Meta's Andromeda delivery system has fundamentally changed how Advantage+ Shopping Campaigns (ASC) work. The old playbook—micro-segmented audiences, low creative volume, constant tweaking—now actively hurts performance.
This 9-point checklist consolidates learnings from 50+ ASC launches where campaigns that follow it consistently outperform by 25%+.
✅ Checkpoint 01: Account Setup & Signal
Your data infrastructure is the foundation of AI-powered delivery.
Checklist Item | Why It Matters |
|---|---|
☐ Pixel + Conversions API (CAPI) firing and deduplicated | Ensures accurate event tracking without double-counting |
☐ CAPI passing high-value parameters (e.g., transaction value) | Enables value-based optimization and better ROAS targeting |
☐ iOS 14+ domain verification and Aggregated Event Measurement set | Required for tracking iOS users post-ATT |
☐ Attribution window set to 7-day click + 1-day view (current default) | Aligns with Meta's recommended attribution for ASC |
Quick Diagnostic
Events Manager → Data Sources → Your Pixel → Test Events
- Are events firing correctly?
- Is deduplication working (check Event Match Quality)?
- Is CAPI Match Quality above 6.0?✅ Checkpoint 02: Campaign Structure
Consolidation is the new segmentation.
Checklist Item | Why It Matters |
|---|---|
☐ One consolidated campaign per offer (maximum data density) | AI needs 50+ conversions/week per ad set for stable learning |
☐ Advantage Campaign Budget (CBO) turned ON | Lets AI allocate spend to best performers automatically |
☐ Daily budget = 5x your target CPA (to hit 50 conversions/week) | Ensures enough signal for the learning phase |
☐ Combine cold + warm audiences (stop segmenting, start consolidating) | More data = better AI optimization |
Budget Calculator
Target CPA: $50
Required Daily Budget: $50 × 5 = $250/day
Weekly Budget: $1,750
Expected Weekly Conversions: 50+
✅ Checkpoint 03: Audience Settings
Let the AI find your buyers—stop building walls.
Checklist Item | Why It Matters |
|---|---|
☐ Broad targeting ON (Age/Gender/Location only; let the AI find the buyer) | Andromeda's AI outperforms human targeting in 2025 |
☐ Manual exclusions active (Exclude past 180-day purchasers) | Prevents wasting budget on existing customers |
☐ Advantage+ Audience "Suggestions" used only as directional signals | Hints, not walls—AI may go broader |
☐ Avoid narrow interest stacks; they now act as "walls" for the AI | Restricts AI's ability to find optimal audiences |
Recommended Exclusions
Past 180-day purchasers (prevent cannibalization)
Existing email subscribers (if running separate retention campaigns)
High-frequency ad viewers with no conversion (optional, test first)
✅ Checkpoint 04: Creative Strategy
Variety is the new targeting.
Checklist Item | Why It Matters |
|---|---|
☐ 8–15+ distinct creative concepts (variety is the new targeting) | AI needs creative diversity to match different buyer personas |
☐ High visual distinctness (ensure each ad triggers a new Entity ID) | Minor "reskins" get grouped as similar by Meta |
☐ Diverse motivators: Mix Problem/Solution, UGC, and Social Proof | Different audiences respond to different psychological triggers |
☐ Vertical/Square formats (9:16 and 4:5) for Reels-first delivery | Reels placements now dominate delivery |
Creative Mix Template
Type | Quantity | Focus |
|---|---|---|
Problem/Solution | 3-4 | Pain point → Your solution |
UGC Testimonials | 3-4 | Real customer stories |
Social Proof | 2-3 | Reviews, ratings, results |
Lifestyle/Aspirational | 2-3 | Emotional connection |
Educational/Demo | 1-2 | Product functionality |
✅ Checkpoint 05: Ad Copy & Audio
Audio keywords now influence delivery.
