether whatever you choose covers research and footage sourcing, because those two consume the most time and most tools leave them with you.

How long does it take to make one faceless documentary Short?

Our pipeline's median is 697.5 seconds — under twelve minutes — from topic to finished file, measured across 23 renders, and it runs unattended. Hand-built, the same video is a multi-hour job dominated by sourcing rather than editing.

What is the hardest part of a faceless workflow?

Footage sourcing. Material has to be relevant, usable and licensable at once, and those three conditions rarely meet in the same clip. Scripting and narration, the parts most tools advertise automating, were never the bottleneck.

Can I really post one video a day?

The rendering can. The subject selection usually cannot, at least not at first — and daily posting built on weak subjects produces exactly the volume-without-variation pattern platforms treat as mass production. Build up over four weeks rather than starting at daily.

What does a faceless documentary Short cost to produce?

A median of $0.185 per finished Short in our ledger of 20 jobs, ranging from $0.020 to $0.414 depending mostly on how much sourcing the subject demanded.

Which parts should stay manual?

Choosing the subject, and watching the finished video once as a viewer rather than as its maker. A pipeline will competently produce the wrong video if you ask for it, and it cannot tell you that a technically correct piece is dead on arrival.