Skip to content

Proof & data

Real workflows, real numbers.

No fabricated testimonials. Every figure below links to the dataset, the video, or the tool page it came from - check it yourself.

Creator workflow

What 12 faceless creators actually made in month one

Our first-month production dataset, published unedited: every finished video, the style mix, the real per-video cost, and the caveats.

39 / 54
renders completed of those started - 72% success rate
15
failed renders, credits auto-refunded
12
distinct creators, 5 languages
~$1.75
Mini 720p provider cost of a standard 30s video (today's default ladder)
The early adopters aren't making memes. They're making content with a script - the exact use case a faceless channel is built on.
Read the full dataset →

Single video, end to end

A 30-second movie recap from a one-line prompt

The exact demo embedded on the recap-video-maker tool page - prompt, cost and duration published next to the video itself.

1 line
of prompt: recap the story beats of a classic underdog sports movie
3
credits consumed (a standard 30s cinematic video)
30s
runtime, vertical 9:16, captions burned in
5/5
scenes generated - the full director-mode draft
Every tool page with a demo shows its real prompt and credit cost, so the result is reproducible instead of aspirational.
See the video and its prompt →

Automation

Series Mode: a channel that publishes on its own schedule

Series Mode generates a recurring video per topic on a schedule. Each video waits for review unless you turn auto-publish on.

1
scheduled series included from the Starter plan
Review
each video waits for approval only while auto-publish is off
Refund
automatic credit refund on a failed render
TikTok · YT · IG
dashboard publishing, all three platforms
Review every video before publishing, or enable full autopilot - the two modes are explicit, not a buried setting.
Open Series Mode →

Live reliability

How the pipeline performs right now

The 72% first cohort above is history; this is now - recomputed daily from production data. Sandbox jobs and unapproved Director drafts are excluded, nothing is hand-picked, and a render stuck in flight for over two hours counts as failed, not ignored.

PeriodStartedDeliveredFailedSuccess rate
Last 7 days440n = 4 - too few to publish a rate (min 20)
Last 30 days3832684.2%
Last 100 renders100901090%

Generation-time percentiles publish at 100 measured renders - currently at 32 of 100.

Measured from scene submission to finished MP4 - excludes the scriptwriting/voiceover step and any time a Director-Mode draft waits for your review.

Publishing success by platform (last 30 days)

  • TikTokn = 3 - too few to publish a rate (min 20)

What failed, and why (last 30 days)

  • 4× Every scene generation failed
  • 2× Final assembly failed at the render provider

6 of 6 failed renders in the last 30 days were refunded automatically - the refund is written in the same database transaction that marks the render failed.

Computed 2026-10-23 (UTC) - refreshed daily.

Download the anonymized per-render data (CSV) and recompute these numbers yourself - job ids are hashed, nothing else identifies a user or video.

Attribution policy: we publish named customer stories only with the customer's explicit written permission. Until then, every case study on this page is grounded in data you can verify - our open production dataset and the demos on Made with ShortLoom.