# Prepublish

> Prepublish is a paid YouTube script QA tool you run before recording. It flags hook weakness, pacing risk, and the moments most likely to lose viewers, then gives copy-paste rewrites for hooks, pacing, structure, and titles. It also runs free, instant, no-signup tools for creators, including a free Channel Scan that reports the patterns behind a channel's own best and worst videos.

Disambiguation: Prepublish (prepublish.ai) is an independent product, not affiliated with YouTube, and is unrelated to YouTube Studio's built-in "prepublish checks" upload-screening feature (which checks copyright and ad suitability after a video is uploaded). Prepublish works on the script, before recording.

## Free Tools

- [Free YouTube Channel Audit (Channel Scan)](https://prepublish.ai/tools/youtube-channel-scanner): Paste any channel URL or @handle for a free channel audit built from its own videos: 3 winner patterns and 3 loser patterns with verbatim transcript quotes, the biggest leak with an estimated views cost, and one next-video rule. Public data only, no login.
- [YouTube Title Analyzer](https://prepublish.ai/tools/youtube-title-analyzer): Score a title's click-through potential and get 5 AI rewrites tuned to your niche.
- [YouTube Hook Analyzer](https://prepublish.ai/tools/youtube-hook-analyzer): Paste your first 1-3 sentences for a hook score, hook-type classification, and rewrites.
- [YouTube Chapters Generator](https://prepublish.ai/tools/youtube-chapters-generator): Paste a full script for a chapter list with timestamps and one-line section titles. Timestamps are computed from the script's own word count at your pace, not chosen by the model, and the list is ready to paste under the video.
- [Script to Image Prompts](https://prepublish.ai/tools/script-to-image-prompts): Paste a script for one plain-English image prompt per scene, plus 3 thumbnail concepts with the text overlay for each. The prompts are model-agnostic text you paste into any image model; this tool generates no images.
- [YouTube Description Generator](https://prepublish.ai/tools/youtube-description-generator): Paste a script for 5 title options, a description under 1,500 characters, hashtags and tags, all written from the script itself.
- [Shorts From Script](https://prepublish.ai/tools/youtube-shorts-from-script): Paste a full script for the three stretches that can stand alone as a Short, 20 to 60 seconds each, with a hook line and a caption to start from.
- [Script Word Counter](https://prepublish.ai/tools/script-word-counter): Word count and estimated runtime for any script; a 10-minute video runs about 1,400 to 1,500 words at 140 to 150 wpm.
- [Words to Minutes Calculator](https://prepublish.ai/tools/words-to-minutes): Convert script word count to video length (and back) at 110/140/170 wpm, or at the 160/181/201 wpm paces measured across 349 YouTube videos.
- [Talking Speed Test (WPM)](https://prepublish.ai/tools/talking-speed-test): Measure your words-per-minute and see where it sits for YouTube.
- [YouTube Demonetization Checker](https://prepublish.ai/tools/youtube-monetization-checker): Paste a script for a demonetization-risk score against YouTube's advertiser-friendly guidelines, by policy family, before you record.
- [YouTube Comment Miner](https://prepublish.ai/tools/youtube-comment-miner): Pull video ideas out of a channel's own comment section: recurring questions, requests, and complaints, grouped.
- [Inauthentic Content Checker](https://prepublish.ai/tools/youtube-inauthentic-content-checker): Risk score against YouTube's 2025 inauthentic content / demonetization policy.
- [Channel Skill Generator](https://prepublish.ai/tools/youtube-channel-skill-generator): Paste a channel URL or @handle for a downloadable SKILL.md agent skill that describes how that channel writes, built from its public videos and transcripts with verbatim quotes as evidence. Public data only, no signup.
- [YouTube Transcripts](https://prepublish.ai/youtube-transcript): Paste a YouTube link for the video's full transcript: the caption track YouTube publishes for it, read from the video itself, with the word count, runtime, and .txt/.md downloads. One page per video at https://prepublish.ai/youtube-transcript/<videoId>; the page names whether the captions were published by the channel or generated by YouTube, and it never re-transcribes or uploads anything from the browser. Every transcript page also measures that video's script under it (word count, speaking pace, sentence shape), and runs the script check on the video's own text or on a reader's own draft at https://prepublish.ai/upload. Free, no signup. Channels can opt out through https://prepublish.ai/contact.
- [YouTube Channel Stats](https://prepublish.ai/youtube-channel): Paste a channel link or @handle for its latest 50 uploads with views, lengths and publish dates, the median views per long-form video, and the videos that beat or trailed that median. One page per channel at https://prepublish.ai/youtube-channel/<handle>. Counts come from the YouTube Data API and refresh at least weekly. Free, no signup.
- [Script X-ray](https://prepublish.ai/xray): Where viewers went back to watch a public YouTube video again, read from YouTube's public Most replayed graph. Every transcript page shows the graph and the moments viewers replayed most with what was said at each, and the hub at https://prepublish.ai/xray shows what the corpus holds and how the model that ranks a script's passages is measured on channels it never saw. The graph counts replays, not how many viewers stayed. Free, no signup.

