@hienlh/ppm 0.9.81 → 0.9.83

package/.opencode/skills/ads-management/references/audience-targeting.md

@@ -1,97 +0,0 @@
-# Audience Targeting Guide
-
-## Targeting Types
-
-### 1. Demographic
-- Age, Gender, Location
-- Income, Education
-- Parental status
-- Home ownership
-
-### 2. Interest-Based
-- Topics and categories
-- Page likes/follows
-- Content engagement
-- Purchase behavior
-
-### 3. Behavioral
-- Purchase history
-- Device usage
-- Travel patterns
-- Life events
-
-### 4. Custom Audiences
-- Website visitors (pixel)
-- Customer lists (email/phone)
-- App users
-- Video viewers
-- Engagement (social)
-
-### 5. Lookalike/Similar
-- Based on custom audiences
-- 1% = most similar
-- 1-10% range available
-
-## Platform-Specific Targeting
-
-### Google Ads
-- Keywords (search intent)
-- In-market audiences
-- Affinity audiences
-- Remarketing lists
-- Customer Match
-
-### Meta Ads
-- Detailed targeting (interests)
-- Custom audiences
-- Lookalike audiences
-- Advantage+ (AI)
-
-### LinkedIn
-- Job title, Function
-- Company, Industry
-- Skills, Groups
-- Seniority level
-
-### TikTok
-- Interest categories
-- Behavioral targeting
-- Custom/Lookalike
-- Creator audiences
-
-## Audience Strategy
-
-### Funnel-Based Targeting
-```
-TOFU (Awareness)
-├── Broad interests
-├── Lookalike 5-10%
-└── In-market audiences
-
-MOFU (Consideration)
-├── Engaged visitors
-├── Lookalike 1-5%
-└── Specific interests
-
-BOFU (Conversion)
-├── Website visitors
-├── Cart abandoners
-├── Customer lookalike 1%
-└── Email lists
-```
-
-## Exclusions (Critical)
-
-- Existing customers (if acquisition)
-- Converted users
-- Irrelevant locations
-- Competitors' employees
-- Low-quality placements
-
-## Testing Approach
-
-1. Start with 2-3 audiences
-2. Same creative across audiences
-3. Sufficient budget per audience
-4. Run 7-14 days minimum
-5. Scale winner, pause losers

