How Andromeda Has Changed Meta Advertising: A Practical Guide

In late 2024, Meta began rolling out Andromeda, a new AI-powered infrastructure that fundamentally changes how ads are delivered on Facebook and Instagram. If you've noticed your campaigns performing differently over the past year—whether better or worse—Andromeda is likely the reason why.

Andromeda isn't just another algorithm update. It's a complete overhaul of Meta's ad delivery system, replacing the previous rule-based approach with deep learning models that can process billions of ad-audience combinations in real time. The system uses neural networks that are reportedly 10,000 times more complex than what Meta was using before, powered by custom chips designed specifically for this purpose.

For advertisers, this means the strategies that used to work well may no longer be optimal. In fact, many of the best practices we've relied on for years—detailed audience targeting, multiple small ad sets, constant manual optimization—can now actually hinder performance rather than help it.  As advertisers we’ve had to  fundamentally rethink our approach to campaign structure, creative strategy, and performance measurement.  Some of these changes can feel counterintuitive.  Here are some key changes that Andromeda has brought on

Campaign structure has simplified dramatically.

We've moved from running multiple campaigns with numerous ad sets to consolidating into fewer campaigns with fewer - oftentimes one - broad ad sets. Where you once might have had 5 ad sets targeting different interests with 3-5 ads each, now you might run one ad set with up to 20-30 creatives, or more. This consolidation gives Andromeda more data to work with and more flexibility to optimize, leading to better results with less management overhead.

Creative strategy has become a primary focus.

With Andromeda handling a lot of the targeting automatically, creative differentiation has become a huge priority.  You probably have heard the phrase “creative is the targeting” - what that means is, instead of defining audiences manually we create ads for different customer personas and let the AI figure out who responds to which message. Each creative essentially acts as its own targeting mechanism—the algorithm reads the creative signals and finds the right audience accordingly.

There are different metrics of success.

Individual ad performance matters less now than campaign and ad set level results. Spend allocation can also function as a performance indicator. When Andromeda consistently allocates budget to certain ads, it's often because those ads are contributing to overall campaign success in ways that aren't captured by last-click attribution. Some of your highest-spending ads could show mediocre direct ROAS but dramatically improve overall campaign performance when active.  More focus should be placed on campaign and account level ROAS, vs individual ad level ROAS.

New ad formats like Flex ads have become valuable despite their limitations.

Flex ads leverage Andromeda's capabilities effectively but provide limited visibility into which specific creative elements drive performance. While the lack of detailed reporting can be frustrating, the performance gains can justify their use, especially when testing broad creative themes.

This guide will walk through each of these changes in detail, explaining not just what's different but why it matters and how to adapt your approach accordingly. The goal isn't to completely abandon everything you know about Facebook advertising. Many fundamental principles - the importance of compelling creative, understanding your customer, testing and iteration - remain as relevant as ever. What's changed is how to apply these principles within Andromeda's AI-driven framework.

For some advertisers, these changes have been challenging. There's less direct control, less visibility into exactly why certain decisions are made, and a need to trust an algorithm whose workings aren't fully transparent and, to be frank, has been beset by bugs and outages.  Not to mention that  Andromeda has not always worked as expected.  The growing pains have been significant for many advertisers.  However, as it continues to improve, and advertisers continue to evolve with the system, Andromeda offers the opportunity for better performance with less manual work.

Let's start with one of the most visible changes: how campaign structure has evolved from complex hierarchies to streamlined, consolidated setups that give the AI room to work.

The Death of Micro-Targeting: From Many Small Ad Sets To Consolidated Power

If you opened a typical Facebook Ads account a couple of years ago, you'd likely see something resembling a complex organizational chart. Campaigns for different objectives, multiple ad sets within each campaign targeting specific interests or demographics, and a handful of ads in each ad set. It wasn't uncommon to see structures like this:

• Prospecting Campaign 1: Women 25-44
  
  • Ad Set 1: Interest in Yoga
  • Ad Set 2: Interest in Pilates
  • Ad Set 3: Interest in Meditation
• Prospecting Campaign 2: Women 25-44
  
  • Ad Set 1: Lookalike of Purchasers
  • Ad Set 2: Lookalike of Engagers
• Prospecting Campaign 3: Broad
  
  • Ad set 1: Open Targeting
• Retargeting Campaign
  
  • Ad Set 1: Cart Abandoners
  • Ad Set 2: Product Viewers
  • Ad Set 3: Past Purchasers

This structure made intuitive sense. We could control budgets for specific audiences, see which interests performed best, and feel confident we were reaching the right people with the right message. Each ad set typically contained 3-6 ads.