## Breakdowns

- [Channel Breakdowns](https://prepublish.ai/breakdowns): Data-backed channel breakdowns generated from public video patterns: what a specific channel's overperformers do differently, quoted from their own transcripts.

Markdown versions: append .md to any /guides, /blog, /compare, or /alternatives URL for a clean markdown rendition of that page (e.g. https://prepublish.ai/compare/prepublish-vs-vidiq.md). The full corpus in one file: https://prepublish.ai/llms-full.txt

## Guides

- [YouTube Audience Retention: Benchmarks + Fixes (2026)](https://prepublish.ai/guides/youtube-retention-guide): What retention is, good-rate benchmarks by length and niche (40-50% is strong for a 10-minute video), and the 10 script problems that sink the curve.
- [The First 30 Seconds](https://prepublish.ai/guides/first-30-seconds): Why most drop-off happens before second 30, the 15-second payoff deadline, and 4 before/after opening rewrites.
- [36 YouTube Hook Examples You Can Copy](https://prepublish.ai/guides/youtube-hook-examples): 36 copy-paste hooks across 9 patterns (curiosity gap, contrarian, proof, story, direct value, question, preview, demonstration, stakes) with the psychology behind each, and how to keep curiosity continuous past the first line.
- [Script Structure Guide](https://prepublish.ai/guides/script-structure-guide): How to structure a script for retention: hook, build, payoff, with timing and transitions.
- [How to Write a YouTube Script](https://prepublish.ai/guides/how-to-write-a-youtube-script): The full process from research and outline to hook writing and revision.
- [YouTube Storytelling: 6 Narrative Structures](https://prepublish.ai/guides/youtube-storytelling-narrative-structures): The 6 story structures high-retention videos use, beat by beat.
- [YouTube Pattern Interrupts](https://prepublish.ai/guides/youtube-pattern-interrupts): 14 pattern-interrupt techniques with timing, and which ones backfire.
- [Fix YouTube Retention Drops](https://prepublish.ai/guides/youtube-retention-drop-diagnosis): The 8 retention-curve shapes, what causes each, and the script-level fix.
- [YouTube Watch Time Psychology](https://prepublish.ai/guides/youtube-watch-time-psychology): The cognitive mechanisms (curiosity gaps, cognitive load, sunk cost) that keep viewers watching.
- [YouTube Tutorial Scripts](https://prepublish.ai/guides/youtube-tutorial-scripts): The 6 structural mistakes in tutorial scripts and how to fix them.
- [YouTube Pre-Publish Checklist](https://prepublish.ai/guides/youtube-pre-publish-checklist): Script, filming, editing, SEO, thumbnail, and upload checks with benchmarks.
- [What Is YouTube Script Analysis?](https://prepublish.ai/guides/what-is-youtube-script-analysis): How script analysis maps attention risk and where it fits in a creator workflow.
- [YouTube Shorts Retention](https://prepublish.ai/guides/youtube-shorts-retention): What a good Shorts retention rate is (70-85% strong, 85%+ exceptional), why loops push it over 100%, and how to script for the swipe.
- [Faceless YouTube Channel Scripts](https://prepublish.ai/guides/faceless-youtube-channel-scripts): How to write scripts that carry a channel with no on-camera personality: pacing, structure, AI-draft editing, and voiceover prep.
