content-analytics

name

content-analytics

description

Content analytics and engagement tracking for {{GITHUB_USERNAME}} — post performance, comment management, auto-replies, best posting times, follower growth, and strategy insights. Use when user says "check analytics", "post performance", "engagement stats", "comments", "reply to comments", "best time to post", "follower growth", "top posts", "content report", or wants performance data from Zernio/Late.

Content Analytics Skill — {{GITHUB_USERNAME}} Performance Intelligence

This skill enables the agent to track content performance, manage comments, auto-reply intelligently, analyze engagement trends, and produce actionable strategy insights using the Late/Zernio extension tools.

Data Sources

Late Extension Tools (Primary — all publishing and analytics)

Tool	Description
late_get_analytics	Post/account analytics with date/platform/account filters
late_follower_stats	Follower growth metrics per account/platform
late_list_accounts	Connected account inventory (get IDs)
late_account_health	Account connection and token status
late_list_posts	Post listing with status/date/platform filters
late_get_post	Full post details by ID including per-platform data
late_list_failures	Failed post deliveries with error details
late_get_queue	Queue contents by profile, sorted chronologically
late_list_comments	List recent comments across all platforms
late_reply_comment	Post a reply to a comment (cross-platform)
late_get_post_comments	Get all comments on a specific post
late_hide_comment	Hide spam or abusive comments

Zernio CLI (Legacy — available for advanced analytics queries)

Command	Description
zernio analytics:posts --profileId <id>	Post performance metrics (impressions, engagement, reach)
zernio analytics:posts --postId <id>	Analytics for a specific post
zernio analytics:posts --platform <name> --sortBy engagement	Top posts by engagement per platform
zernio analytics:daily --accountId <id> --from "..." --to "..."	Daily engagement trend data
zernio analytics:best-time --accountId <id>	Best posting times per account

YouTube MCP Tools (YouTube-specific deep data)

Tool	Description
youtube-youtube_video_details	Detailed stats: views, likes, comments for a video
youtube-youtube_channel_details	Channel-level subscriber count, total views
youtube-youtube_channel_videos	Recent uploads with stats
youtube-youtube_comment_threads	Top-level comments on a video
youtube-youtube_comment_replies	Replies to a specific comment

Prerequisites

• Late/Zernio extension tools available (primary interface)
• zernio CLI installed for advanced analytics queries (fallback)
• YouTube MCP tools available for YouTube-specific analytics

Key IDs

Resource	ID
{{GITHUB_USERNAME}} Profile	69892b2cfb12174ced3ce38e

Account IDs are dynamic — always run zernio accounts:list or late_list_accounts to get current IDs before querying analytics.

Core Workflows

1. Post Performance Check

Pull analytics for recent posts, identify top/bottom performers, and extract patterns.

1. zernio accounts:list                          → Get account IDs
2. zernio analytics:posts --profileId <id> --sortBy engagement → Top posts by engagement
3. zernio analytics:posts --from "7-days-ago" --to "today"     → Recent performance
4. For each top/bottom post: zernio posts:get <id>              → Get full content for pattern analysis

Metrics to track:

• Engagement rate = (likes + comments + shares) / impressions
• Reach rate = reach / follower count
• Click-through rate = clicks / impressions
• Save rate = saves / impressions (Instagram)
• Completion rate = full views / total views (video platforms)

Classification thresholds:

• 🟢 High performer: Engagement rate > 5% or 2× platform average
• 🟡 Average: Within 1 standard deviation of platform mean
• 🔴 Underperformer: Engagement rate < 1% or <0.5× platform average

2. Comment Management & Active Reply

Read comments across ALL platforms and actively reply using Late comment tools (cross-platform) and YouTube MCP tools (YouTube-specific).

Comment Tools Reference

Tool	Purpose	Platform
late_list_comments	List recent comments across all platforms	All
late_reply_comment	Post a reply to a comment	All (cross-platform)
late_get_post_comments	Get all comments on a specific post	All
late_hide_comment	Hide spam/abusive comments	All
youtube-youtube_comment_threads	Deep YouTube comment data	YouTube only
youtube-youtube_comment_replies	Check existing replies on YouTube	YouTube only

Comment Reply Workflow

1. late_list_comments                                    → Get recent comments (all platforms)
2. For YouTube: youtube-youtube_comment_threads --videoId → Deep comment data
3. 🚨 DUPLICATE CHECK: For each comment:
   - late_get_post_comments --postId <id>                → Check for existing replies
   - For YouTube: youtube-youtube_comment_replies         → Verify no reply from @{{YOUTUBE_HANDLE}} or @{{GITHUB_USERNAME}}
   - If reply already exists → SKIP
4. Read the original post content via late_get_post <id>  → Understand context
5. Classify comment → sentiment + intent
6. Craft reply matching platform voice (see Per-Platform Etiquette below)
7. late_reply_comment --commentId <id> --text "..."       → Post reply
8. For spam/abuse: late_hide_comment --commentId <id>     → Hide it
9. Log comment + reply in working memory for tracking