Today, that same account might look like this:

• Prospecting Campaign
  
  • Ad Set 1: Broad Targeting, no limit on number of ads other than Meta’s, which recently changed from 50 to 150
  • Ad Set 2: Broad Targeting, creative testing sandbox

The transformation is striking, and at first glance, it might seem like we're doing less work or being less sophisticated. In reality, we're working with Andromeda's capabilities rather than against them.

The logic behind consolidation becomes clear when you understand how Andromeda operates. The AI needs data to learn effectively, and splitting your budget across multiple ad sets dilutes the signal it receives.

"Data density" has become a critical concept in the Andromeda era. It refers to the concentration of conversion events flowing through a particular campaign or ad set. Higher density means faster learning and better optimization.  One campaign getting 20 purchases per day gives Meta much clearer signal vs 3 campaigns giving it 6-7 per day.

While we're removing most targeting restrictions, strategic exclusions remain important.  Andromeda friendly campaign structures are intended as “full funnel” - you no longer have to break out different pieces of your funnel in different campaigns.  That said, we still strongly recommend excluding all current customers from your campaigns so Meta uses your budget to find new customers.  Let email and SMS handle repeat purchasers.  

Creative as the New Targeting: Personas Over Parameters

With Andromeda handling audience selection automatically, creative has become the primary mechanism for reaching different customer segments. This represents an inversion of older Facebook advertising strategy. Where we once spent cycles crafting detailed audience parameters and then created a few ads to test, we now use broad targeting and let our creative variations do the work of finding the right people.

The concept is deceptively simple: Andromeda reads every element of your creative - the visuals, copy, tone, format - as signals about who might respond to that ad. A technical tutorial video naturally finds people seeking education. An emotional brand story reaches those who connect with narratives. A problem-agitation-solution ad attracts people actively looking for solutions.  Your creative is no longer just your message; it's your targeting mechanism.

In place of Meta’s old targeting options, successful advertisers are now using creative differentiated by persona to find different audience segments.  A persona represents a specific customer mindset, need state, or position in the buying journey. Each creative is designed to resonate with a particular persona, and Andromeda handles matching that creative to the right people.

Common persona frameworks that work well with Andromeda:

Awareness Level Framework:

• Problem Unaware: Educational content that introduces the problem
• Problem Aware: Content that agitates the problem and introduces the concept of solutions
• Solution Aware: Demonstrations and comparisons of your specific solution
• Product Aware: Social proof, testimonials, and trust-building content
• Most Aware: Offers, urgency, and direct response messaging

Emotional State Framework:

• Frustrated: Focus on pain points and immediate relief
• Curious: Educational and how-to content
• Skeptical: Data, proof, and third-party validation
• Excited: Aspirational content and transformation stories
• Analytical: Features, specifications, and detailed comparisons

Use Case Framework:

• Professional Use: ROI-focused, efficiency gains, business benefits
• Personal Use: Lifestyle benefits, convenience, personal enjoyment
• Gift Giving: Packaging, presentation, recipient happiness
• Problem Solving: Specific solutions to specific problems

The key is that each creative speaks distinctly to a different persona.  You may have heard that creative volume is the name of the game with Andromeda.  And while there is some truth to that, even more than volume is diversity and quality.  Meta recently raised the amount of ads you can have in an ad set to 150 - but this isn’t about creating 150 random variations—it's about systematically covering different angles, personas, and messages to give Andromeda maximum flexibility in matching content to users.  This includes both format diversity (static images, short form video, long form video, carousels, etc) as well as persona/angle diversity.  

But diversity is the name of the game, more than volume.  Andromeda appears to group similar creatives together, treating minor variations as essentially the same ad. This means testing blue versus red backgrounds or slightly different headlines is largely pointless. The algorithm needs conceptually different creatives to properly segment audiences.