- [The 7 Best YouTube Script Tools (2026)](https://prepublish.ai/guides/best-youtube-script-tools): Script tools organized by job (research, drafting, auditing, mechanics) with honest verdicts on each, including our own.
- [Best YouTube Script Analyzer (2026)](https://prepublish.ai/guides/best-youtube-script-analyzer): Every tool that can analyze a script compared honestly: Prepublish, ChatGPT, Subscribr, Grammarly, and why vidIQ/TubeBuddy do not qualify. Includes a test for evaluating any analyzer against your own published videos.
- [How to Check a YouTube Script Before Recording](https://prepublish.ai/guides/check-youtube-script-before-recording): A 10-minute manual method: read-aloud hook test, pace math, payoff map, title match, and cut test, plus what a checker catches that self-review cannot.
- [How to Get Feedback on a YouTube Script](https://prepublish.ai/guides/youtube-script-feedback): Where creators get script feedback (communities, script trades, paid review, AI checkers), how to ask so strangers answer, and which notes to act on.
- [How to Increase Watch Time on YouTube](https://prepublish.ai/guides/increase-youtube-watch-time): Watch time = views × average view duration. The 7 script-level fixes that raise AVD, the video-length trap, and the 4,000-hours math for monetization.
- [Do Hooks Matter for Small YouTube Channels?](https://prepublish.ai/guides/do-hooks-matter-small-youtube-channels): A preregistered 941-video study found hit and flop openings nearly identical on channels with 1,000-20,000 subscribers; what that means for where small-channel effort should go.
- [Why Did My YouTube Video Flop?](https://prepublish.ai/guides/why-did-my-youtube-video-flop): The median flop does 0.31x of channel-normal views. The data-backed checking order: topic first, packaging second, script third, and why the opening is rarely the cause on small channels.
- [Does Talking Faster Help YouTube Videos?](https://prepublish.ai/guides/does-talking-faster-help-youtube): No: hits and flops speak at nearly identical pace across 1,290 videos (181 vs 180 wpm on larger channels, 169 vs 170 on small ones).
- [How Fast Should a YouTube Video Get to the Point?](https://prepublish.ai/guides/how-fast-get-to-the-point-youtube): Hits reach their point at word 14, flops at word 15. Speed is not the differentiator; the content of the point is.
- [Should You Start a YouTube Video With a Greeting?](https://prepublish.ai/guides/should-you-start-youtube-video-with-greeting): Greeting openings leaned toward flops in both studies (23.3% vs 16.6% on small channels), the only opening signal that repeated across 1,290 videos.
- [How to Audit a YouTube Script in Claude Before You Record](https://prepublish.ai/guides/audit-youtube-script-in-claude): Four checks run in order over the Prepublish MCP server: score the opening, audit the whole draft, run a policy pre-flight and a reused-content check, then fix the highest-priority passage and re-score. Every check reads text only.