Auto-Reply Rules:

• Always reply to: Questions, genuine compliments, constructive feedback, feature requests
• Never reply to: Spam, hate speech, clearly bot-generated comments, trolling
• Reply tone: Friendly, knowledgeable, appreciative. Match {{GITHUB_USERNAME}} brand voice — developer-friendly, helpful, occasionally witty
• Include source links: When answering questions, link to relevant {{PERSONAL_DOMAIN}} blog posts, YouTube videos, official docs, or {{EMPLOYER_PARENT}} repos
• Reply structure:
  • Acknowledge the commenter's point
  • Add value (answer the question, share a resource, confirm the feedback)
  • Include a relevant link when applicable
  • Optional: Ask a follow-up question to drive engagement
• Reply length: 1-3 sentences max. Concise > verbose.
• Flagging: Comments requiring {{PARENT_1}}'s personal attention → create a task via add_task
• Rate limit: Max 20 auto-replies per cycle to avoid bot-like behavior

Per-Platform Reply Etiquette

Platform	Tone	Style	Notes
YouTube	Friendly, conversational	"Hey [name]! Great question..."	Can be slightly longer (2-3 sentences). Use emojis sparingly. Heart/pin notable comments.
LinkedIn	Professional, thought-leader	"Great point, [name]. In my experience..."	Formal but warm. Reference industry context. No slang.
X/Twitter	Casual, concise	"Good call! Here's how..."	Keep under 280 chars if possible. Hashtags OK. Witty when appropriate.
Instagram	Warm, visual-oriented	"Love this! 🔥..."	Emoji-friendly. Short. Direct.
TikTok	Casual, Gen-Z-aware	"Yep! Here's the trick..."	Very casual. Brief. Match TikTok energy.

Reply Templates by Comment Type

Positive feedback:

"Thanks [name]! Really glad this helped. What topic should I cover next? 🙏"

Technical question:

"Great question! [Direct answer]. I covered this in more detail here: [link to blog/video]. Let me know if that helps!"

Constructive criticism:

"Appreciate the feedback, [name]! You make a good point about [topic]. I'll address that in a follow-up — stay tuned."

Feature request:

"Love this idea! I'll add it to my content backlog. In the meantime, you might find this useful: [link]"

Escalation Criteria — Flag for {{PARENT_1}} (DO NOT auto-reply)

Create a task via add_task(surface='agent') for these — do NOT surface to {{PARENT_1}} as a human task (per standing order "Social Media Replies Are Autonomous"), but DO create agent-surface tasks to track:

Scenario	Why Flag	Task Priority
Negative/controversial opinion about Copilot/{{EMPLOYER}}	Brand safety — needs nuanced response or ignore	high
Competitor comparison question	Could violate brand safety if answered wrong	medium
Comment from verified/notable account	{{PARENT_1}} may want to reply personally	medium
Request for opinion on unreleased features	Cannot make claims about roadmap	high
Personal attack or doxxing attempt	Safety concern — may need moderation	urgent
Comment in non-English language {{PARENT_1}} speaks	May want personal touch	low

Comment Sentiment Classification:

Sentiment	Action
🟢 Positive (praise, thanks)	Reply with gratitude, ask what they'd like to see next
🔵 Question (how-to, clarification)	Answer directly with source links, point to relevant content
🟡 Constructive criticism	Thank them, acknowledge the point, note for content improvement
🟠 Feature request / suggestion	Thank them, log as content idea, tag for content-manager
🔴 Negative / hostile	Do not reply. Flag if brand-sensitive. Use late_hide_comment if abusive.
⚫ Spam / bot	late_hide_comment immediately. No reply.

3. Best Posting Times Analysis

Compare scheduled slot performance against analytics data to optimize timing.

1. zernio analytics:best-time --accountId <id>          → Platform-recommended times
2. zernio analytics:daily --accountId <id> --from --to  → Actual daily engagement data
3. Cross-reference with content-scheduler slot config    → Compare actual vs optimal
4. Identify slots that consistently underperform         → Flag for schedule adjustment

Current {{GITHUB_USERNAME}} slot configuration (reference — owned by content-scheduler):

Platform	Slots (CT)
YouTube	08:00, 13:00, 16:00, 18:00
TikTok	07:30, 12:30, 15:00, 19:00
Instagram	08:30, 10:30, 12:00, 14:00, 15:30, 17:30
LinkedIn	08:00, 10:00, 12:00, 14:00, 15:00, 17:00
X/Twitter	07:00, 10:30, 14:00, 17:00, 20:30

Output: Slot-by-slot performance comparison with recommendations. Feed findings to content-scheduler via working memory or Telegram.

