Content Factory Skill
Turn one article or transcript into a week of posts in your voice, not AI voice.
par GetKits
Voyez la différence
Blog post in, on-voice posts out
Avant
Lena Zhou's 1,400-word Fernway blog post: 7 onboarding emails on a timer got 9% clicks. Making them shorter changed nothing. 11 onboarding calls showed two stuck points. 4 emails sent when users stall got 31%.
Après
I made our onboarding emails shorter. Click rate didn't move. Then I listened back to 11 setup calls. Nobody mentioned the emails. They mentioned two screens where they got stuck. I'd been editing the wrong thing. X post 3 of 4 (Confession, 215 chars). The same pack adds a 7-post thread, a LinkedIn post, a newsletter section, a video script and a RED note: 做SaaS的，新手引导邮件我只留了4封
One blog post in, a week of drafts out
Avant
Source: Lena Zhou's 1,400-word Fernway blog post, "We sent 7 onboarding emails. Nobody needed 3 of them." Key facts: 412 trials (Jun 1 to Jul 12) got 7 emails on a day-0 to day-14 timer. 9% average click rate, 22% invited a client in week one. Shortening the emails changed nothing. 11 onboarding calls showed two stuck points. New sequence: 4 emails, each sent when a user stalls. 389 trials over six weeks: 31% click rate, 38% invited a client. A 41-word plain-text note got 44% clicks and 63 replies. Support tickets fell "roughly by a third". Settings: goal newsletter signups, all six channels, week of Oct 12, America/Toronto, no X Premium.
Après
Voice: Lena Zhou, last updated 2026-10-08 | Pack ID: 2026-W42-onboarding-emails SOURCE BRIEF Thesis: Onboarding emails work when each one targets one place users get stuck, sent at that moment. Ideas: 1. Send by behavior, not by day | Proof: 7 timed emails to 4 triggered; clicks 9% to 31%; invites 22% to 38% | Type: number 2. Shortening didn't help, listening did | Proof: "Nothing moved"; 11 calls, two stuck points | Type: story 3. One email, one job, one button | Proof: old welcome had 6 links; 4% clicked past the first | Type: number To confirm: tickets "roughly by a third"; signup link X THREAD Hook formula: Before and after 1/ We sent new agencies 7 onboarding emails on a timer. 9% clicked. We cut it to 4, each sent when someone gets stuck. 31% clicked, and client invites in week one went from 22% to 38%. Below: what didn't work, then each email's one job. (posts 2 to 6: what failed, then one email per post) 7/ Stop scheduling onboarding emails by day. Write one email per place people get stuck. Send it when they get stuck there. The full sequence, including the 41-word note, is in Thursday's Fernway Notes: [confirm link] X POSTS Post 3 | Confession | 215 chars I made our onboarding emails shorter. Click rate didn't move. Then I listened back to 11 setup calls. Nobody mentioned the emails. They mentioned two screens where they got stuck. I'd been editing the wrong thing. RED NOTE 标题：做SaaS的，新手引导邮件我只留了4封 封面文字：4封邮件，点击率31% CALENDAR | Mon 2026-10-12 08:30 | LinkedIn | I assumed our onboarding emails were too long | Onboarding | Confession | 2026-W42-onboarding-emails | Ready for review | | Tue 2026-10-13 09:00 | X thread | 7 timed emails, 9% clicked. 4 triggered, 31% | Onboarding | Before and after | 2026-W42-onboarding-emails | Ready for review | (7 more rows) OPEN ITEMS - [confirm] Support tickets "roughly by a third": left out of every piece until confirmed - Asset needed: screenshot of the 41-word email (VIDEO SCRIPT, RED NOTE) - Anti-AI fixes made: 6
Four samples in, a VOICE PROFILE out
Avant
S1 (X): We lost 14 trial accounts in July. Not to a competitor. To a password reset email that landed in spam. Boring problems kill more startups than big ones. S2 (newsletter): I used to think our onboarding problem was copy. So I rewrote the welcome email four times. Same result. Then I watched six agencies set up Fernway on a call... I'd spent a month on adjectives. S3 (LinkedIn): A client asked me last week why Fernway doesn't have a chatbot. Honest answer: we tried one... Not every feature needs to exist. Most need to be shorter. S4 (X): Agency owners don't want "a world-class client experience". They want the client to stop emailing them asking where the files are.