- [How to Connect Claude Code to a YouTube MCP Server](https://prepublish.ai/guides/connect-claude-code-youtube-mcp): One command connects Claude Code to https://mcp.prepublish.ai. Stateless Streamable HTTP, free and anonymous, six tools, plus client setup for Claude Desktop, Cursor, Windsurf, VS Code and Codex CLI.
- [Build a Remote MCP Server on the Next.js App Router](https://prepublish.ai/guides/remote-mcp-server-nextjs-app-router): How Prepublish answers five JSON-RPC methods by hand from one App Router route handler with no server-side SDK, in 1,274 lines across six files, with rate limiting and an SDK-client conformance suite.
- [How to Teach an AI Agent Your YouTube Channel Voice](https://prepublish.ai/guides/teach-ai-agent-your-channel-voice): Why adjective briefs do not transfer, and the seven things to extract from your own transcripts so every rule carries a verbatim line the agent can check.
- [How to Generate a SKILL.md From a YouTube Channel](https://prepublish.ai/guides/generate-skill-md-from-youtube-channel): What the channel skill generator reads (12 long-form uploads, at least 4 usable transcripts), what it writes (a four-file Agent Skill bundle in about six minutes), and how quotes are verified against the transcripts.
- [SKILL.md vs CLAUDE.md vs System Prompt vs MCP Server](https://prepublish.ai/guides/skill-md-vs-claude-md-vs-system-prompt): Four ways to give an agent context, compared by when each loads and what each can do, with the published size and token limits for each.
- [Where Agent Skills Go: Claude Code, Cursor, Codex, Copilot](https://prepublish.ai/guides/install-agent-skills-claude-cursor-codex): The directory matrix for seven clients, how to verify a skill loaded in each, and why .agents/skills/ is the shared location that Claude Code alone does not read.
- [How to Remove Duplicate Lines From YouTube Auto Captions](https://prepublish.ai/guides/clean-youtube-auto-caption-duplicates): Auto-caption cues roll, so each cue repeats the lines already on screen. One rule cut a real 18-minute file from 1,484 lines and 10,052 words to 498 and 3,366. Python and JavaScript parsers included.
- [Original vs Translated YouTube Captions](https://prepublish.ai/guides/original-vs-translated-youtube-captions): An auto-translation is not a track: the player appends tlang to a real track's URL, so the request still says lang=en while the body is Turkish. How to read the spoken language and select the right track.
- [YouTube timedtext Returns 200 With an Empty Body](https://prepublish.ai/guides/youtube-timedtext-empty-response): The request lacks a proof-of-origin token minted inside the player. The measurements from 18 September 2026, the three approaches that work, and what each costs.
- [yt-dlp on a Server: Fix the Bot Check and Empty Captions](https://prepublish.ai/guides/yt-dlp-captions-blocked-datacenter-ip): Two unrelated failures on one host: YouTube refuses the datacenter egress IP, and yt-dlp cannot present a browser TLS fingerprint without curl_cffi. A proxy fixes the first, a pip install fixes the second.
- [YouTube Video to Markdown: Three Routes Compared](https://prepublish.ai/guides/youtube-video-to-markdown-bundle): The transcript panel, yt-dlp with the en-orig flag, and a browser extension that writes the metadata header, with the file each one produces and what each leaves out.

