SEVEN:THREE
Sign In
PRE-PUBLISH WORKFLOW

From cut to published — one workflow.

SEVEN:THREE predicts your video's Potential Reach across the click, the watch, and the audience fit — then generates the fixes and publishes the optimized cut to YouTube. Long-form and Shorts, same pipeline.

01UPLOAD

Drop in your cut. Video, audio, and transcript ingested together.

02PREDICT

Potential Reach across click, watch, and audience fit — in under a minute.

03OPTIMIZE

Approve AI-recommended edits: cuts, overlays, B-roll, loudness.

04PUBLISH

Ship the optimized cut and metadata directly to YouTube.

STAGE 1 · UPLOAD & ANALYZE

Video, audio, and transcript — analyzed together.

SEVEN:THREE ingests the full cut as a single multimodal signal, not three separate feeds. That's why it can catch things that transcript-only or thumbnail-only tools miss.

  • Drag-and-drop MP4 / MOV
  • Auto-generated transcript with editable overrides
  • Long-form and Shorts detected automatically
VIDEO
  • Thumbnail / first frame
  • Pacing & shot rhythm
  • Visual craft
AUDIO
  • Energy curve
  • Loudness (LUFS)
  • Dead-air detection
TRANSCRIPT
  • Promise-to-payoff
  • Satisfaction signals
  • Chapter arcs
STAGE 2 · PREDICT POTENTIAL REACH

One score. Three stages. No mystery meat.

Potential Reach is a single 0–100 number backed by three named stages you can act on. Every drop is tied to a timestamp, a segment, or a viewer cluster.

  • Earn the click — packaging, thumbnail or first frame, title synergy
  • Keep & satisfy — retention curve, drop-off risks, promise-to-payoff
  • Right viewers — audience fit against your real subscribers
POTENTIAL REACH
74
EARN THE CLICK
82

Packaging over-indexes vs. channel baseline.

KEEP & SATISFY
62

Reveal delivers, but a slow post-hook stall costs viewers.

RIGHT VIEWERS
79

Strong overlap with slow-TV documentary + solo adventure.

See a full sample prediction
STAGE 3 · OPTIMIZE

Three editors, one approval loop.

Every drop-off risk maps to a fix in one of three editors — Cut, Stylize, or Audio Mix. Approve individually or globally, then SEVEN:THREE renders the final cut and re-scores the result.

  • Cut editor — timestamp-precise removes, trims, and pacing tightening
  • Stylize editor — animated captions, brand logo, callouts / lower thirds / quote cards, and a subtle brightness-and-zoom pulse to the music
  • Audio Mix editor — music / voice / SFX lanes with AI music and AI voiceover generation, per-clip volume, fades, loops, and auto-ducking under voice
  • Re-scored preview before you commit
RECOMMENDED ACTIONS
CUT
Remove segment0:12–0:37+58%Tighten pacing1:48–3:18+22%Trim intro0:00–0:08+12%Loop-friendly endingShorts
STYLIZE
Animated captionsonCallout card2:40–2:55+6%Pacing pulse to musicsubtle
AUDIO MIX
AI music bed-8 dBAI voiceover0:00–0:12hookDuck under voiceauto
Re-scored Potential Reach after all approvals: 89 (+15).
STAGE 4 · PUBLISH

Ship the optimized cut directly to YouTube.

SEVEN:THREE generates title variants, description, tags, and chapter markers based on the analysis. You review, tweak if needed, and publish — no leaving the app.

  • Title variants scored for CTR
  • Description and tags generated from transcript signals
  • Chapter markers pulled from the retention curve
  • Scheduled or immediate publish
PUBLISH TO YOUTUBE
TITLE
I Walked 400 Miles to Find a Cabin That Doesn't Exist
CTR score 88 · 2 alt titles saved
DESCRIPTION
31 days. 400 miles. One cabin at the end of a trail that shouldn't be there. A solo hike through Patagonia and what I found when I got there…
CHAPTERS
0:00 · 3:18 · 8:42 · 14:05 · 19:20 · 24:30
SCHEDULE
Thu · 9:00 AM local
FORMAT-AWARE

Same workflow, different scoring for Shorts.

Shorts don't use thumbnails and the retention curve compresses into seconds. SEVEN:THREE swaps the signals automatically — no separate product, no separate account.

Click signal
LONG-FORM
Thumbnail strength
SHORTS
First-frame hook (1–3s)
Retention curve
LONG-FORM
% across full runtime
SHORTS
Compressed across 15–60s
Audience segments
LONG-FORM
Long-form viewer clusters
SHORTS
Shorts-native feed clusters
Global edits
LONG-FORM
Progress bar, chapter beds
SHORTS
Loop-friendly ending, text-card trims
FAQ

Quick answers.

What files can I upload?
MP4 and MOV up to 900 MB for the fast path, larger files via the proxy pipeline. Transcript is auto-generated; you can override.
How long does analysis take?
Most 15–30 minute videos complete in a few minutes. Shorts complete in 30 to 90 seconds.
Do you post to my channel automatically?
Only when you approve. Publish is a one-click confirmation after you've reviewed the metadata and thumbnail.
Do you store my video?
The source cut is stored only for the duration of the analysis and cleaned up after render. Reports and scores stay in your workspace.
Long-form vs Shorts — same workflow?
Same workflow, format-aware scoring. Shorts skip thumbnail analysis and score the first-frame hook instead.

Predict reach before you publish.

Upload a cut, run the prediction, fix what won't work, and publish — all in one workflow.