Related feature
Produces the script that metadata generation reads. Script keywords become title keywords.
Back to features
Feature page
SEO Metadata that writes itself from the script you already approved.
SEO metadata is the structured information attached to a YouTube video — title, description, tags, chapters, and hashtags — that determines how search engines and recommendation algorithms discover your content. In Outbox, metadata generation is stage 8 of 9, turning your finalized script into a complete, publish-ready metadata package.
TL;DR: Outbox reads your finalized script and generates a complete YouTube metadata package — title, description, tags, chapters, Shorts hooks, and engagement question — in one pipeline stage. No spreadsheet keyword research. No copy-pasting timestamps. No guessing which tags matter.
Keyword-front-loaded titlesDuration-aware chaptersStage 8 of 9Multi-provider LLM
Pipeline stage
Metadata is stage 8 of 9.
Active
01
Analyze
02
Script
03
Voiceover
04
Align
05
Captions
06
Edit
07
Render
08
Metadata
09
Publish
Workflow
What does YouTube SEO metadata actually solve?
Most creators finish editing and then rush through the upload form. Title? Whatever sounds good. Description? A sentence and three links. Tags? Copy them from the last video. Chapters? “I’ll add those later” (they never do).
That metadata gap costs views. YouTube’s search and recommendation systems rely on structured signals to decide who sees your video. Weak metadata means your content competes with one hand tied behind its back.
Manual alternative
- 1Open a keyword research tool (TubeBuddy, VidIQ, Ahrefs).
- 2Search for relevant terms. Compare volume and competition.
- 3Write a title that front-loads the primary keyword.
- 4Draft a description that includes secondary keywords naturally.
- 5Manually list tags — hope they're relevant and not too broad.
- 6Rewatch your video and note timestamps for chapters.
- 7Format chapter timestamps in 0:00 — Label format.
- 8Write a comment-bait engagement question.
- 9Copy everything into the YouTube upload form.
- 10Realize you forgot hashtags. Add three. Publish.
Outbox result
One stage instead of ten disconnected steps.
Analyze -> Script -> Voiceover -> Align -> Captions -> Edit -> Render -> Metadata -> Publish
Mechanics
How SEO metadata generation works in Outbox
01
Read the finalized script
The same script that drove voiceover, captions, and editing.
02
Extract keywords and structure
Identify primary keyword, secondary terms, and content arc.
03
Generate the metadata package
Title, description, tags, chapters, Shorts hooks, engagement question.
04
Assemble YouTube description
Combine workspace links, SEO description, chapters, CTA, and hashtags.
05
Pass to Publish stage
Metadata attaches to your video and ships to YouTube.
Generated metadata
Six fields. One stage. Zero guessing.
Ready to publish
TitleSupabase Auth in Next.js: Complete Setup Guide
DescriptionSet up Supabase authentication in a Next.js 14 app…
Tagssupabase, nextjs, authentication, react
Chapters6 chapters · 0:00 to 11:15
Engagement question
What’s your go-to auth setup for Next.js? Supabase, NextAuth, or Clerk? Drop it below.
Output
What metadata does Outbox generate?
Every pipeline run produces a metadata artifact containing six fields designed for YouTube search and discovery:
| Field | What it does | Example |
|---|---|---|
| Title | Keyword-front-loaded, max 100 characters. Optimized for search intent and click-through. | "Next.js Auth Tutorial: Add Google Login in 12 Minutes" |
| Description | SEO-friendly summary, max 500 characters. Natural keyword placement without stuffing. | "Learn how to add Google OAuth to your Next.js app using NextAuth.js." |
| Tags | Relevant keywords extracted from script content. Used for YouTube search indexing. | nextjs, google-auth, nextauth, oauth-tutorial |
| Chapters | Timestamped section markers from script structure. Duration-aware — skipped under 5 minutes. | 0:00 Introduction · 1:24 Install NextAuth · 3:48 Configure Provider |
| Shorts hooks | Compelling clip segments with start/end timecodes for YouTube Shorts repurposing. | "The auth config trick nobody mentions" — 4:12 to 4:38 |
| Engagement question | Conversational CTA designed to drive comments and boost algorithmic engagement. | "What auth provider do you use? Drop it in the comments." |
Chapters
Smart chapter generation
Chapters are one of YouTube’s strongest SEO signals — they create additional search entry points, improve watch time through navigation, and get extracted by Google for video rich snippets. But most creators skip them because manually timestamping a 15-minute video is tedious.