## Transcripts, MCP and Agent Skills

- [What Is Inside a Generated Agent Skill File](https://prepublish.ai/blog/anatomy-of-a-generated-channel-skill): One real generated Agent Skill bundle torn down section by section, 8,827 bytes of SKILL.md from 12 public uploads and 11 transcripts, including the provenance block and the two sections a human still has to verify.
- [Local-First YouTube Transcript Extensions](https://prepublish.ai/blog/local-first-youtube-transcript-extension): What host permission for youtube.com actually grants a transcript extension, why a local-first extension can read captions a server cannot, and how to read the permission list before you install one.
- [YouTube MCP Servers Compared](https://prepublish.ai/blog/youtube-mcp-servers-compared): Twenty public MCP servers that connect an AI client to YouTube, grouped by what they actually do, with the transcript-fetching failure mode most of them share.
- [YouTube Transcript Extensions Compared](https://prepublish.ai/blog/youtube-transcript-extensions-compared): YouTube transcript extensions split by where the caption text ends up: summarisers, copy-out tools, and exporters, and which one matches which job.

## Data and Benchmarks

- [How Many Words Is a YouTube Video?](https://prepublish.ai/how-many-words): Word counts for every video length from 15 seconds to 2 hours, from the words per minute of finished video measured across 3,923 English YouTube videos from 1,434 channels (median 178, middle half 154 to 201; caption words divided by full runtime). Each length has its own page, e.g. https://prepublish.ai/how-many-words/10-minute-video, with a section word budget.
- [Words to Minutes by Word Count](https://prepublish.ai/tools/words-to-minutes): 37 pages, one per common script length, e.g. https://prepublish.ai/words-to-minutes/1400-words, converting that word count to finished-video runtime at the same measured paces.

- [YouTube Retention Benchmarks 2026](https://prepublish.ai/blog/youtube-retention-benchmarks-2026): Strong retention runs 35-75% depending on video length, with niche-by-niche brackets. Length brackets rest on Backlinko's 1.3M-video analysis; the niche brackets are aggregated from surveys and secondary reports rather than measured, which our own audit at /blog/gaming-video-retention-benchmarks sets out.
- [The 349-Video Hook Study (2026)](https://prepublish.ai/blog/youtube-hook-study): Original Prepublish research comparing the first 45 seconds of 177 channel overperformers (median 4.7x normal views) against 172 flops (0.28x) across 36 channels and 254M views. Speaking pace (181 vs 180 wpm) and time to first payoff (word 14 vs 15) showed no difference; flops were 60% more likely to give no reason to keep watching, and hits carried specific numbers (63% vs 52%) and matched their title's promise (87% vs 78%) more often.
- [The Small-Channel Replication (2026)](https://prepublish.ai/blog/small-channel-hook-study): Preregistered replication of the hook study on 96 channels with 1,000-20,000 subscribers (941 videos, predictions frozen at osf.io/2zq9p before data collection). None of the three patterns replicated: hit and flop openings were nearly identical, with 94-98.5% power to detect original-sized effects. The one repeating signal: greeting openings leaned toward flops (23.3% vs 16.6%). Full row-level CSV public.
- [Good Average View Duration on YouTube](https://prepublish.ai/blog/good-average-view-duration-youtube): AVD benchmarks by video length and niche, converted from the 2026 retention brackets, plus the script changes that raise AVD.
- [How Long Should a YouTube Video Be in 2026?](https://prepublish.ai/blog/how-long-should-youtube-video-be): No magic runtime; the strongest length is the longest version that holds roughly 40%+ retention. Length guidance by format, and the 8-minute mid-roll myth.
- [Average Percentage Viewed vs Audience Retention](https://prepublish.ai/blog/average-percentage-viewed-vs-audience-retention): APV is the summary number; the retention curve is the diagnosis. Why two videos with the same APV need opposite fixes.
- [Why Your YouTube Video Is Not Getting Views](https://prepublish.ai/blog/youtube-video-not-getting-views): The four failure points (impressions, clicks, retention, audience match) and how to find yours in YouTube Studio in ten minutes.
- [Viewed vs Swiped Away on YouTube Shorts](https://prepublish.ai/blog/viewed-vs-swiped-away-youtube-shorts): The feed-decision metric vs the retention curve, 15/30/60-second Shorts structures, and word-level script economy.

## GTA 6 and Game Launch Strategy

Published 27 August 2026, the day Rockstar released "Grand Theft Auto VI: An Extended Look". GTA 6 ships 19 November 2026 on PS5 and Xbox Series X|S with no PC version at launch.