package/.opencode/skills/ads-management/references/campaign-setup-and-bidding.md

@@ -1,88 +0,0 @@
-# Campaign Setup & Bidding Strategies
-
-## Google Ads Campaign Types
-
-| Type | Best For | Min Data |
-|------|----------|----------|
-| Search | High-intent, bottom-funnel | Keyword list + negatives |
-| Display | Retargeting, awareness | Creative assets |
-| Video | YouTube reach | 15-30s video |
-| Performance Max | Full-funnel, e-com/lead gen | 30+ conv/mo, product feed |
-| Demand Gen | Top-funnel social-style | Strong video/image creative |
-
-### Search Campaign Setup
-- Use Responsive Search Ads (RSA): 15 headlines, 4 descriptions min
-- Tight ad groups: 3-5 closely related keywords per group
-- Enable all assets: sitelinks, callouts, structured snippets, images
-- Match types: Exact + Phrase for control; Broad only with Smart Bidding + data
-
-### Performance Max Setup
-- Needs 30+ conversions/month (50+ ideal)
-- Add campaign-level negative keywords (up to 10,000)
-- Provide diverse assets: headlines, descriptions, images, logos, videos
-- Create intentional video (auto-generated performs poorly)
-- Separate campaigns by product category/margin tier
-- Audience signals: Customer Match + remarketing lists
-- Learning period: 2-4 weeks; wait for 50 conversions before changes
-
-### Demand Gen Setup
-- Separate prospecting vs remarketing campaigns
-- Tailor creative per placement: short videos for Shorts, carousels for Discover
-- Launch with Maximize Conversions → switch to tCPA after 50+ conversions
-- Allow 60-90 days to stabilize
-
-## Meta Ads Campaign Structure
-
-### Recommended: 2-Campaign System
-- **Campaign A: Creative Testing** — small budget, 3-5 ad sets, ABO
-- **Campaign B: Scaling Winners** — CBO, broad targeting, top creatives only
-
-### Advantage+ Campaigns
-- Best for e-commerce with catalog, $5K+/mo spend
-- Set existing customer budget cap (e.g., 20%)
-- Requires: trained Pixel, product catalog, CAPI
-- Expected: 12-32% ROAS improvement vs manual
-
-### Key Rules
-- Weekly budget ≥ 50× target CPA to exit learning phase
-- 72-hour rule: don't touch campaigns for 3 days post-launch
-- Broad targeting outperforms narrow interest stacking for $5K+/mo accounts
-
-## Bidding Strategies
-
-### Google Ads
-| Strategy | When to Use | Min Data |
-|----------|-------------|----------|
-| Maximize Clicks | New campaigns, no conversion data | None |
-| Manual CPC + eCPC | Testing phase, low volume | None |
-| Maximize Conversions | Have conversion tracking | 15-20 conv/mo |
-| Target CPA | Stable lead gen | 30+ conv/mo |
-| Target ROAS | E-com with values | 50+ conv/mo |
-
-**Rules:**
-- Don't set tCPA/tROAS below actual average
-- Change budget and bid targets separately
-- Allow 4-week stabilization after bid strategy change
-
-### Meta Ads
-- **CBO**: preferred for prospecting (algorithm distributes budget)
-- **ABO**: for retargeting (manual control per ad set)
-- Vertical scaling: increase budget max 20%/day
-- Horizontal scaling: duplicate winning ad sets with new audiences
-- Scaling threshold: ROAS > target for 3+ consecutive days
-
-## Budget Management
-
-- Google: daily budget × 30.4 = monthly ceiling (can overspend 2× daily)
-- Meta: start $20-50/day per ad set, prove CPA, then scale
-- Reserve 20% of budget for testing
-- Retargeting always-on: even $5-10/day pays back
-- 60/40 (Google/Meta) common starting split for e-commerce
-
-## Quality Score (Google)
-
-Three pillars: Expected CTR + Ad Relevance + Landing Page Experience
-- Match keyword → ad copy → landing page theme
-- Mobile page speed critical: 1s delay = ~20% fewer conversions
-- Assets improve CTR → improves Expected CTR component
-- Target QS 7+ on core keywords; 9-10 unlocks lowest CPCs

package/.opencode/skills/ads-management/references/competitor-analysis-and-tools.md

