コンテンツファクトリー Bot

アフター

「新規代理店に、タイマーで7通のオンボーディングメールを送っていました。クリック率は9%。4通に減らしました。」 切り口1のフック案：日数ベースのドリップメールか、ユーザーが止まった時に送るメールか。代理店運営の創業者の質問に約140件の返信が付きましたが、データで答えた人はいません [リンク]。あなたの強み：9月29日のテスト、クリック率9%から31%。 切り口4件とスキップリスト付きの切り口レポートの冒頭です。読み取り専用：投稿、いいね、フォローは一切しません。

アフター

Pack check: no row uses this source. Pack ID: 2026-W46-file-requests. VOICE PROFILE loaded. SOURCE BRIEF Thesis: Clients ask "where is it?" when files have no fixed home. Give each client one place. Ideas: 1. One place per client cuts file requests | Proof: 18 agencies, median 11 a week before the Files tab, 3 after | Type: number 2. The clients were lost | Proof: "I'd assumed clients were lazy. They were lost." | Type: story 3. Four details in the tab | Proof: one per client, sorted by project, "new" badge, last-updated date | Type: example 4. A customer said it plainly | Proof: Maya Chen, Northwind Creative | Type: quote To confirm: how the 18 agencies were chosen; Maya Chen's OK to be quoted beyond the blog X THREAD Hook formula: Specific result 1/ 18 agencies counted every "where is that file?" email from their clients for two weeks. The median was 11 a week. Two weeks after each client got one Files tab, it was 3. (posts 2 to 5: one tab detail each; post 6: takeaway and blog link) X POSTS Post 3 | Confession | 202 chars I'd assumed the clients who kept asking agencies for files were lazy. 18 agencies counted those emails for us. 11 a week. After each client got one place to look, 3. They weren't lazy. They were lost. LINKEDIN POST Hook formula: Cold open | 1,210 chars "My clients stopped asking because they stopped having to." Maya Chen, Northwind Creative, was one of the 18 agencies that counted. NEWSLETTER SECTION Subject line options: 11 file emails a week, then 3 (29) | Why clients ask where the files are (35) | What 18 agencies counted (24) Link line: All 18 agencies' numbers: [confirm link] VIDEO SCRIPT Length: 35 seconds | Spoken words: 84 [0:00-0:03] HOOK | Say: "Eleven emails a week asking where the files are." | On screen: "11 a week" RED NOTE 标题：每周11封“文件在哪”，后来只剩3封 CALENDAR | Mon 2026-11-09 08:30 | LinkedIn | My clients stopped asking because they stopped having to | Retention | Cold open | 2026-W46-file-requests | Ready for review | (8 more rows: thread Tue, video and X post Wed, newsletter and X post Thu, X post and RED Fri, X post Sat) OPEN ITEMS - [confirm] Maya Chen agreed to be quoted outside your blog (LINKEDIN POST) - [confirm] How the 18 agencies were picked (X THREAD post 1) - [confirm link] Blog post (X THREAD post 6, NEWSLETTER SECTION) - Asset needed: screenshot of one client's Files tab (VIDEO SCRIPT, RED NOTE) - Anti-AI fixes made: 5 Approve adding these 9 rows to Content calendar as Ready for review?

アフター

PERFORMANCE REVIEW Week: 2026-W41 | Pack: 2026-W41-chatbot | Shipped: 8 of 9 (Sat X post killed: "too close to the thread") | Time per pack: 52 min (target under 45) | Date | Channel | Name | Hook formula | Impr. | Replies | Saves | Clicks | Depth per 1k | | Mon 10-05 | LinkedIn | We turned off our support chatbot | Confession | 14,200 | 61 | 88 | 0 | 10.5 | | Tue 10-06 | X thread | Our chatbot wrote 200-word answers | Curiosity detail | 9,800 | 34 | 41 | 57 | 7.7 | | Wed 10-07 | X post | Most features need to be shorter | Contrarian | 3,100 | 4 | 6 | 0 | 3.2 | | Wed 10-07 | Video | A 200-word answer to a 40-word doc | Curiosity detail | 2,400 | 9 | 12 | 0 | 8.8 | | Thu 10-08 | X post | Tickets went up 6%. Churn didn't move | Specific result | 4,600 | 18 | 22 | 0 | 8.7 | | Fri 10-09 | X post | A client asked why we have no chatbot | Cold open | 5,200 | 22 | 19 | 0 | 7.9 | | Fri 10-09 | RED | 我们把客服机器人关掉了 | Confession | 1,900 | 11 | 45 | 0 | 29.5 | Median depth per 1k (excluding newsletter and RED): 8.3, last week 6.1 Newsletter: 212 / 3,080 = 6.9%, 4-week median 5.4% RED: compared only with RED. 29.5 vs 21.0 last week. Friday pieces are 8 hours old or less; their numbers will still move. Top piece: LinkedIn, Monday. Data: most replies and saves; 23 of 61 replies asked what replaced the chatbot. Guess: Confession hook, with the 6% in the first three lines. Bottom piece: Wednesday X post. Data: lowest depth, and the only post with no number. Guess: an opinion without proof reads as a platitude. Questions from replies: 1. "What did you replace the chatbot with?" (23 replies, LinkedIn) 2. "Did you track CSAT?" (4 replies, thread) 3. "How short should a help doc be?" (3 replies, thread) Next week: - Do more: Confession hooks with a number in the first three lines - Do less: opinion posts with no number. Add one or skip the slot - Test one thing: newsletter link in the thread's last post as well as the reply; compare with this week's 57 clicks Approve writing this review to Notion?

アフター

REPLY DRAFTS 1. Link: [link] | Author: Priya Nair, Head of CS at Orbit What they said: asked what replaced the chatbot Draft: "Mostly shorter docs. The invite-a-client doc is 40 words, and that was always the real answer. [confirm: anything else that replaced it]" Intent: answer | Risk: needs [confirm] 2. Link: [link] | Author: Jun Park, founder of Lumen What they said: asked whether you tracked CSAT Draft if yes: "We did. CSAT went from [confirm] to [confirm] over the same three weeks." Draft if no: "We didn't track it then. Fair hole in the story." Intent: add proof | Risk: needs [confirm] 3. Link: [link] | Author: Marco What they said: the bot was badly configured Draft: "Maybe. But the doc it summarised was 40 words. Any setup that turns 40 words into 200 is solving the wrong problem." Intent: push back | Risk: none Skipped: 1 (spam offer) Post these yourself. Nothing was liked, replied to or followed.