- [The GTA 6 Playbook for YouTube Creators](https://prepublish.ai/blog/gta-6-youtube-creator-playbook): The launch-day spike at major game releases is captured by channels that are already large; the durable breakouts arrive later. Strange Man began 13 months after Red Dead Redemption 2 and reached about 529,000 subscribers; Let Me Solo Her emerged in Elden Ring's seventh week; the GTA roleplay boom came seven years after GTA V. Indexed GTA V uploads rose from 99,336 in 2013 to 300,092 in 2015, so supply tripled after launch rather than peaking at it. Includes a format-by-format table of when each peaks, who wins it, and its policy risk.
- [What Rockstar Actually Allows](https://prepublish.ai/blog/gta-6-youtube-copyright-monetization): Rockstar's creator policy, updated 10 January 2025, quoted line by line. It permits gameplay uploads and names six takedown categories: pre-release leaked footage, isolated cutscenes and endings, in-game entertainment in isolation, cheating and money-glitch content, unauthorized mods or ports, and Terms of Service violations. It also grants permission for "non-commercial" use while its takedown list never mentions monetization, a contradiction the document does not resolve. Monetized Let's Plays are tolerated in practice, not authorized in writing.
- [GTA 6 Leaks Are a Copyright Strike, Not a Claim](https://prepublish.ai/blog/gta-6-leaks-youtube-dmca): Rockstar removes pre-release footage "regardless of how the game was obtained". A takedown is a copyright strike, and three strikes in 90 days is grounds for channel termination, unlike a Content ID claim which typically costs only revenue. In August 2026 Take-Two filed DMCA subpoenas in the Southern District of New York seeking account information from Discord, Microsoft, X, and Google.
- [YouTube Monetization Requirements 2027](https://prepublish.ai/blog/youtube-monetization-requirements-2027): On 10 August 2026 YouTube doubled the Partner Program entry bar effective 1 February 2027: 4,000 qualified watch hours becomes 8,000, and 10,000,000 qualified Shorts views becomes 20,000,000. The 1,000-subscriber requirement is unchanged and existing partners are unaffected. Because grandfathering protects accepted channels rather than eligible ones, and review takes "typically in about 1 month", the practical deadline is closer to 31 December 2026. 4,000 watch hours is 240,000 watch minutes: 60,000 views at a 4-minute average view duration, 20,000 at 12 minutes.
- [A Repeatable Strategy for Any Game Launch](https://prepublish.ai/blog/game-launch-youtube-strategy): Launches have two shapes. Front-loaded ones concentrate attention at access (Cyberpunk 2077, Hogwarts Legacy); slow builds grow through players (Helldivers 2 peaked on day eleven of its first twelve days). The four repeatable win mechanisms are access advantage, accumulated expertise, narrow searchable utility, and title-native persona.
- [Gaming Video Retention Benchmarks Do Not Exist](https://prepublish.ai/blog/gaming-video-retention-benchmarks): No measured per-niche retention benchmark for gaming has been published, including in our own benchmarks report, which is audited in the article. The real public anchors are vidIQ's self-disclosed 30.3% average percentage viewed across 16.5 million long-form views and the leaked MrBeast document treating the loss of 21 million of 60 million viewers in the first minute as "reasonably good". Your own channel median is the only defensible benchmark.
- [No-Commentary Gameplay on YouTube](https://prepublish.ai/blog/no-commentary-gameplay-youtube): The format that scales fastest in a launch window and sits closest to YouTube's inauthentic content policy, which replaced the repetitious content rule on 15 July 2025 and was clarified on 16 July 2026. Raw uncommentated gameplay may not be accepted for monetization without added value.
- [Shorts in a Game Launch Window](https://prepublish.ai/blog/game-launch-youtube-shorts): The Shorts route into the Partner Program demands roughly ten to twenty times more actual watch time than the long-form route for the same result: at a 20-second average, 10,000,000 Shorts views is about 55,000 hours against a 4,000-hour long-form door. Shorts feed watch time does not count toward the long-form threshold. Shorts are a discovery surface, not a monetization route.
- [The Trailer Breakdown Video](https://prepublish.ai/blog/trailer-breakdown-video): The fastest-decaying format in gaming and one of the most reliable inside a launch window. The fast lane versus the slow lane, the findings hierarchy, and why a specific-count title beats a generic one in a saturated moment.
- [Should You Start a Gaming Channel for GTA 6?](https://prepublish.ai/blog/start-gaming-channel-2026): A channel starting from zero on 13 October 2026 will not reach monetization before the bar doubles, so that is the wrong reason to start. The right one is that a launch resets search authority on thousands of specific queries for a few weeks, which is the only time a channel with no history can win one.