@@ -1,67 +0,0 @@
-# Competitor Analysis & Third-Party Tools
-
-## Free Competitor Research
-
-| Tool | Platform | What It Shows |
-|------|----------|---------------|
-| Meta Ad Library | Meta | All active FB/IG ads by competitor |
-| Google Ads Transparency Center | Google | Active competitor search ads |
-| Google Keyword Planner | Google | Search volume + forecasts (free) |
-
-## Paid Competitor Intelligence
-
-| Tool | Best For | Price |
-|------|----------|-------|
-| SpyFu | Google Ads keyword/ad history | $39+/mo |
-| SEMrush | Full PPC + SEO suite | $99+/mo |
-| PowerAdSpy / BigSpy | Cross-platform creative intel | $49+/mo |
-| iSpionage | Google/Bing ads + landing pages | $59+/mo |
-| Adbeat | Display ad intelligence | $249+/mo |
-
-## Ad Management & Optimization Tools
-
-### Google Ads
-| Tool | Use Case | Price |
-|------|----------|-------|
-| Optmyzr | Automation, QS tracking, bid rules | $66-208/mo |
-| WordStream | Simplified management for SMBs | Custom |
-| DataSlayer | Reporting: Ads → Sheets | $19+/mo |
-| Google Ads Editor | Bulk edits offline | Free |
-
-### Meta Ads
-| Tool | Use Case | Price |
-|------|----------|-------|
-| Madgicx | AI automation, audience segmentation | $49-149/mo |
-| Birch (Revealbot) | Rule-based automation, bulk mgmt | $49-249/mo |
-| AdEspresso | Split testing, campaign creation | $49-259/mo |
-| BestEver.ai | Creative analysis, hook scoring | $99+/mo |
-
-### Attribution & Analytics
-| Tool | Use Case | Price |
-|------|----------|-------|
-| Triple Whale | E-commerce attribution | $129+/mo |
-| Northbeam | Multi-channel attribution | Custom |
-| Hyros | High-ticket/info product attribution | $300+/mo |
-| Motion | Creative performance analytics | $100+/mo |
-
-## Competitor Analysis Workflow
-
-1. **Identify competitors**: search your keywords, check who's bidding
-2. **Audit their ads**: Meta Ad Library + Google Transparency Center (free)
-3. **Keyword gaps**: SpyFu or SEMrush — find keywords they rank for that you don't
-4. **Creative analysis**: PowerAdSpy/BigSpy for cross-platform creative patterns
-5. **Landing page review**: manually visit competitor landing pages; note CTAs, offers, layout
-6. **Budget estimation**: SpyFu/SEMrush provide estimates (not exact, but directional)
-7. **Document findings**: messaging patterns, offers, unique angles, gaps to exploit
-
-## Google vs Meta — When to Use Which
-
-| Scenario | Platform |
-|----------|----------|
-| High purchase intent, active search | Google Ads |
-| Demand generation, brand awareness | Meta Ads |
-| B2B long sales cycle | Meta (nurture) + Google (BOFU) |
-| E-commerce with strong visuals | Both: Meta discovery, Google capture |
-| Niche interest-based targeting | Meta |
-
-**Rule:** Google captures demand. Meta creates demand. Best ROI = both, funnel-aligned.