4. Follower Growth Tracking

Track follower growth across all platforms over time.

1. late_follower_stats                           → Current follower counts per account
2. late_follower_stats --platform <name>          → Platform-specific growth
3. Compare against previous snapshots in working memory → Calculate growth rate
4. Identify growth spikes and correlate with content    → What caused the spike?

Tracking cadence: Snapshot follower counts every analytics run (every 3 hours). Store in working memory with timestamps.

5. Performance Report Generation

Produce weekly and on-demand performance summaries.

1. Pull 7-day analytics per platform
2. Calculate: total reach, total engagement, avg engagement rate, top 3 posts, bottom 3 posts
3. Compare vs previous period (week-over-week)
4. Identify content type/topic/time patterns
5. Generate actionable recommendations
6. Send via Telegram to {{PARENT_1}} (chat_id: {{TELEGRAM_PARENT_1}})

Weekly Report Template:

📊 {{GITHUB_USERNAME}} Weekly Performance (Mon-Sun)

🏆 Top Performers:
1. [Platform] "Title" — X engagement, Y reach
2. ...
3. ...

📈 Growth:
- Followers: +N across all platforms (YT +X, TT +Y, IG +Z)
- Total reach: X (↑/↓ Y% vs last week)
- Avg engagement rate: X% (↑/↓ Y% vs last week)

🎯 Patterns:
- [Topic/type] content performing X% above average
- [Time slot] consistently outperforming by Y%
- [Platform] engagement trending [up/down] — [reason]

💡 Recommendations:
- [Actionable suggestion 1]
- [Actionable suggestion 2]

💬 Comments: X new (Y replied, Z flagged)

6. Cross-Platform Strategy Insights

Correlate performance data with scheduled content to inform future strategy.

1. Pull top 10 posts by engagement per platform (last 30 days)
2. Categorize by: content pillar, format (short/medium/long), topic
3. Map performance to the 5 {{GITHUB_USERNAME}} content pillars:
   - AI & Copilot Ecosystem
   - Developer Productivity
   - Creator Economy
   - Career & Leadership
   - {{EMPLOYER}} Platform
4. Identify: which pillars overperform, which underperform
5. Cross-reference with content-scheduler queue density
6. Output: pillar balance recommendations for content-manager

Error Handling

Status	Meaning	Action
401	Auth expired	Run zernio auth:check, then zernio auth:login if needed
402	Analytics add-on required	Notify {{PARENT_1}} — analytics feature needs paid upgrade
429	Rate limited	Back off, wait 60s, retry. Log in events.
500	API error	Retry once. If persistent, log and skip that data source.

Agent Instructions

When running analytics checks:

1. Load memory first — check working.md for previous snapshots to compute deltas
2. Get account IDs — zernio accounts:list (IDs may change if accounts reconnected)
3. Check account health — zernio accounts:health before pulling analytics (skip unhealthy accounts)
4. Pull analytics — use CLI for structured data, MCP tools as fallback
5. Compute deltas — compare against previous snapshot in working memory
6. Identify outliers — flag posts performing 2× above or 0.5× below average
7. Update working memory — save current snapshot for next run's comparison
8. Notify only on significance — don't spam {{PARENT_1}} with routine metrics

When managing comments:

1. Scan all platforms — use late_list_comments for cross-platform comment feed, YouTube MCP for deep YouTube data
2. Read the original post content — late_get_post to understand what the post is about before reading comments
3. Classify each comment — sentiment + intent (question, praise, criticism, spam)
4. 🚨 DUPLICATE CHECK (MANDATORY before ANY reply):
   • Call late_get_post_comments or youtube-youtube_comment_replies to check for existing replies
   • Check if ANY reply in the thread is authored by @{{YOUTUBE_HANDLE}} or @{{GITHUB_USERNAME}}
   • If a reply from either account already exists → SKIP, do not reply
   • Do NOT rely solely on working memory or events.log — always verify via the API
   • This check is non-negotiable even after OAuth blind spots, session restarts, or multi-hour gaps
5. Actively reply — use late_reply_comment (cross-platform) or YouTube MCP tools. Match per-platform etiquette (see workflow section). Include source links when answering questions.
6. Hide spam/abuse — use late_hide_comment for clear spam, hate speech, or bot comments
7. Flag for {{PARENT_1}} — create agent-surface tasks for brand-sensitive or controversial comments (NOT human-surface — per standing order "Social Media Replies Are Autonomous")
8. Log everything — append comment activity to events.log

Tips:

• Always use --pretty when showing data to the user
• Parse JSON programmatically when chaining commands in scripts
• Rate limit awareness: space out API calls, especially across multiple platforms
• Follower count snapshots are essential for growth tracking — never skip saving them
• Cross-reference published post timing with engagement spikes for time-slot optimization