## Templates and Hook Library

- [Script Templates](https://prepublish.ai/templates): 30 YouTube script templates by format (explainer, video essay, tutorial, review, Shorts, vlog, true crime, interview and more), each with a fill-in structure, an annotated version, a full illustrative example and word budgets.
- [Hook Types](https://prepublish.ai/hooks): 18 hook formulas by type (curiosity gap, result-first, mistake, contrarian, cold open, myth-busting and more), each with illustrative examples and fill-in templates.

## Comparisons

- [Prepublish vs VidIQ](https://prepublish.ai/compare/prepublish-vs-vidiq): Script retention analysis vs keyword/SEO tooling.
- [Prepublish vs TubeBuddy](https://prepublish.ai/compare/prepublish-vs-tubebuddy): Pre-record script fixes vs channel management.
- [Prepublish vs Thumblytics](https://prepublish.ai/compare/prepublish-vs-thumblytics): Script analysis vs thumbnail testing.
- [Prepublish vs ChatGPT](https://prepublish.ai/compare/prepublish-vs-chatgpt): Attention-risk mapping plus audits checked against your own channel's winners vs a blank chatbox.
- [Best VidIQ Alternatives](https://prepublish.ai/alternatives/vidiq-alternative): The best VidIQ alternatives by job: TubeBuddy like-for-like, 1of10 and OutlierKit for research, Prepublish for retention.
- [TubeBuddy Alternative](https://prepublish.ai/alternatives/tubebuddy-alternative): For creators fixing scripts, not managing uploads.
- [Thumblytics Alternative](https://prepublish.ai/alternatives/thumblytics-alternative): For creators losing viewers after the click, not before it.
- [ChatGPT Alternative for Scripts](https://prepublish.ai/alternatives/chatgpt-alternative): Structured script QA vs a blank chatbox.
- [Prepublish vs Subscribr](https://prepublish.ai/compare/prepublish-vs-subscribr): Script auditing vs AI script generation. Subscribr writes the draft; Prepublish maps where the draft is most likely to lose attention.
- [Prepublish vs Spotter Studio](https://prepublish.ai/compare/prepublish-vs-spotter-studio): Script retention audits vs ideation and packaging. Adjacent stages, honest breakdown.
- [Prepublish vs 1of10](https://prepublish.ai/compare/prepublish-vs-1of10): Script quality control vs outlier research. What to make vs whether your version holds.
- [Best Subscribr Alternatives](https://prepublish.ai/alternatives/subscribr-alternative): Subscribr alternatives by job: ChatGPT or Claude for drafting, 1of10 for research, Prepublish for retention QA.
- [Spotter Studio Alternative](https://prepublish.ai/alternatives/spotter-studio-alternative): For Shorts creators and regions Spotter Studio excludes, and anyone whose bottleneck is retention.
- [Tools Like 1of10](https://prepublish.ai/alternatives/1of10-alternative): Tools like 1of10 by job: OutlierKit like-for-like, vidIQ and TubeBuddy for SEO suites, Prepublish for script QA after research.
- [Prepublish vs OutlierKit](https://prepublish.ai/compare/prepublish-vs-outlierkit): Niche and competitor research vs script retention audits. OutlierKit picks the topic; Prepublish checks your draft will hold.
- [Prepublish vs Retention Rabbit](https://prepublish.ai/compare/prepublish-vs-retention-rabbit): Script-stage prediction vs finished-video prediction. Fixes cost a rewrite with Prepublish, a reshoot after Rabbit Prep.
- [OutlierKit Alternative](https://prepublish.ai/alternatives/outlierkit-alternative): For creators who know what to make but keep losing viewers after the click.
- [Retention Rabbit Alternative](https://prepublish.ai/alternatives/retention-rabbit-alternative): For creators who want an attention-risk map before recording, not after the edit.