Outbox generates chapters automatically and adjusts strategy based on video length:
| Video duration | Chapter behavior | Why |
|---|---|---|
| Under 5 minutes | Chapters omitted entirely | Short videos don't benefit from navigation markers. Chapters would feel forced. |
| 5–8 minutes | Optional — only if distinct sections exist | Borderline length. Chapters help if content shifts meaningfully. |
| 8+ minutes | Required — one chapter per ~5 minutes | Longer videos need navigation. First chapter always starts at 0:00. |
Override per workspace with three chapter preferences: off (never generate), on (always generate), or auto (let Outbox decide).
Advantage
Why script-based metadata beats post-production guessing
Most metadata tools work backwards. You upload a finished video, then a tool analyzes the audio or asks you to describe the content manually. By that point, you’ve lost the richest signal: the script.
Outbox flips the order. Metadata generates from the same script that produced your voiceover, captions, and editing — the full-resolution text of what your video actually says.
| Approach | Signal quality | Result |
|---|---|---|
| Manual keyword research | Based on what you think the video is about | Titles and tags may not match actual content |
| Post-upload AI tools | Based on audio transcription or brief description | Lossy — transcription errors, missed context |
| Outbox script-based generation | Based on the full, approved script text | Titles, descriptions, and tags reflect what the video actually covers |
Assembly
The YouTube description assembly
Outbox doesn’t just generate metadata fields — it assembles them into a formatted YouTube description block that’s ready to publish. Your workspace settings control the links section — add your website, social profiles, or affiliate URLs once, and they appear in every video’s description with UTM parameters attached.
Assembled YouTube description
🔗 Important links
https://outbox.run?utm_source=youtube
https://outbox.run/docs/getting-started
https://outbox.run/docs/getting-started
📝 About this video
Set up Supabase authentication in a Next.js 14 app…
📋 Chapters
0:00 Introduction
1:32 Install Supabase client
3:45 Configure auth providers
6:18 Add server-side sessions
9:02 Write RLS policies
1:32 Install Supabase client
3:45 Configure auth providers
6:18 Add server-side sessions
9:02 Write RLS policies
💬 What’s your go-to auth setup?
#supabase #nextjs #authentication
Configuration
Workspace-level metadata settings
Different channels need different metadata strategies. Outbox lets workspace admins configure three settings that shape how metadata generates across all pipeline runs:
| Setting | What it controls | Example |
|---|---|---|
| Description links | URLs at the top of every YouTube description, with UTM parameters | Your site, docs, social profiles |
| Chapter preference | Whether chapters are always on, always off, or auto (duration-based) | auto — let Outbox decide by length |
| CTA style | Tone of the engagement question — conversational, direct, or playful | conversational |
Set these once per workspace. Every video inherits them. Override per-run when needed.
Comparison
SEO metadata vs. the manual alternative
| Dimension | Manual workflow | Outbox SEO Metadata |
|---|---|---|
| Tools required | Keyword research tool + spreadsheet + YouTube Studio | One pipeline stage |
| Time per video | 15–30 minutes (research, write, format, paste) | Automatic — generates and attaches in seconds |
| Title quality | Varies by creator experience and rush level | Keyword-front-loaded, click-optimized, under 100 chars |
| Chapter accuracy | Manual timestamp scrubbing — often skipped entirely | Auto-generated from script structure, duration-aware |
| Tag relevance | Copy-pasted from last video or guessed | Extracted from actual script content |
| Consistency across videos | Depends on who uploads that day | Same generation rules, same workspace config, every run |
| Shorts repurposing | Rewatch, find hooks, note timestamps manually | Shorts hooks generated with timecodes automatically |
| Description formatting | Cobbled together from copy-paste templates | Assembled from links + description + chapters + CTA + hashtags |
Signals
YouTube SEO signals that actually matter
Not all metadata carries equal weight. Here’s what YouTube’s algorithm prioritizes — and how Outbox addresses each signal:
| Signal | Why it matters | How Outbox handles it |
|---|---|---|
| Title keyword placement | YouTube weighs the first 3–5 words heavily for search ranking | Titles are keyword-front-loaded by default |
| Description keyword density | YouTube indexes the first 200 characters for search | SEO description leads with primary keyword |
| Chapter markers | Create additional search entry points; Google extracts for rich snippets | Auto-generated from script with keyword-rich labels |
| Tag relevance | Tags help YouTube understand category and suggest related videos | Tags extracted from actual script content |
| Engagement signals | Comments boost algorithmic ranking. First 24 hours matter most. | Engagement question drives comment responses |
| Hashtags | YouTube displays up to 3 hashtags above the title | Auto-generated from tags, formatted with # prefix |
Repurposing
Shorts hooks: built-in repurposing
Every metadata run identifies the most compelling moments in your video and marks them as Shorts hooks — complete with start and end timecodes.