Instead of testing "Save 20% Today" versus "Get 20% Off Now", test fundamentally different approaches:

• A founder story video versus a customer testimonial
• A problem-focused pain point ad versus an aspirational lifestyle ad
• A technical demonstration versus an emotional transformation story
• Professional studio content versus authentic UGC

Each of these gives Andromeda a distinctly different signal to work with, allowing it to find different pockets of your potential audience.

Measuring Creative Performance in the Andromeda Era

While ROAS will always be what drives direct response Meta advertisers, the introduction of Andromeda calls for some additional interpretations of success.

• Spend Distribution: The percentage of budget Andromeda allocates to each creative can be more telling than conversion metrics. High spend generally indicates the algorithm has found a strong audience match.
• Stability Over Time: Creatives that maintain consistent spend and performance over weeks are particularly valuable. They've found a sustainable audience pocket.  It can also take Meta some time in the Andromeda era to properly match creatives to the right pockets of users, so having patience with ads - we recommend waiting at least 5 -7 days before making any decisions on ads - will often be rewarded.  
• Assist Impact: Some creatives might not drive direct conversions but improve overall campaign performance when active. Test pausing high-spend, low-ROAS creatives to see if overall performance drops.  Make sure you are assessing ROAS on a campaign level - high spend/low ROAS creatives can be driving more upper funnel behavior while lower spending ads in the same campaign hit a lower funnel audience and therefore get more attributed conversions.  But without those high spending top of funnel creatives the lower funnel ads can dry up.  

The shift to creative-as-targeting requires rethinking your entire content strategy. It's no longer about finding the one perfect ad but building a comprehensive library that covers every angle of your customer's journey and mindset. This approach gives Andromeda the tools it needs to find and convert your ideal customers at scale.

In the next section, we'll explore how new ad formats like Flex ads fit into this strategy, offering powerful automation at the cost of transparency.

Flex Ads: Perfect for Andromeda, Frustrating For Transparency

Flex ads represent Meta's most aggressive push into automated creative optimization. Unlike traditional ads where you control exactly what users see, Flex ads allow Meta to dynamically combine different creative elements such as headlines, primary text, images, video to create personalized combinations for each user. It's Andromeda's capabilities pushed to their logical extreme.

When you create a Flex ad, you're essentially giving Meta a toolkit of creative components and letting the algorithm assemble them in real-time based on what it predicts will resonate with each specific user. The combinations are nearly endless, and that's both the power and the problem.

The performance gains from Flex ads can be substantial. By testing countless combinations automatically, they often find winning formulas that manual testing would never discover. The algorithm can identify non-obvious patterns.  Perhaps your technical headline performs best with your lifestyle imagery, or your emotional copy resonates when paired with product close-ups.  If you haven’t tried them yet they are well worth testing, despite their transparency issues.  

And that is where frustration sets in: Meta provides almost no visibility into which specific combinations drive performance. You can see overall Flex ad performance, but not whether your emotional headlines outperform technical ones, or whether video consistently beats static images.

This opacity makes it nearly impossible to extract learnings that inform broader creative strategy. You know the Flex ad works, but not why it works or how to replicate that success in other formats.  And yet, they can provide a performance boost that can make them worth using.

Additionally, Flex ads likely represent the future direction of Meta advertising - increased automation, improved performance, reduced transparency. As Andromeda evolves, we can expect:

• More automated formats that assemble creative dynamically
• Further reduction in manual creative control options
• Improved performance from algorithmic optimization
• Continued challenges in understanding the "why" behind results

It’s been a bumpy and frustrating ride up to now, with many bugs, outages and some truly questionable algorithmic decisions from Meta’s AI…but, for better or worse, this is where things are headed.  Advertisers who embrace this trade-off early, developing strategies that balance performance with learning, will be best positioned for continued success.

Conclusion: Navigating Your Own Path in the Andromeda Era

The changes Andromeda has brought to Meta advertising are profound and, for many advertisers, challenging to navigate. We've moved from a world of granular control and transparent reporting to one of AI-driven optimization and algorithmic decision-making. Campaign structures have consolidated, creative has become the primary targeting mechanism, and success metrics have evolved beyond simple attribution models.