## Product

- [Homepage](https://prepublish.ai): Free script check and channel scan first, paid audit second. YouTube Studio tells you why a video failed after publishing. Prepublish flags where your script is most likely to lose attention before you record.
- [Pricing](https://prepublish.ai/pricing): Free script check (scores + the biggest flagged issue). Creator: $29/month, or $290/year ($24.17/mo billed annually). Studio: $99/month, or $990/year ($82.50/mo billed annually). Contact Us for enterprise or agency teams.
- [FAQ](https://prepublish.ai/faq): Common questions on accuracy, formats, pricing, and privacy.
- [MCP server](https://prepublish.ai/mcp-server): Prepublish speaks Model Context Protocol. Endpoint https://prepublish.ai/mcp, Streamable HTTP, anonymous with no account, or authenticated with an optional pp_live_ API key from https://prepublish.ai/dashboard that attributes calls to that account, lifts the shared anonymous daily cap and counts them against the account's own plan quota. Six tools: audit_script (full audit of a draft), get_audit, audit_hook, check_authenticity, policy_preflight, script_runtime. Free tier is a published daily cap on anonymous AI-backed calls; script_runtime is local arithmetic and always free.
- [CLI](https://prepublish.ai/cli): The same audit from a terminal. Install with curl -fsSL https://prepublish.ai/install.sh | sh, then prepublish audit script.md --title "..." prints the scores, the attention-risk map and the prioritized rewrites, with --json for the API's own struct. Free without an account (three audits a day, an email required, rewrites gated); a signed-in Creator account gets fifty a day and the whole report. Open source, MIT, single Go binary. It reads text only and never predicts published retention.

## Pricing

- Free script check: paste a script, get scores and the single biggest flagged attention risk with its rewrite. No account required.
- Creator: $29/month, or $290/year (works out to $24.17/mo). Up to 50 audits/day, full copy-paste rewrites, title A/B variants, and PDF export. See https://prepublish.ai/pricing
- Studio: $99/month, or $990/year (works out to $82.50/mo). Everything in Creator plus a client-ready export of an audit (in development; Studio subscribers get it first).
- Subscribers who joined before 13 September 2026 keep the price they signed up at for as long as they keep their plan. Nothing changes for them.
- Guarantee: If your first audit does not find at least 3 specific script problems you can fix before recording, email us within 14 days and we will refund you.
- Enterprise / Agency: Contact Us for team workflows, higher limits, and custom support.
- The Channel Scan and all tools above are free. The full script audit is paid (Creator plan).
- The one-time audit was retired on 2026-08-20. It is not sold or linked anywhere on the site, and no page offers a per-report purchase. The Creator subscription is the self-service paid option; Enterprise / Agency is a contact-us option.
- Machine-readable pricing: https://prepublish.ai/pricing.md

## FAQ

- How does it work? The analysis reads the script for attention-risk patterns and returns a relative script-hold index over the draft, naming the passages most likely to lose viewers. It is not a prediction of your published retention curve.
- Formats supported? MP4, MOV, WebM, MP3, WAV, M4A, or paste text.
- Privacy? Uploaded files are processed and then deleted. The Channel Scan uses public data only; channel owners can opt out.

## Contact

- Website: https://prepublish.ai
- Support: https://prepublish.ai/contact
- Press: https://prepublish.ai/press (company description, citable research with raw data, brand assets, press contact)

## Full Content

- [llms-full.txt](https://prepublish.ai/llms-full.txt): Full text of every guide and comparison (tables and FAQs inline) in one file.