Après
VOICE PROFILE Name: Lena Zhou (Fernway) Last updated: 2026-10-08 Confidence: medium (4 samples, 3 formats, no video or Chinese sample) Rhythm - Average sentence length: 8 words (range 2 to 24) - Fragments: frequent, used as the turn: "Same result." (S2) Openers - Type: an event with a number or a person - Verbatim: "We lost 14 trial accounts in July." (S1) Closers - Type: one-line principle, under 10 words - Verbatim: "Most need to be shorter." (S3) Signature moves (use at most one per piece) - Self-deprecating cost line: "I'd spent a month on adjectives." (S2) Taboo words - From the user: [ask Lena] - From the anti-AI list, absent from every sample: delve, leverage, journey, seamless, unlock, crucial - Mocked in S4: "world-class" Stances - Small, boring fixes beat big launches (S1, S2) - Shorter beats more (S3) Formatting - Line breaks every 1 to 3 sentences. No lists, emoji, hashtags or bold. Contractions always. Platform adjustments - LinkedIn: one more paragraph of context, ends on a principle, never a question - Video: [unknown] - RED: [unknown, ask for one Chinese sample] Calibration: "Which sounds most like you, A, B or C, and what is off?"
AI-sounding draft, rewritten in voice
Avant
Mode C. Channel: LinkedIn. Lena's VOICE PROFILE loaded. Facts in the draft notes: 7 emails cut to 4, clicks 9% to 31%. Draft from a generic prompt: In today's fast-paced SaaS landscape, onboarding is more crucial than ever. It's not about sending more emails. It's about sending the right ones. Here's what we learned: - Personalization is key - Timing matters - Less is more The result? Engagement went through the roof. What do you think? #SaaS #Onboarding #Growth
Après
REWRITE We cut our onboarding emails from 7 to 4. Clicks went from 9% to 31%. What changed most is timing. Each email now waits until someone gets stuck, instead of firing on day 3 or day 7. Less email, sent when it's needed. Weak spot: the draft never says which email did the work. Add the 41-word note and its numbers if you can. TELLS REMOVED - Opener: "In today's fast-paced SaaS landscape" became "We cut our onboarding emails from 7 to 4." - Banned word: "crucial" removed; the numbers carry the stakes - Negation flip: "It's not about sending more emails. It's about sending the right ones." cut - Triplet list: "Personalization is key / Timing matters / Less is more" became one point, timing, the only one with proof - Colon reveal: "The result? Engagement went through the roof." became "Clicks went from 9% to 31%." - Engagement bait: "What do you think?" removed - Hashtag stack: three hashtags removed (profile: no hashtags)
Contenu du Kit
- content-factory.zip
À propos de ce Kit
Give this skill 3 to 5 things you wrote and it builds a VOICE PROFILE you reuse from then on. Give it one article, transcript or podcast and it returns a week of drafts in that voice: an X thread, four X posts, a LinkedIn post, a newsletter section, a short video script, a Xiaohongshu (RED) note in Chinese and a 7-day calendar. A full pack takes about 10 minutes, and you edit drafts instead of starting from a blank page.
What you get
- The
content-factoryskill folder for Claude, ChatGPT, Codex and Cursor. - Three modes: build a voice profile, build a content pack, and fix a draft that sounds like AI.
- A voice profile template with 18 fields measured from your samples, each one tied to a line you actually wrote, plus a calibration step that checks it with you.
- 13 hook formulas, each with an example, when it works and how it fails, plus a four-question hook test to choose between them.
- Platform formats: limits and structure for X threads and posts, LinkedIn, newsletter sections and short video, and a guide written in Chinese for RED, with 标题 formulas and the words that get notes throttled.
- An anti-AI tells list: more than 70 phrases and words, and 18 patterns readers now spot, each with its fix, in English and Chinese.
- A content calendar template that matches the Notion database used by the Content Factory Bot and Dot.
- A full worked example: one blog post in, the complete pack out, with notes on every judgment call, including the invented details it caught.
Who it is for
- Founders and operators who write from real work and want to post on X and LinkedIn every week.
- Newsletter writers who want each issue to feed a week of social posts.
- Creators who reach Chinese-speaking readers on Xiaohongshu as well as English channels.
- Ghostwriters and content leads who manage someone else's voice and need it to stay consistent.
What it will not do
- Post, schedule or tag anyone. It drafts. You publish.
- Invent numbers, quotes or stories. Anything it cannot trace to your source comes back marked
[confirm]. - Pad a thin source. If there is not enough in it, you get three questions instead of filler.
- Translate your English posts into Chinese word for word. The RED note is written for Chinese readers from the source itself.
- Write long SEO articles, sales pages or ads.
Avis
Aucun avis pour l'instant.
Autres Kits Contenu
Content Factory Kit
One long piece in, a week of on-voice posts out, scheduled in Notion and reviewed weekly.
Content Factory Bot
A Grok Bot that scouts angles on X Monday and drafts next week's posts by Thursday.
Content Factory Dot
Your dot keeps your Notion content calendar full and packs every interview you share.