But here's what's crucial to remember: while these strategies represent current best practices in the Andromeda era, there's still no one-size-fits-all approach for every account.

Your Data Tells Your Story

The frameworks and strategies outlined in this guide provide a strong starting point, but your account data should always be your North Star. Every business has unique characteristics that affect how Andromeda performs:

• Your customer journey complexity - High AOV/long consideration products with long sales cycles may see different patterns than impulse-purchase e-commerce brands
• Your creative production capabilities - A brand with unlimited UGC might leverage different strategies than one relying on quarterly photo shoots
• Your margin structure - High-margin businesses can afford to test more aggressively than those operating on thin margins
• Your competitive landscape - Saturated markets may require different approaches than emerging categories

Pay attention to what your specific data tells you. If maintaining some audience segmentation consistently outperforms full consolidation in your account, don't abandon it just because consolidation is the trend. If certain "rules" of Andromeda don't seem to apply to your campaigns, trust your results over general wisdom.

Testing Remains Essential

The Andromeda era hasn't eliminated the need for testing—it's changed what and how we test. Instead of micro-testing audiences and headlines, we're testing fundamental strategic approaches:

• Consolidation levels - Maybe your sweet spot is 3-5 campaigns, not 1 or 2
• Creative volume thresholds - Perhaps 15 or even 5 ads per ad set performs better than 50 for your brand
• Exclusion strategies - Excluding existing customers may not be right for your business.  Or, perhaps tighter exclusions including site visitors and engagers are the right direction for you  

Document your tests, measure over sufficient time periods (remember that 5-7 day period of patience), and let data guide your evolution. What works in Q1 might not work in Q4. What works for your hero product might not work for new launches.

Patience and Persistence in Adaptation

The transition to Andromeda-optimized advertising isn't a switch you flip.  Some accounts see immediate improvements with consolidated structures and broad targeting. Others need several rounds of testing to find their optimal setup. This is normal.

Try to be patient with the system.  This can be really hard - trust me, I know!  Andromeda's bugs and outages have been frustrating for everyone. Algorithm changes that worked yesterday might break today. That high-performing campaign might suddenly tank for no visible reason. But this is the life of a Meta advertiser.

What matters is maintaining a learning mindset. Each test teaches you something about how Andromeda interprets your specific business. Each creative iteration helps you understand what resonates with your audience. Each measurement evolution reveals new insights about your true performance drivers.

The Human Element Still Matters

While we've handed over much control to AI, human judgment remains irreplaceable. Andromeda can't understand:

• Your brand values and what creative directions are off-limits
• Your business goals beyond what conversion data tells it
• Your customer feedback from channels outside Meta
• Your strategic priorities for market positioning

You still need to make strategic decisions about creative direction, offer development, and brand messaging. You still need to interpret performance in the context of your broader business. You still need to know when to override the algorithm's preferences because you understand something it doesn't.

Looking Forward

Andromeda isn't the end state of Meta advertising, but rather another step in ongoing evolution. As AI capabilities expand, we can expect even more automation, less manual control, and new formats we haven't imagined yet. The advertisers who thrive will not only best adapt to today’s best practices, but retain a flexible mindset and willingness to change when the landscape changes.

The fundamentals of good marketing haven't changed. Understanding your customer, crafting compelling messages, and delivering value still matter. What's changed is how we execute these fundamentals within an AI-mediated system.

If you're feeling overwhelmed by these changes, you're not alone. Every advertiser is figuring this out in real-time, dealing with the same frustrations and uncertainties. Remember that Andromeda, for all its sophistication, is just a tool. A powerful tool that's reshaping our industry, but still just a tool. Your strategic thinking, creative insight, and understanding of your customer remain the true drivers of advertising success. Andromeda simply helps you execute these more efficiently…when you learn to speak its language.

Adapt what works, test what's uncertain, and always let your account's unique data guide your decisions. The Andromeda era is here whether we're ready or not. The question isn't whether to adapt, but how quickly you can evolve while staying true to what makes your advertising effective for your specific business.

Welcome to the new reality of Meta advertising. It's frustrating and it’s powerful.  Time to make it work for you.