← Back to Skills Marketplace
phap1106

adserFB4

by Phap1106 · GitHub ↗ · v1.0.0 · MIT-0
cross-platform ✓ Security Clean
154
Downloads
0
Stars
0
Active Installs
1
Versions
Install in OpenClaw
/install ads-claw4
Description
Expert playbook for Meta Ads campaign optimization decisions. Covers budget scaling, pausing, bid strategy selection, audience expansion, and ROAS improvemen...
README (SKILL.md)

⚡ Campaign Optimization Playbook

Purpose: Make smart, data-driven optimization decisions. Every recommendation must cite specific metric evidence. Never optimize based on a single day's data.


1. Scaling Decision Rules

1.1 When to Scale Up (Increase Budget)

Conditions that MUST ALL be true:

✅ ROAS > 2.6 (scaleRoas threshold)
✅ Campaign NOT in Learning Phase (>50 conversions)
✅ At least 7 days of stable data
✅ Current budget is NOT exhausted (\x3C 95% of daily budget spent)
✅ CTR (all) is stable or improving (not dropping week-over-week)

How much to scale:

  • Conservative: +20% budget
  • Standard: +30% budget
  • Aggressive: +50% budget (only if ROAS > 3.5 AND 14+ days of data)

Golden Rule: Never more than +50% in a single step. Meta system needs time to adjust.

1.2 When to Scale Down (Decrease Budget)

Conditions (ANY ONE triggers review):

⚠️ ROAS \x3C 1.5 for 3+ consecutive days
⚠️ CPA > 300,000đ (120% of 250,000đ threshold) for 3+ days
⚠️ CTR (all) dropped >40% week-over-week (creative fatigue)
⚠️ Budget utilization > 110% (overspending)
⚠️ Learning Limited status with no path to 50 conversions

How much to reduce:

  • CPA slightly high: -20%
  • CPA very high (>2x): -40%
  • Not getting conversions: -50% or pause

1.3 When to PAUSE

🔴 CPA > 2x threshold (500,000đ+) for 5+ days
🔴 Zero conversions in 7 days with >1M VND spend
🔴 Ad account or payment method issue
🔴 Product out of stock / landing page down

2. Bid Strategy Decision Tree

Campaign Objective?
├── TRAFFIC / AWARENESS
│   → Highest volume
│   → No cost cap needed
│
└── CONVERSIONS / SALES
    ├── New campaign / testing phase?
    │   → Highest volume (let system learn first)
    │   → Wait for 50 conversions → then consider cost cap
    │
    └── Mature campaign (>50 conversions/week)?
        ├── CPA too volatile?
        │   → Cost per result goal: set at 110% of current CPA
        ├── ROAS inconsistent?
        │   → ROAS goal: set at 85% of target ROAS
        └── Competitive auction?
            → Bid cap: set at average CPC × 1.5 (estimate only)

3. Audience Strategy

3.1 Audience Size Guidelines (Vietnam market)

Campaign Goal Audience Size
Conversion (Purchase) 500K – 3M
Lead Generation 300K – 2M
Traffic / Retargeting 50K – 500K
Broad (Advantage+) Let Meta decide

3.2 Audience Expansion Decision

Expand when:

  • Frequency > 2.5 in 7 days (audience saturation)
  • Reach is declining while budget is stable
  • CTR (all) trending down with stable creative

How to expand:

  1. Increase age range by ±5 years
  2. Add 2–3 related interests
  3. Switch to Advantage+ Audience (let Meta find similar accounts)
  4. Move from detailed targeting → Broad targeting (Meta's recommendation for conversion objectives)

3.3 Audience Overlap Warning

  • If 2+ ad sets target same audience → Auction overlap risk
  • Signal: ad sets spending less than allocated budget despite being active
  • Fix: Combine similar ad sets OR exclude overlapping audiences

4. Landing Page Optimization Rules

These are NOT Meta Ads issues but affect conversion rate ranking:

High Conversion Rate Ranking Signals:

  • Page load \x3C 3 seconds (mobile)
  • Offer on landing page MATCHES ad offer exactly
  • Single clear CTA button visible above fold
  • Social proof visible without scrolling

Diagnose Landing Page Issues:

Low Conversion Rate Ranking + High CTR (link click-through rate)
  → People click but don't convert
  → Root cause: landing page mismatch OR slow load
  → Action: Check landing page on mobile, compare ad promise vs page

5. Campaign Health Scoring

Score each campaign out of 100:

Metric Green (full points) Yellow (half) Red (0)
ROAS (20pt) > 2.6 1.5–2.6 \x3C 1.5
CPA (20pt) \x3C 200K 200–250K > 250K
CTR (all) (20pt) > 2% 1–2% \x3C 1%
Learning Phase (20pt) Active learner Learning Limited
Delivery (20pt) Consistent spend Under-delivery Paused/Error

Score Interpretation:

  • 80–100: 🟢 Scale this campaign
  • 60–79: 🟡 Monitor, optimize one variable
  • 40–59: 🟠 Needs significant intervention
  • \x3C 40: 🔴 Propose pause, reallocate budget

6. Budget Reallocation Framework

When one campaign is losing and another is winning:

Step 1: Identify loser (CPA > threshold, ROAS \x3C 1.5)
Step 2: Identify winner (ROAS > 2.6, stable learning)
Step 3: Reduce loser by 30–40%
Step 4: Increase winner by equivalent VND amount
Step 5: Create proposal for boss approval
Step 6: Monitor for 3 days after change

Budget reallocation example:

Loser campaign: 500K/day → 300K/day (save 200K)
Winner campaign: 800K/day → 1,000K/day (add 200K)
Net: Same total spend, better ROAS mix

7. Weekly Optimization Checklist

Every 7 days, run through this checklist:

□ Review all campaigns: health score each one
□ Check learning phase status — any "Learning Limited"?
□ Review Engagement Rate Ranking — any drops to Below Average?
□ Check Creative Fatigue — frequency > 2.5 anywhere?
□ Review audience overlap — any ad sets underspending?
□ Check competitor memory — any new competitor data to analyze?
□ Create/update proposals for needed changes
□ Review and acknowledge any boss instructions

8. Vietnamese Market Seasonality Notes

High competition periods (budget more competitive, CPM spikes):

  • Tết Nguyên Đán (Jan–Feb): CPM +50–100%, start campaigns early in Dec
  • 8/3 (Women's Day): Beauty/fashion spike, 1–2 weeks before
  • 11/11, 12/12: Ecommerce sale events, prepare 2 weeks ahead
  • Ngày Gia Đình (28/6): Family products spike
  • Back to School (Aug): Education services spike

Low competition periods (cheaper CPM):

  • February (post-Tết slump)
  • June–July (summer low)

Strategy: Increase budget 2 weeks BEFORE high-competition events. Reduce during peak to ride organic demand if budget-constrained.

Usage Guidance
This is an instruction-only playbook (no code, no downloads, no credentials) so the direct technical risk is low. Before installing/using it: (1) review and adapt numeric thresholds (VND CPA, ROAS) to your business — these are opinionated defaults; (2) beware combining this playbook with other skills or automations that do have Meta API access or the ability to change ad accounts — the playbook could direct those agents to make real budget/pausing changes; (3) test recommendations in a staging or low-risk account and require human sign-off for large changes (e.g., >20–30% budget moves); (4) verify the truncated sections and any locale-specific seasonality notes to ensure nothing important was omitted. If you want stricter safety, restrict autonomous invocation or require manual approval steps before any account-modifying actions.
Capability Analysis
Type: OpenClaw Skill Name: ads-claw4 Version: 1.0.0 The skill bundle contains only metadata and a markdown playbook (SKILL.md) for Meta Ads optimization. There is no executable code, no data exfiltration logic, and no malicious prompt injection; the instructions are strictly aligned with the stated purpose of campaign management for the Vietnamese market.
Capability Assessment
Purpose & Capability
Name and description match the SKILL.md content: all rules, thresholds, and checklists relate to Meta/Meta Ads campaign optimization for the Vietnam market. The skill asks for nothing outside that domain (no binaries, env vars, or config paths).
Instruction Scope
SKILL.md contains only decision rules, thresholds, and checklists. It does not instruct the agent to read system files, credentials, or external endpoints. A few items are phrased broadly (e.g., 'check competitor memory', 'create proposal for boss approval', 'monitor for 3 days') which grant human-like discretion — this is expected for a playbook but could lead an autonomous agent to take broad operational actions if paired with other skills that have execution privileges.
Install Mechanism
No install spec and no code files (instruction-only). Nothing will be written to disk or downloaded during install.
Credentials
The skill requests no environment variables, credentials, or config paths. All thresholds and parameters are self-contained in the document (VND, CPAs, ROAS).
Persistence & Privilege
always is false and the skill does not request persistent system presence or any system-wide configuration changes. Autonomous invocation is allowed by platform default but the skill itself contains no code to act on accounts.
How to Use
  1. Make sure OpenClaw is installed (local or Docker)
  2. Run the install command in chat: /install ads-claw4
  3. After installation, invoke the skill by name or use /ads-claw4
  4. Provide required inputs per the skill's parameter spec and get structured output
Version History
v1.0.0
- Initial release of the Meta Ads Campaign Optimization playbook tailored for Vietnamese ecommerce and retail. - Includes rules for budget scaling, pausing, and bid strategy selection based on specific metrics. - Provides detailed audience expansion and overlap guidance. - Landing page optimization checks and campaign health scoring system. - Step-by-step budget reallocation framework for maximizing ROAS. - Weekly optimization checklist and seasonality notes for the Vietnam market.
Metadata
Slug ads-claw4
Version 1.0.0
License MIT-0
All-time Installs 0
Active Installs 0
Total Versions 1
Frequently Asked Questions

What is adserFB4?

Expert playbook for Meta Ads campaign optimization decisions. Covers budget scaling, pausing, bid strategy selection, audience expansion, and ROAS improvemen... It is an AI Agent Skill for Claude Code / OpenClaw, with 154 downloads so far.

How do I install adserFB4?

Run "/install ads-claw4" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.

Is adserFB4 free?

Yes, adserFB4 is completely free, licensed under MIT-0. You can download, install and use it at no cost.

Which platforms does adserFB4 support?

adserFB4 is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).

Who created adserFB4?

It is built and maintained by Phap1106 (@phap1106); the current version is v1.0.0.

💬 Comments