package/.opencode/skills/ads-management/references/creative-asset-generation.md

@@ -1,110 +0,0 @@
-# Creative Asset Generation for Ads
-
-## AI Image Generation Workflow
-
-### Step 1: Search Prompt Database
-```bash
-python3 .opencode/skills/ai-artist/scripts/search.py "product shot ad" --domain examples
-python3 .opencode/skills/ai-artist/scripts/search.py "lifestyle marketing" --domain style
-```
-
-### Step 2: Generate Images
-
-**Standard (Flash)** — fast iterations, search-based prompts:
-```bash
-.opencode/skills/.venv/bin/python3 .opencode/skills/ai-multimodal/scripts/gemini_batch_process.py \
-  --task generate --model gemini-2.5-flash-image \
-  --prompt "<prompt>" --aspect-ratio <ratio> --size 2K --output <path>
-```
-
-**Creative (Pro)** — unique art direction, complex scenes:
-```bash
-.opencode/skills/.venv/bin/python3 .opencode/skills/ai-multimodal/scripts/gemini_batch_process.py \
-  --task generate --model gemini-3-pro-image-preview \
-  --prompt "<prompt>" --aspect-ratio <ratio> --size 4K --output <path>
-```
-
-### Aspect Ratios by Platform
-| Platform/Placement | Ratio | Resolution |
-|-------------------|-------|------------|
-| Meta Feed | 1:1 | 1080x1080 |
-| Meta Feed (tall) | 4:5 | 1080x1350 |
-| Meta Stories/Reels | 9:16 | 1080x1920 |
-| Google Display | 16:9 | 1200x628 |
-| Google Display (square) | 1:1 | 1200x1200 |
-| LinkedIn Sponsored | 16:9 | 1200x628 |
-| TikTok In-Feed | 9:16 | 1080x1920 |
-
-### Ad Image Prompt Patterns
-```
-Product shot: "Studio photo of [product] on [surface], soft shadow, white background,
-  commercial photography, 4K, no text"
-
-Lifestyle: "Young professional using [product] in [setting], natural light,
-  warm tones, candid, 4K"
-
-Before/After: "Split image: left side [problem state], right side [solution state],
-  clean divider, commercial style"
-
-Social proof: "Flat lay of [product] surrounded by 5-star review cards,
-  modern minimalist, overhead shot"
-```
-
-### Brand Consistency
-- Define style prompt once: `"dark academia, muted tones, film grain"` — reuse across all creatives
-- Use same seed number for cohesive campaign variations
-- Upload brand color hex + logo reference for color accuracy
-
-## AI Video Generation (Veo 3.1)
-
-### Generate Ad Videos
-```bash
-.opencode/skills/.venv/bin/python3 .opencode/skills/ai-multimodal/scripts/gemini_batch_process.py \
-  --task generate-video --model veo-3.1-generate-preview \
-  --prompt "<video_prompt>" --output <path>
-```
-
-### Video Ad Formats
-| Length | Use Case | Platform |
-|--------|----------|----------|
-| 6s | Bumper / brand recall | YouTube |
-| 15s | Awareness, product intro | All |
-| 21-34s | UGC / performance | TikTok, Reels |
-| 30-60s | Demo, testimonial | YouTube, Meta Feed |
-
-### Video Prompt Patterns
-```
-Product demo: "15-second video: hands unboxing [product], close-up details,
-  placing on desk, using it. Warm lighting, modern office. Native audio."
-
-UGC-style: "Casual selfie-style video: person talking to camera about [product],
-  natural lighting, authentic feel, subtle background music."
-
-Before/after: "Split-screen video: left shows [problem], right shows [solution with product].
-  Smooth transition. Professional but approachable."
-```
-
-### Hook Strategies (First 3 Seconds)
-- **Problem hook**: "Struggling with [pain point]?"
-- **Curiosity**: "You won't believe what this does..."
-- **Social proof**: "10,000 customers switched to this"
-- **Visual break**: Bold text overlay, unexpected motion
-- **Direct address**: "Hey [audience], stop scrolling"
-
-## Creative Testing Framework
-
-### Testing Hierarchy
-1. **Hook** — same body, 3 different openers; measure 3s hold rate
-2. **Format** — static vs video vs carousel for same offer
-3. **Angle** — benefit, social proof, fear, urgency
-
-### Fatigue Detection
-- CPM rising + CTR falling + frequency >3 = fatigue
-- Refresh creatives every 2-4 weeks
-- Always have 2-3 new variants in pipeline
-
-### Winning Patterns (2025-2026)
-- UGC/lo-fi outperforms polished brand ads
-- Vertical 9:16 outperforms 1:1 on mobile
-- First-frame text overlay boosts hold rate 15-20%
-- Real faces > stock/AI avatars for trust

package/.opencode/skills/ads-management/references/measurement-and-attribution.md

