If you’ve been running Facebook or Instagram ads for your wine, beverage, lifestyle, or food business and have noticed some changes lately—perhaps your ads aren’t performing quite as they used to, or your costs have shifted—you’re not imagining things. Meta has implemented a significant update to how its advertising system works, and it’s worth understanding what this means for your marketing in 2026.  

The good news? This isn’t something to panic about. However, we want to be candid: we’re still in the process of testing and adapting to these changes ourselves, and the results have been mixed across different clients and industries. Some of the conventional wisdom about “trusting the algorithm” may not apply universally, especially for small businesses with unique offerings and niche audiences. 

What is the Andromeda Update? 

In December 2024, Meta introduced what they call “Andromeda”—a next-generation AI system that fundamentally changes how ads are selected and shown to people on Facebook and Instagram. According to Meta’s engineering team, this system has achieved a 6% improvement in recall and an 8% improvement in ads quality. 

Think of it this way: Previously, Meta’s advertising system was like a skilled matchmaker working from a limited pool of options. Now, with Andromeda, it’s more like having an incredibly intelligent assistant who can consider millions of possibilities in milliseconds and make highly personalized recommendations for each individual person. 

The system is powered by advanced artificial intelligence and specialized computer hardware (including NVIDIA’s Grace Hopper Superchip) that allows Meta to process vastly more information when deciding which ads to show to which people. 

The Key Shift: Creative is Now Your Targeting  

Here’s the most important thing to understand: Meta’s advertising has shifted toward creative-based matching rather than traditional audience-based targeting. 

In the past, you might have spent considerable time creating specific audience segments—targeting wine enthusiasts aged 35-55 in specific locations, for example. You’d build multiple ad sets for different demographics and interests.  

  • Interest targeting and custom/lookalike audiences still exist, but they don’t carry the same weight they used to. Narrow interest stacks can actually limit performance because they shrink the pool the AI can optimise against.  
  • In many cases, broad audiences (e.g., location + age + gender) give the system more room to learn and drive results — especially when paired with strong creative signals.

Under Andromeda, Meta now places greater weight on how visually unique each creative appears and what your ad content actually communicates. The AI analyzes your images, videos, copy, and even audio to understand what your ad is really about, then matches it with people whose behaviour and interests suggest they’d be interested—regardless of traditional targeting parameters. 

 However, here’s where we urge caution: While Meta promotes this as a more intelligent system, we’re finding that completely handing over control doesn’t always work for specialised products like craft wines, artisanal beverages, or premium food brands. The algorithm doesn’t always understand the nuance of your specific market position, price point, or customer sophistication level. 

What This Means for Your Business Running Meta Ads

1. Variety Matters More Than Perfection 

One beautifully polished ad won’t cut it anymore. Andromeda learns by testing different creative approaches to find what resonates with different people. 

For your wine or food brand, this means creating multiple variations: 

  • Different product presentations (bottle shots vs. lifestyle imagery vs. food pairings) 
  • Various messaging angles (heritage and tradition vs. innovation vs. sustainability) 
  • Diverse formats (static images, short videos, carousels, customer testimonials) 
  • Different storytelling approaches (behind-the-scenes, educational content, seasonal themes) 

The goal is to give Meta’s AI a diverse palette to work with. Think of it as having different conversations with different types of customers, rather than using one pitch for everyone. 

2. Simpler Campaign Structures Work Better (But Not Always) 

Meta’s guidance suggests that campaigns packed with multiple lookalikes, stacked interests, or heavy exclusions may no longer perform as efficiently. They recommend broader targeting with fewer restrictions. 

Campaign Structure has Shifted 

Pre-Andromeda:
Many segmented campaigns and ad sets with tight targeting. Often able to use short bursts of budget to drive results. 

Post-Andromeda: 

Fewer campaigns, fewer ad sets, lots of creatives inside broad ad sets. Need to allocate budget to allow for test, learn and optimising. 

This matters because: 

  • Meta’s Advantage+ frameworks perform better when each ad set has lots of data flowing into it — this helps with learning and optimisation.  
  • Tiny, fragmented audiences starve the learning process because the system has less data to evaluate per segment. 

Our experience has been more nuanced. While some clients have seen success with simplified structures, have found that completely removing audience controls leads to wasted spend on unqualified clicks. We monitor this daily so we’ve been trying to understand how accurate the system is and it’s not always right. 

