Meta's Andromeda Update: Why Your E-commerce Facebook Ads Stopped Working

Updated January 2026: Six months after Meta's Andromeda algorithm update devastated e-commerce Facebook ads, the dust has settled. Some NZ online stores are thriving. Others are still bleeding money. The difference comes down to whether they adapted their creative strategy.

If your e-commerce store's Facebook ads are still underperforming - or if you adapted but want to optimise further - this updated guide covers what's actually working for NZ online retailers in 2026.

The Story So Far

Meta's Andromeda update rolled out in July 2025 and hit e-commerce advertisers hard. The algorithm was designed to handle the flood of AI-generated content by rewarding genuine creative diversity over volume.

E-commerce stores got hit hardest because most relied on the same approach: product photos, feature-focused copy, let Meta optimise. That stopped working overnight.

Six Months Later: E-commerce Recovery Data

We've tracked performance across our e-commerce clients since July 2025. Here's what the data shows:

NZ E-commerce Benchmarks (January 2026)

| Metric | July 2025 (Crisis) | January 2026 (Adapted) | | ----------------- | ------------------ | ---------------------- | | Average ROAS | 1.8x | 3.2x | | Average CPM | $30-40 | $18-25 | | Cost per purchase | $35-50 | $18-28 | | Average CPC | $1.50-2.00 | $0.80-1.20 |

The businesses hitting these improved benchmarks share common characteristics. The ones still struggling are making the same mistakes.

What's Working for NZ E-commerce in 2026

1. Advantage+ Shopping Campaigns

Meta's Advantage+ Shopping campaigns have improved significantly since July 2025. They now work _with_ Andromeda rather than against it.

Setup that works:

• Single campaign with $75-150/day minimum budget
• Complete product catalog with high-quality images
• Conversions API properly implemented
• 14-21 day learning phase (don't touch it)

Results we're seeing:

• 20-35% lower cost per purchase than manual campaigns
• Better new customer acquisition
• Improved catalog-wide performance (not just bestsellers)

2. Customer Motivation-Based Creatives

The winning e-commerce accounts create ads around _why_ people buy, not _what_ they're buying.

Example: NZ Skincare Brand

Instead of "Our moisturiser has hyaluronic acid," they created separate ads for:

• Problem-focused: "Tired of dry, flaky skin this winter?"
• Result-focused: "Hydrated, glowing skin in 7 days"
• Social proof: "Why 2,000+ Kiwi women switched to us"
• Value-focused: "Premium skincare without the premium price"
• Convenience: "Delivered to your door, subscription optional"
• Trust: "Made in NZ, dermatologist tested"

Each ad has completely different copy, different imagery, different call-to-action. This is what Andromeda rewards.

3. Video Content (But Smarter)

Short-form video is outperforming static images under Andromeda, but only when done right:

What works:

• Under 15 seconds
• Hook in first 2 seconds
• Product in use (not just product shots)
• Sound-off optimised (captions/text overlay)
• UGC-style production (polished but not corporate)

What doesn't work:

• Long-form brand videos
• Overly produced content
• Videos without clear product focus
• Slow builds without early hook

4. Catalog Creative Diversity

For stores with large product ranges, catalog ads have become more effective - but need creative diversity applied:

Approach:

• Multiple catalog ad sets, each with different creative overlays
• Segment by product category with category-specific messaging
• Use lifestyle images alongside product shots
• Dynamic creative optimisation with diverse templates

5. First-Party Data Integration

E-commerce stores with strong first-party data are crushing it:

• Email list custom audiences for retargeting
• Purchase history for lookalike audiences
• Conversions API sending complete customer journey data
• Customer segmentation by purchase behaviour

What's Still Not Working

AI-Generated Creative Spam

Ironically, Andromeda was built to penalise exactly this. Stores using AI tools to generate dozens of similar product images are seeing the worst performance.

The problem: AI variations look different to humans but semantically similar to Andromeda. The algorithm sees them as duplicates.

The fix: Use AI for ideation and production efficiency, but ensure genuine strategic diversity in messaging.

Feature-Only Copy

"Our product has X feature" isn't enough. You need to connect features to customer outcomes.

Instead of: "100% merino wool"

Try: "Stay warm on the coldest Auckland commute - without the bulk"

Ignoring the Learning Phase

E-commerce advertisers are notorious for making changes too quickly. Under Andromeda, this kills performance.

Rule: No significant changes for 14-21 days after launch. Let the algorithm learn.

Broad Discounting

"20% off everything" ads don't provide the messaging diversity Andromeda needs. Discount ads can work, but need to be part of a diverse creative mix, not the entire strategy.

E-commerce Creative Framework for 2026

Minimum Viable Creative Set (15 ads)

Problem-aware (3 ads):

• Different problems your products solve
• Different customer pain points
• Different frustrations with alternatives

Solution-aware (3 ads):

• How your product solves each problem
• Key differentiators from competitors
• Unique value proposition angles

Social proof (3 ads):

• Customer testimonials
• Review highlights
• User-generated content

Objection handling (3 ads):

• Price concerns
• Quality doubts
• Shipping/returns worries

Urgency/offer (3 ads):

• Limited stock
• Seasonal relevance
• Special offers

Format Mix

• 5 static images (different styles)
• 5 short videos (under 15 seconds)
• 3 carousels (storytelling format)
• 2 UGC-style content

Setting Up for Success

Technical Requirements

Conversions API: Essential, not optional. Server-side tracking captures data that browser tracking misses.

Product catalog: Complete, accurate data. Missing fields hurt performance.

Website speed: Slow sites kill conversion rates. Andromeda can't fix a bad landing experience.

Budget Minimums

Post-Andromeda, we've found these minimums for effective e-commerce campaigns:

• Testing new creative: $50-75/day minimum
• Scaling winners: $100-200/day sweet spot for NZ market
• Advantage+ Shopping: $75/day minimum to exit learning phase effectively

The Opportunity

Here's the good news: most NZ e-commerce businesses still haven't adapted properly. The ones that have are seeing better results than pre-Andromeda.

This is a competitive advantage if you move now. The gap between adapted and non-adapted e-commerce advertisers will only widen throughout 2026.

Next Steps

• Audit your current creative diversity (do you have 15+ genuinely different messages?)
• Review your campaign structure (are you using Advantage+ Shopping?)
• Check your technical setup (Conversions API, complete catalog)
• Plan your creative production for Q1 2026

Need help getting your e-commerce Facebook ads profitable again? Get in touch for a campaign audit. We've helped dozens of NZ online stores recover and exceed their pre-Andromeda performance.