@@ -1,63 +0,0 @@
-# Measurement, Attribution & Tracking
-
-## Conversion Tracking Setup
-
-### Google Ads
-- Use Google Tag (gtag.js) or GTM — never rely on imported GA4 goals alone for bidding
-- Set up Enhanced Conversions (hash first-party data) — improves accuracy 10-20%
-- Assign conversion values even for leads (avg deal size × close rate)
-- Primary vs secondary conversions: only primary drives bidding
-- For e-com: implement dynamic remarketing tag with product IDs
-
-### Meta Ads (Pixel + CAPI)
-- **Use both** Pixel + Conversions API in parallel — never replace one with the other
-- Deduplication: use `event_id` parameter on both sides; 48h dedup window
-- Target Event Match Quality (EMQ) ≥ 7.0, dedup rate ≥ 90%
-- Send: email (hashed), phone, first/last name, zip, country for max match
-- Priority events: ViewContent → AddToCart → InitiateCheckout → Purchase
-- CAPI setup: Meta native gateway, Shopify integration, or GTM server-side
-- Validate in Events Manager → Diagnostics
-
-## Attribution Models
-
-| Model | Best For | Notes |
-|-------|----------|-------|
-| Data-Driven (DDA) | Google Ads default 2025+ | ML-based credit distribution |
-| First-touch | Brand awareness campaigns | Identifies discovery channels |
-| Last-touch | Short sales cycles, direct response | Simple but biased |
-| Linear | B2B long consideration | Equal credit all touchpoints |
-
-### Post-iOS 14.5 Reality
-- Meta reported ROAS is understated; use 7-day click / 1-day view window
-- Server-side tagging via GTM bypasses cookie/browser restrictions
-- Proper attribution reduces wasted spend by ~27%
-- Multi-touch improves CPA efficiency 14-36%
-
-## Key Metrics
-
-| Metric | Formula | Benchmark |
-|--------|---------|-----------|
-| CTR | clicks / impressions | >1% search, >0.5% display |
-| CVR | conversions / clicks | >2% |
-| CPC | cost / clicks | Varies by industry |
-| CPM | cost / 1000 impressions | Varies |
-| ROAS | revenue / ad spend | >2x minimum, >4x scale |
-| CPA | cost / conversions | ≤ target |
-| MER | total revenue / total ad spend | Holistic metric |
-
-## Measurement Best Practices
-
-- **Triangulate signals**: platform data + GA4 + revenue CRM + incrementality tests
-- **MER (Marketing Efficiency Ratio)**: total revenue / total ad spend — simpler than blended ROAS
-- Use server-side tagging for accuracy
-- Add qualitative data: surveys, interviews, sales feedback
-- Focus on useful/honest reporting over vanity metrics
-
-## Google Ads Scripts for Automation
-
-High-value free scripts:
-- **Budget Pacing** — alert on over/underspend
-- **Quality Score Tracker** — log QS changes to Google Sheets
-- **Broken URLs** — scan for 404 destination URLs
-- **Search Term Harvester** — auto-add converting terms as keywords
-- Source: developers.google.com/google-ads/scripts/docs/intro

package/.opencode/skills/ads-management/references/optimization-playbook.md

@@ -1,83 +0,0 @@
-# Ads Optimization Playbook
-
-## Daily Checks
-
-- [ ] Spend on pace?
-- [ ] Any anomalies in metrics?
-- [ ] Disapproved ads?
-- [ ] Budget depleted early?
-
-## Weekly Optimization
-
-### Performance Review
-| Metric | Action if Low |
-|--------|---------------|
-| CTR < 1% | Test new creative/copy |
-| CVR < 2% | Check landing page |
-| CPC too high | Refine targeting |
-| ROAS < 2x | Pause or restructure |
-
-### Optimization Actions
-1. Pause ads with 0 conversions (100+ clicks)
-2. Increase budget on ROAS > target
-3. Add negative keywords (search)
-4. Refresh creative (fatigue check)
-5. Test new audiences
-
-## A/B Testing Framework
-
-### What to Test
-1. **Creative** (highest impact)
-2. **Audience** (high impact)
-3. **Copy** (medium impact)
-4. **Bidding** (medium impact)
-5. **Placements** (lower impact)
-
-### Testing Rules
-- One variable per test
-- Minimum 100 conversions per variant
-- Run 7-14 days minimum
-- 95% confidence before deciding
-- Document all tests
-
-## Scaling Strategy
-
-### Vertical Scaling
-- Increase budget 20-30% at a time
-- Wait 3-5 days between increases
-- Monitor CPA/ROAS changes
-
-### Horizontal Scaling
-- Duplicate winning ad sets
-- Test new audiences
-- Expand to new platforms
-- Create variations of winners
-
-## Troubleshooting
-
-### High CPC
-- Narrow targeting
-- Improve Quality Score
-- Test different placements
-- Adjust bid strategy
-
-### Low CTR
-- Test new headlines
-- Refresh creative
-- Check audience fit
-- Review ad relevance
-
-### Low ROAS
-- Exclude low-value audiences
-- Focus on bottom-funnel
-- Test higher-value products
-- Review attribution window
-
-## Budget Allocation
-
-| Performance | Action |
-|-------------|--------|
-| ROAS > 4x | Scale aggressively |
-| ROAS 2-4x | Scale gradually |
-| ROAS 1-2x | Optimize first |
-| ROAS < 1x | Pause or restructure |