We’re currently testing a hybrid approach: 

  • Using broader targeting than before, but not eliminating it entirely 
  • Maintaining some strategic exclusions (like excluding existing customers for acquisition campaigns) 
  • Keeping geographic targeting when it’s essential to your business model 
  • Building in more creative variations, but within thoughtfully structured campaigns 

The “one-size-fits-all” approach of letting the algorithm do everything doesn’t always work for specialty brands. Your $75 natural wine deserves different treatment than a $12 mass-market bottle, and the algorithm doesn’t always recognize that distinction without guidance. 

3. Creative Fatigue Happens Faster 

You may notice your ads burning out more quickly than before. Assets that used to last six to eight weeks may now fade in two to three with alerts that performance is impacted. We know that creative fatigue is different for other advertising mediums so are wary about these system warnings, however we’re battling a machine algorithm so rationale arguments don’t hold up here.   

This means you need a more consistent creative refresh cycle. Plan to update your ads every 2-4 weeks with fresh concepts. For seasonal businesses like wineries with vintage releases or beverage brands with summer promotions, this actually aligns well with natural marketing cycles. However, investing in creative and having people at the heart is important – those days of two photo shoots a year to feed the creative libraries just don’t cut it. 

4. Data Quality Matters More Than Ever 

High-quality conversion data and budget allocation directly impact performance. Make sure your Facebook Pixel is properly installed and that you’re tracking meaningful conversions—whether that’s purchases, reservations, newsletter signups, or tasting room visits. 

Clean, accurate tracking helps Andromeda learn what actually drives results for your specific business. 

 

Practical Steps for 2026 – Where to start with Meta ads 

We’ve already been working with Andromeda for the last year, trying to figure out how to best use it and we’ve seen success here for many clients. If you’re not aware of this as yet, here’s how to adapt your Meta advertising strategy: 

Invest in Creative Production You don’t need a Hollywood big brand budget, but you do need variety. Consider: 

  • People first – raw, authentic and human in nature
  • Behind-the-scenes footage from your vineyard, brewery, or kitchen 
  • Quick smartphone videos showing products in use 
  • Customer testimonials and reviews 
  • Educational content (wine pairing tips, brewing processes, recipe ideas) 

 

Embrace Automation—But Monitor Closely Meta’s Advantage+ campaigns are designed to work with Andromeda, and they can be effective. However, we recommend a cautious approach: 

  • Start with a small test budget before committing fully 
  • Run Advantage+ campaigns alongside some manual campaigns for comparison 
  • Check placements regularly—the algorithm may put your premium brand in contexts that don’t align with your positioning 
  • Monitor audience quality through post-click behaviour, not just click-through rates 
  • Be prepared to add manual guardrails if you’re seeing budget waste 

Full automation works best for brands with broad appeal and flexible positioning. For specialised, premium, or regionally-focused businesses, strategic human oversight remains important. 

 

Think Conceptually Different, Not Just Visually Different Don’t just change colours or swap a photo. Create ads with genuinely different: 

  • Value propositions (quality vs. convenience vs. experience vs. sustainability) 
  • Emotional tones (celebratory vs. educational vs. aspirational) 
  • Use cases (gifts vs. personal enjoyment vs. entertaining) 

Be Patient with Learning Give new campaigns at least 5-7 days before making judgments. The Andromeda system needs time to gather data and optimize. Patience is now a competitive advantage, as ads may need more time to build momentum. 

Monitor the Right Metrics Focus on campaign-level performance rather than obsessing over individual ad performance. Look at your overall return on ad spend and cost per acquisition across your creative portfolio. 

 

What We’re Still Learning 

We want to be transparent: as a marketing consultancy, we’re still actively testing and refining our approach to Andromeda because we’re not convinced that the system always knows best. We’ve been in digital marketing for over 2 decades and there is still something to input when we understand our audienes.  Here are some challenges we’re navigating: 

