WorkBuddy Practical: 20 - Social Media Growth: Building a Growth Loop

Social Media Growth Loop Workflow

Skills support specific stages, they do not replace human judgment. We examine eight scenarios:

Scenario 1: Tracking Trends, but Undecided on What to Write

Trending lists tell you "what people are reading," but not "why your account should write about it." Chasing trends yields generic copy; relying on gut feeling misses user intent.

• WeChat Popular Search: Analyze popular posts matching your niche.
• Lark/Redbook Hot Notes: Scrape trending notes and data signals.
• Redbook Comments Insight: Extract queries, objections, and gaps from comment sections.
• Inspiration Hunter: Save quick thoughts into a unified inbox.

Prompt Example

Compile a topic pool for "AI Office Automation" this week. Do not write articles.
Gather high-engagement posts from WeChat Official Accounts and Redbook over the past 30 days. Log titles, dates, core promises, structural layouts, and links.
Extract from comments: repeating queries, objections, failures, and raw user quotes.
Align with my account profile: non-technical professionals, focusing on real workflows and validations.
Output 12 topic candidates, each detailing: target reader, problem addressed, existing content gaps, my new evidence, platform fit, and creation cost.
Do not assume high views mean the topic fits my account.

Execution and Outputs

WorkBuddy compiles cross-platform tables, groups comments into query clusters, and rates them for trend, account alignment, new value, and token costs. The deliverable is a topic board for manual selection.

To find viral topics from low-follower accounts, use the viral-topic Skill, which fetches recent viral posts from YouTube and X (Twitter) matching target niches.

Scenario 2: Generating Hooks, Not Clickbait

"Give me 20 viral titles" produces exaggerated promises that do not match the article. A title is a commitment to the reader.

• WeChat Title Generator & Scorer.
• Redbook Copywriter title and tag modules.
• Video Hook Planner.

Prompt Example

Read approved-article.md. Generate titles based only on facts in the text.
Generate: 8 WeChat titles, 8 Redbook titles, 5 video hooks.
For each candidate, output:
1. Target audience; 2. Promise; 3. Verification paragraph in text;
4. Pattern (question, metric, list, case, counter-intuitive);
5. Ratings for credibility, specificity, platform fit, and exaggeration risks.
Filter out unproven metrics, absolute promises, and misleading claims.
Do not pick a final title; let me confirm the promises first.

A/B test single variables (e.g. question vs. outcome titles) instead of changing titles, covers, and publish times all at once.

Use viral-title for WeChat title scoring and optimizations.

Scenario 3: Official Account Covers from Blank Canvas

Covers must convey themes, match resolutions and safe zones, and align with branding. Proclaiming "make a premium cover" yields unrelated graphics or distorted logos.

• WeChat Cover Generator: Analyze competitor styles and propose visual layout drafts.
• WeChat Image Generator: Adjust cover cuts, in-text cards, and follow graphics.
• Image Generation Skills: Render verified visual briefs.

Prompt Example

Create a WeChat cover brief for "Knowledge Archiving: From Clutter to Reusable Assets".
Target reader: Knowledge workers. Core message: from hoarding to reusable flows.
Brand colors: #1677FF, white, black. Exclude purple gradients and sci-fi glowing effects.
Output 3 composition routes mapping: subject, hierarchy, cover text, colors, safe zone check, and matching paragraph. I will confirm before rendering images.
Once rendered, verify text spelling, logo dimensions, and focal points.

Scenario 4: Redbook is More than "Cutting Long Text into Cards"

Converting long articles for Redbook by simply splitting paragraphs and adding emojis yields poor mobile legibility.

• Redbook Cover Designer.
• Redbook Card Renderer: Render structured text into vertical cards.
• Redbook Operations Copilot: Run pre-flight checks and reviews.

Workflow:

1. Extract factual points from long articles.
2. Select a single core question, removing secondary details.
3. Design slide transitions: Cover promise -> Problem resonance -> Solutions -> Examples -> Gaps -> Checklist.
4. Output slide wireframes and text limits before rendering cards.
5. Verify font sizes, margin padding, and line wraps on mobile.
6. Verify numbers and names across titles, copy, and tags.

Convert approved-article.md into an 8-page Redbook card layout. Do not add facts.
Page 1: Core promise; Page 2: Target problem; Page 3-6: One action + one example per page; Page 7: Common mistakes; Page 8: Checklist card.
Return text and visual hierarchy for each slide first.

Scenario 5: Converting Long Articles into Video Scripts

"Convert to 60s script" often yields a rushed reading script lacking camera angles, pauses, visual cues, or asset specifications.

• Short Video Topic Scraper.
• Short Video Script Factory.
• AI Video Director: Compile分镜 storyboard descriptions and prompts.

Prompt Example

Convert this article into a 60-second video script. Target: first-time WorkBuddy users. Goal: explain why task briefs are crucial.
Output timeline table: duration, shot size, camera cues, narration copy, screen text, asset sources, and transitions.
Include hook in first 3 seconds; show product proof within 20 seconds.
Provide a call to action at the end. Specify real, screenshot, or AI-generated assets. Do not fabricate feedbacks.

Scenario 6: Compliance checks and publishing boundaries

• WeChat Restricted Word Filter: Scan for compliance risks.
• WeChat Markdown Formatter: Format and upload to draft folder.
• De-AI Editor: Polish text style and remove filler words.

Scan this article for restricted words and suggest corrections. Evaluate AI-like patterns, polish the text to sound natural, and format the markdown.

Staged workflows are recommended: Fact Check -> Citations & Copyright -> Branding & Compliance -> Links verification -> Mobile preview -> Manual validation -> Publish. Automated comments, bulk direct messaging, and bypassing platform security limits are not recommended.

Scenario 7: Continuous Style Evolution via Reviews

Self-improvement relies on analyzing draft edits against published versions to update style guides. Use Official Account Evolution or Redbook Operations Copilot:

Compare the original draft, the manual edits, and published metrics.
Log facts first, then list 3 testable hypotheses. Do not mistake correlation for causation.
Analyze performance across: topic, title, cover, hook, layout, timing, and channels.
Design 2 single-variable experiments for the next run, defining criteria and limits.
Update style-guide.md with verified rules; do not save temporary trends as permanent rules.

Feed the draft and final published copy to generate retrospective reports and update style-guide.md.

A Baseline Social Media Skill Stack

Sometimes trend discovery is not enough: find low-follower hits too

Search for high-performing posts from smaller accounts as well as general hot topics. Record the account size, publication time, engagement signals, and source links so a low-follower hit is not mistaken for a platform-wide trend.

Acceptance method

Check that every candidate title keeps the verified fact, names a concrete audience or problem, and does not rely on an exaggerated promise. Compare it with the source fact pack before publication.

Is the result usable?

An official-account cover is usable only when the title remains legible on a phone, the visual hierarchy matches the article, required brand elements are present, and no unsupported claim is introduced. A platform draft is ready only after checking format, links, disclosures, and the human publishing decision.