If your Facebook and Instagram ads suddenly stopped performing over the past year, you’re not imagining it.
Meta rolled out a major update called Andromeda that fundamentally changed how ads are delivered, who sees them, and how the algorithm prioritizes creative. Businesses across the board saw performance shifts — campaigns that were working suddenly needed different approaches.
Here’s what Meta Andromeda actually is, how it’s affecting your ads, and what you need to do to adapt.
What Is Meta Andromeda?
Meta Andromeda is an AI-driven ads retrieval system that decides which ads are eligible to be shown to a user. It started rolling out in late 2024 and became a core component of Meta’s platform throughout 2025.
Here’s what changed: Instead of starting with advertiser-defined audiences — the interests, lookalikes, and demographics you used to set — Andromeda works in reverse. It evaluates your ad creative, copy, format, and historical engagement data to predict which users are most likely to engage and convert.\
The system shifted Meta from audience-first advertising to creative-first matching.
Your targeting settings still matter, but the creative is now the primary signal Meta uses to find your audience. The algorithm looks at visuals, themes, hooks, and language to determine relevance — then finds users who are likely to respond.
What Andromeda Changed for Advertisers
Andromeda meant advertisers saw four major shifts, whether they realized what was going on or not. Step one of fixing a problem is knowing what the issue might be to begin with, so you know where to start troubleshooting.
1. Broad targeting started outperforming narrow targeting
Detailed targeting options that used to drive performance stopped working, and broad targeting started winning instead. This is because AI performs better when it has a larger pool to work with, so broader campaigns give AI more options to match you with users who will convert.
2. Simplified account structures started winning
Hyper-segmented campaigns with dozens of ad sets stopped performing well too. Advertisers who consolidated campaigns into fewer, broader ad sets with larger budgets saw better results. Fewer campaigns mean more data per campaign, which helps Andromeda learn faster and identify patterns.
3. Creative fatigue accelerated
Ads that used to perform well for four or even six weeks started declining in just two to three. Meta’s AI began rotating users away from ads they’d seen multiple times more aggressively, which means you need fresh creative more often to stay in rotation.
4. Tracking quality became non-negotiable
This one catches a lot of advertisers off guard. Andromeda relies on your conversion data to learn which creatives are working and why. If your Meta pixel is firing inconsistently, your Conversions API isn’t set up, or you have deduplication issues in Events Manager, the algorithm is essentially flying blind — and it will show in your CPMs. Clean, accurate data signals are now as important as the ads themselves.
Introducing GEM: Meta’s Central AI Brain
In addition to these shifts in the way ads perform, in mid-2025, Meta rolled out GEM (Generative Ads Recommendation Model) to work alongside Andromeda.
Here’s the difference between them:
- Andromeda decides which ads can be shown in a user’s feed and creates a list of ones that might be relevant to a user.
- GEM determines which ads should be shown based on patterns across user behavior, engagement data, and conversion signals.
Think of it like this: Andromeda decides what ads make it onto the shelf. GEM learns what people buy and decides which ads to take off the shelf and put into people’s feeds.
According to Meta, GEM is 4x more efficient at driving ad performance gains compared to original recommendation models.
For advertisers, this means ads are now evaluated within broader user journeys. Long-term patterns matter more than short-term performance fluctuations, and fast testing cycles or frequent edits can hurt performance by interrupting the system’s learning.
Why Your Meta Ads Aren’t Performing Right Now
If your ad performance declined after Andromeda rolled out, it’s likely one or more of these issues.
You’re running the same creative too long. You used to be able to run the same winning ad for months and still see results. Now, you need to rotate fresh creative every two to three weeks to stay on top.
Your creative is too generic. Stock photos, text-heavy static images, and templated designs are getting deprioritized. Meta’s AI favors creative that feels authentic, visually engaging, and tailored to the target audience.
You’re not testing enough variations. Andromeda and GEM reward advertisers who test multiple creative angles. If you’re running one ad per campaign, you’re losing out to competitors testing several variations. More creative inputs means AI has more options to find what resonates.
Your account structure is too complex. Dozens of hyper-segmented ad sets split your budget and limit learning. Meta’s AI needs volume to identify patterns, so consolidating campaigns into fewer, broader ad sets helps the system learn faster.
You’re editing campaigns too often. Frequent edits reset the learning phase and interrupt pattern recognition. What looks like early underperformance might just be the system learning. Patience is now a competitive advantage.
Your budget is too low. Low spend limits the AI’s ability to learn. Meta performs best when budgets allow campaigns to generate consistent conversion data. If your daily budget is too small relative to your conversion events, the system can’t detect trends.
Your tracking is off. If your pixel and Conversions API aren’t sending clean, deduplicated data, Andromeda can’t properly evaluate your creative performance — and you’ll see CPMs rise as a result. Check Events Manager and audit your event match quality score; anything below a 7 means Meta is struggling to match your conversion events to real users.
How to Adapt Your Meta Ads Strategy After Andromeda
That’s a lot to take in, and it might feel like we’re telling you to reinvent the wheel a bit. We get it, it’s been an adjustment for us too.