  • The Quality vs. Volume Trade-off Broad targeting often generates more impressions and clicks, but we’re finding that conversion quality can suffer. A click from someone who doesn’tunderstand your price point or brand positioning is wasted budget. We’re working to find the right balance between giving the algorithm data to learn from and maintaining lead quality. 
  • Creative Fatigue is Real, But Expensive . The need for constant creative refresh is legitimate, but producing 10-20 genuinely different concepts every few weeks is resource-intensive for small businesses. We’re testing whether strategic variations (same concept, different executions) can extend creative lifespan while still satisfying the algorithm’s need for diversity.
  • The Learning Phase Never Seems to End . Meta claims campaigns need 50 conversions per week to optimize effectively. For small businesses with modest budgets or longer sales cycles (like wine club memberships), reaching this threshold is challenging. We’re exploring whether different conversion events or campaign strategies can work around this limitation. 
  • Algorithm Transparency Issues : Unlike previous systems where you could see exactly who you were targeting and why, Andromeda operates more as a black box. When performance drops, it’s harder to diagnose whether it’s creative fatigue, audience saturation, seasonal factors, or something else entirely. 

 

We’re sharing this not to discourage you, but to set realistic expectations. The advertising landscape is evolving and finding what works requires ongoing testing and adjustment—not just implementing a new playbook and moving on. 

What About Small Meta Advertising Budgets? 

If you’re working with a modest advertising budget (which many small wine, beverage, and food businesses are), Andromeda can actually work in your favour. The system is designed to be more efficient, potentially delivering better results with the same spend—as long as you provide it with quality creative inputs and long enough timeframe to test and learn. 

However, according to industry research, the algorithm works best with at least 50 conversions per week to have enough data to learn effectively. If you’re spending $500-1000 per month, be realistic about timelines and focus on one clear objective rather than trying to do everything at once. 

Meta’s Andromeda update represents a significant shift toward AI-driven, creative-focused advertising. The promise is more intelligent ad delivery and better results with less manual targeting work.  

The reality is more complicated. While some businesses are seeing improvements, others—particularly those in specialised niches are finding that complete automation leads to wasted spend and lower-quality leads. There is often suggestions for targeting, copy and creative enhancements (AI driven) that are not aligned with brand look and feel, and often enter the irresponsible marketing territory for wine. Yes, it may be more fun and playful, but there’s a brand risk element to also consider in the way that language is used and the AI models are not across this. 

For small businesses in the wine, beverage, lifestyle, and food sectors, our current recommendation is: 

  • Don’t abandon targeting entirely - Use broader parameters than before, but maintain strategic controls that protect your brand positioning and budget. Uploading custom audiences and lookalike audiences can help the AI to learn about your ideal customer 
  • Invest in creative diversity - This part of Meta’s guidance seems sound; varied, authentic content does perform better. Make sure that you have some budget allocated to creative so that the algorithms can test with a few options 
  • Test cautiously - Run controlled experiments rather than overhauling everything at once. We’re still monitoring results carefully , not every business or industry is the same
  • Monitor quality, not just volume - More clicks mean nothing if they’re not from qualified prospects 
  • Maintain manual oversight - The algorithm is a tool, not a replacement for strategic thinking about your unique market position 

The brands that will thrive aren’t necessarily those that trust the automation most completely, but those that find the right balance between leveraging AI capabilities and maintaining human strategic oversight. 

This is particularly true for businesses that offer craft products, premium experiences, or specialised services. Your competitive advantage often lies in understanding nuances that algorithms, no matter how sophisticated, may not fully grasp. 

The advertising landscape is always evolving, and Meta’s Andromeda update is just the latest chapter. We’re committed to testing, learning, and adapting alongside our clients to find what actually works—not just what Meta’s marketing materials claim should work. 

If you’re working with us or another marketing agency, make sure they’re approaching Andromeda with the same healthy scepticism and commitment to testing that you’d expect for any significant strategic change. 

Sources: 

  • Meta Engineering Blog: “Meta Andromeda: Supercharging Advantage+ automation with the next-gen personalized ads retrieval engine” (December 2024) 
  • Clutch.co: “What Meta’s Andromeda Update Means for Businesses in 2026” 
  • AdExchanger: “What Meta’s Andromeda Update Actually Changes – And What It Doesn’t” (January 2026) 
  • Disruptive Digital: “How the Meta Andromeda Update Changed Everything and What We Did About It” 
  • Multiple industry analysis from digital marketing agencies and Meta advertising specialists 

Note: Meta’s algorithms and best practices continue to evolve. This guide reflects current understanding as of February 2026. We recommend staying informed through Meta’s official business resources and working with experienced marketing professionals to optimise your specific campaigns.