Checklist Item | Why It Matters |
|---|---|
☐ 5+ headlines and primary text variations for GEM to test | Generative Expansion Module (GEM) optimizes copy combinations |
☐ Audio Keywords: Use voiceovers with keywords your audience searches for | Andromeda analyzes audio—keywords in voiceovers affect targeting |
☐ Messaging addresses different persona "pain points" | Different motivators for different audience segments |
☐ Hook strength: Clear value or movement in the first 3 seconds | 3-second thumb-stop determines if users watch |
Hook Formula
[PATTERN INTERRUPT] + [SPECIFIC BENEFIT] + [CURIOSITY GAP]
Example: "I stopped spending $500/month on [X]...
and here's what happened to my [desired outcome]"✅ Checkpoint 06: Landing Page Consensus
Ad-to-page consistency is now algorithmically measured.
Checklist Item | Why It Matters |
|---|---|
☐ Semantic Matching: Page copy mirrors the ad's specific promise | Meta checks if landing page matches ad claims |
☐ Mobile page load under 3 seconds | Slow pages = high bounce = wasted spend |
☐ Visual continuity from ad to "above the fold" on the site | Reduces cognitive friction, improves CVR |
☐ Minimal form fields to reduce friction | Each field = 10-15% fewer completions |
Landing Page Audit
1. Does the headline match the ad promise?
2. Is the primary CTA visible without scrolling?
3. Are trust signals (reviews, badges) visible?
4. Is the page mobile-optimized?
5. Does it load in under 3 seconds?✅ Checkpoint 07: Conversion Health
Verify your signal quality before launch.
Checklist Item | Why It Matters |
|---|---|
☐ Purchase/Lead events verified in Events Manager | Ensures you're optimizing for the right action |
☐ Test transaction completed to verify value-based signals | Confirms revenue data is passing correctly |
☐ Deduplication score is high (no double-counting between Pixel/CAPI) | Duplicate events confuse AI optimization |
☐ Monitoring "New Customer Acquisition Cost" (nCAC) specifically | True efficiency metric for growth |
Verification Steps
1. Events Manager → Test Events → Trigger a test conversion
2. Verify event appears within 30 seconds
3. Check that value parameter matches actual order value
4. Confirm deduplication (event should appear once, not twice)✅ Checkpoint 08: Launch & Patience
The first 7 days are critical—don't touch anything.
Checklist Item | Why It Matters |
|---|---|
☐ Start with your calculated 5x CPA budget (avoid "budget-starving") | Under-budgeting prevents learning phase completion |
☐ Strict "No Edit" rule for 7 days (GEM needs a longer sequence window) | Edits reset learning—costs 2-3 days each time |
☐ Document baseline CPA/ROAS for 7-day window | Need comparison point for optimization decisions |
☐ Monitor "Spend Distribution" to see which creative the AI prioritizes | Reveals which creatives AI considers most effective |
Launch Day Checklist
☐ Budget set to 5x target CPA
☐ All tracking verified
☐ Calendar blocked: NO EDITS for 7 days
☐ Baseline metrics documented
☐ Spend distribution monitoring scheduled❌ Common Mistakes to Avoid
Mistake | Why It Hurts |
|---|---|
❌ Micro-segmenting audiences | Fragments data, starves AI of signal |
❌ Making tiny "reskin" tweaks | Meta groups these as "Similar"—doesn't count as new creative |
❌ Pausing ads in the first 48 hours | Interrupts "Retrieval Test" phase—AI was still learning |
❌ Low creative volume (running only 1-3 ads) | Kills AI's ability to match creatives to different user types |
📊 Quick Reference: Budget & Timeline
Phase | Duration | Action |
|---|---|---|
Pre-Launch | 1-2 days | Complete this checklist |
Learning Phase | Days 1-7 | NO EDITS. Document metrics. |
Evaluation | Day 8 | Analyze spend distribution, CPA, ROAS |
Optimization | Day 8+ | Pause underperformers, add new creative concepts |
🎯 Performance Benchmarks
After completing this checklist, expect:
CPA improvement: 15-30% lower than non-optimized launches
Learning phase completion: 90%+ of campaigns exit learning within 7 days
Creative efficiency: Clear spend distribution showing 3-5 winning creatives
ROAS stability: Less variance after learning phase completes