Here are steps to analyze your ads and update them to work well with Andromeda.
1. Make creative strategy your core lever
Test creative angles tailored to different customer personas — not just micro-variations of the same ad. Use diverse formats, such as images, videos, carousels, user-generated content, and testimonials, and try different angles like emotional appeals, social proof, or logical arguments. The more creative variations you give Meta’s AI, the better it can match your ads to the right users.
2. Refresh creative at least every two weeks
Even if an ad is still performing, plan to rotate it out within two weeks. Create multiple variations of the same concept — different images, different hooks, different CTAs — and test them simultaneously.
3. Simplify your account structure
Consolidate campaigns and ad sets. Fewer campaigns with broader targeting and consolidated budgets allow Andromeda and GEM to learn faster. It’s common now to see accounts with one or two campaigns outperforming accounts with dozens.
4. Let campaigns learn
Before launching new campaigns or assets, set a minimum no-touch window — at least a week or 50 to 75 conversions, whichever comes first. Look at three- to seven-day performance trends instead of daily spikes before making any changes.
5. Prioritize native, authentic creative
Ditch stock photos and overly polished ads. User-generated content, behind-the-scenes footage, and authentic testimonials perform better under Andromeda. Video outperforms static images, and carousel ads give the algorithm more signals to work with.
6. Audit your tracking before anything else
Run both pixel and Conversions API with deduplication enabled. Check your event match quality score in Events Manager monthly. Pick one primary conversion event and optimize for it consistently — switching between events confuses the algorithm.
7. Use budget as a signal
Make sure your daily budget is realistic and aligned with your conversion event. High-intent events like purchases require more ad spend per learning cycle than upper-funnel actions like clicks or engagements.
How HustleFish Handles Meta Ads After Andromeda
We’ve been managing Meta ads through algorithm updates for years, so we’ve gotten pretty good at adapting when things change. Here’s what we’ve started going for clients since Andromeda rolled out.
- We’re producing more creative variations. Instead of two to three ads per campaign, we’re testing five to 10 to give the algorithm options and prevent fatigue.
- We’re shortening creative refresh cycles. What used to be monthly is now every two weeks.
- We’re simplifying account structures. Fewer campaigns, broader targeting, consolidated budgets.
- We’re auditing tracking setups before touching creative. Sometimes fixing pixel and CAPI configurations made more of a difference than any creative change.
- We’re letting campaigns learn. We set minimum no-touch windows and avoid constant tweaking that resets the learning phase.
- We’re leaning into video and carousel formats. These perform better under Andromeda’s creative-first system.
If your Meta ads aren’t performing like they used to, we can help. We’ll audit your campaigns, identify what’s not working, and build a strategy that works with Andromeda — not against it.
Not sure how to adapt your Meta ads strategy? Let’s audit your campaigns and show you what’s not working.
Frequently Asked Questions
What is Meta Andromeda?
Meta Andromeda is an AI-driven ads retrieval system that decides which ads are eligible to be shown to users. It prioritizes creative quality and relevance over advertiser-defined targeting. It rolled out in late 2024 and became a core component of Meta’s ad platform in 2025.
When did Meta Andromeda roll out?
Andromeda began rolling out in late 2024 and became a core part of Meta’s advertising infrastructure throughout 2025. GEM (Generative Ads Recommendation Model) rolled out in mid-2025 with broad impact by Q4 2025.
Why are my Facebook ads not working after the Andromeda update?
A few things could be happening. Creative fatigue happens faster now, so if you’ve been running the same ads for weeks, they’re probably worn out.
Your account structure might be the culprit too; if you have hyper-segmented ad sets or constantly tweak your campaigns, Meta’s AI doesn’t have enough data to learn from.
How often should I refresh my ad creative after Meta’s Andromeda?
Every two to three weeks to start, and adjust from there if needed. Even if performance is doing well, plan to rotate your creative proactively to prevent fatigue.
Do video ads perform better after the Andromeda update?
Yes. Video ads, carousel ads, and interactive formats perform better under Andromeda because they’re more engaging and give the AI more signals to work with.
What is GEM and how does it work with Andromeda?
Andromeda decides what can be shown; GEM decides what should be shown. Both rely on user behavior to narrow down ad options, and together, they make Meta’s ad delivery more efficient.
Should I use broad targeting or detailed targeting after Andromeda?
Broad targeting performs better under Andromeda. Meta’s AI now needs a larger pool of users to find the best matches based on your creative. Detailed targeting that used to work well doesn’t perform as strongly now.
How long should I let campaigns learn before making changes?
At least a week or around 50 to 75 conversions, whichever comes first. Early volatility is normal, so look at three- to seven-day performance trends instead of daily fluctuations before making adjustments.
Does my tracking setup affect Andromeda performance?
Yes, significantly. Andromeda relies on conversion data to learn which creatives are working.
If your pixel is firing inconsistently or your Conversions API isn’t properly configured, performance will suffer regardless of how good your creative is. Audit your event match quality score in Events Manager — anything below a 7 means Meta is struggling to match your conversions to real users.