Instead of rewatching your 15-minute tutorial to find a 30-second clip worth sharing on Shorts, Outbox flags them for you:
Shorts hook
“The auth config trick nobody mentions”
4:12 → 4:38·26 seconds
Shorts hook
“Why your RLS policies are wrong”
9:02 → 9:28·26 seconds
These hooks become the starting point for repurposing workflows — cut, publish to Shorts, drive traffic back to the full video.
Pipeline
How metadata fits the full production flow
Related feature
Voices the script. Video duration from the voiced track informs chapter density rules.
Related feature
Generates timed subtitles from the same script. Caption text complements SEO description content.
Related feature
Consumes the metadata package and attaches it to the YouTube upload.
Related feature
Controls workspace-level metadata settings — description links, chapter preferences, CTA style.
Stage isolation
If you edit a script mid-pipeline, Outbox re-runs from the point of impact. Metadata regenerates from the updated script — new title, new chapters, new tags. You are not manually reconciling stale metadata with a new script.
Audience
Who uses SEO metadata in Outbox?
Faceless YouTube channel operators
You publish 10–15 videos per week across multiple channels. Writing unique, keyword-optimized metadata for every video doesn't scale. Outbox generates it from the script — consistent quality, zero per-video effort.
SaaS founders shipping product content
You recorded a demo of your new feature. The video editing is done. Now you stare at the YouTube upload form and write a title that's just the feature name. Outbox generates "Next.js Auth Tutorial: Add Google Login in 12 Minutes" instead of "Auth Feature Demo."
Developer advocates building tutorial libraries
Your channel has 200 tutorials. The first 50 have great metadata because you had time. The last 150 have "Tutorial — [Feature Name]" as the title. Outbox generates keyword-front-loaded titles and chapters for every video, regardless of volume.
Agencies managing client channels
Fifteen clients. Fifteen different keyword strategies. Fifteen description templates with different links and CTAs. Outbox workspace settings handle per-channel configuration so your operators don't need to remember which client wants what.
FAQ
Common questions about SEO metadata
What AI model generates the metadata?
Outbox supports multiple LLM providers — currently Claude (Anthropic) and GPT-4.1 Mini (OpenAI). The provider is configurable per deployment. The generation prompt enforces consistent formatting rules regardless of which model runs.
Can I edit the generated metadata before publishing?
Yes. If auto_publish is disabled in your workspace, the pipeline pauses after metadata generation. You review the title, description, tags, and chapters — edit anything — then approve for publish. Most users edit titles the first few runs, then trust the output.
Does Outbox do keyword research?
Not in the traditional sense. Outbox extracts keywords directly from your script content. The result: tags and title keywords that match what your video actually covers, not what a keyword tool thinks you should rank for.
How does Outbox handle videos in different languages?
Metadata generates in the same language as your script. If your script is in German, the title, description, and tags will be in German. Language detection is automatic.
Can different workspaces have different metadata strategies?
Yes. Each workspace has its own description links, chapter preference, and CTA style. Agency teams managing multiple client channels configure each workspace independently.
What if the generated title doesn't match my keyword strategy?
Edit it. The pipeline pauses for review if auto-publish is off. Over time, you'll find the generated titles are close to what you'd write — but faster and more consistent than doing it manually at 11 PM before a deadline.
Does metadata regenerate if I change the script?
Yes. Outbox re-runs from the point of change. If you edit the script, all downstream stages — voiceover, alignment, captions, editing, rendering, metadata, and publish — re-execute with the updated content. Metadata always reflects the current script.
Get started
Raw footage in. Discoverable video out.
Every video ships with search-optimized metadata that would have taken you 20 minutes to write manually. Titles, descriptions, tags, chapters, Shorts hooks, and engagement questions — all generated from the script you already approved.