package/.opencode/skills/ads-management/references/platform-specs.md

@@ -1,64 +0,0 @@
-# Ad Platform Specifications
-
-## Google Ads
-
-### Ad Types
-| Type | Best For | Specs |
-|------|----------|-------|
-| Search | High intent | 3 headlines (30 chars), 2 descriptions (90 chars) |
-| Display | Awareness | 1200x628, 300x250, 336x280 |
-| Video | Engagement | 15-30 sec, 16:9 |
-| Performance Max | Full funnel | Multiple assets |
-
-### Bidding Strategies
-- Manual CPC: Control bids
-- Target CPA: Optimize conversions
-- Target ROAS: Optimize value
-- Maximize Conversions: Volume focus
-
-## Meta Ads (Facebook/Instagram)
-
-### Ad Types
-| Type | Placement | Specs |
-|------|-----------|-------|
-| Image | Feed/Stories | 1080x1080, 1080x1920 |
-| Video | Feed/Reels | 1:1/9:16, 15-60 sec |
-| Carousel | Feed | 2-10 cards |
-| Collection | Mobile | Hero + products |
-
-### Objectives
-- Awareness: Reach, Brand awareness
-- Traffic: Link clicks, Landing page views
-- Engagement: Post engagement, Messages
-- Conversions: Purchases, Leads
-
-## LinkedIn Ads
-
-### Ad Types
-| Type | Best For | Specs |
-|------|----------|-------|
-| Sponsored Content | Awareness | 1200x627 |
-| Message Ads | Direct response | 500 char body |
-| Text Ads | Traffic | 25 char headline |
-| Dynamic Ads | Personalization | Auto-generated |
-
-### Targeting
-- Job title, Function, Seniority
-- Company size, Industry
-- Skills, Groups, Interests
-
-## TikTok Ads
-
-### Ad Types
-| Type | Best For | Specs |
-|------|----------|-------|
-| In-Feed | Engagement | 9:16, 5-60 sec |
-| TopView | Awareness | 9:16, up to 60 sec |
-| Spark Ads | Authenticity | Boost organic |
-| Branded Effects | Interaction | AR filters |
-
-### Best Practices
-- Native content style
-- Hook in first 2 sec
-- Sound-on design
-- Trend participation

package/.opencode/skills/ads-management/scripts/.env.example

@@ -1,13 +0,0 @@
-# Google Ads API credentials
-GOOGLE_ADS_DEVELOPER_TOKEN=your_developer_token
-GOOGLE_ADS_CLIENT_ID=your_client_id.apps.googleusercontent.com
-GOOGLE_ADS_CLIENT_SECRET=your_client_secret
-GOOGLE_ADS_REFRESH_TOKEN=1//your_refresh_token
-GOOGLE_ADS_CUSTOMER_ID=1234567890
-GOOGLE_ADS_LOGIN_CUSTOMER_ID=your_mcc_id
-
-# Meta/Facebook Ads API credentials
-META_APP_ID=your_app_id
-META_APP_SECRET=your_app_secret
-META_ACCESS_TOKEN=your_system_user_token
-META_AD_ACCOUNT_ID=act_123456789