Meta's Andromeda Update: What's Actually Changing

If you run Meta ads, you've heard the noise: Andromeda is coming. Meta's been unusually public about this one, and that's a signal something big is shifting. Most of the chatter online is speculation. Here's what's actually changing.

What Andromeda Actually Is

Andromeda isn't a single feature update. It's a redesign of Meta's core ranking algorithm—the system that decides which ads get shown, to whom, and how often. The algorithm is Meta's most jealously guarded asset. The fact that they're being transparent about this change suggests they believe it will benefit advertisers. Or they're hedging against the inevitable outcry when performance dips during the transition.

Meta's had the same core ranking system for nearly a decade. Andromeda modernises that system using more recent machine learning techniques and a rethink of how ads compete for inventory.

The Three Things Actually Changing

1. Algorithm Retraining (The Speed Play)

The old algorithm was trained on patterns from 2015-2020. That's not ancient, but in machine learning terms it's stale. Andromeda retrains on much fresher data—behaviour patterns from the last 2–3 years. This means the algorithm is faster at recognising what works right now, not what worked five years ago.

What this means: creative trends evolve faster than the algorithm used to adapt. Andromeda closes that gap. Seasonal content, trend-jacking, and moment-relevant creative will be rewarded faster. Evergreen creative that worked for years might plateau sooner.

2. Broader Audience Matching (The Precision Paradox)

The old system was built around niche targeting. If you picked a tight audience—interests, lookalikes, custom audiences—the algorithm would optimise within that bucket. Andromeda flips this. It's designed to find relevant people outside your defined audience.

This sounds backwards. You're spending money to find people outside your target? But here's what Meta learned: in a world of Advantage+ and broad targeting, tight audience segments are often the constraint, not the feature.

What this means: interest-based targeting and narrow lookalikes become less powerful. If you're still relying on detailed targeting stacks, you'll feel this immediately. Broad campaigns and Advantage+ will pull ahead. Campaign structure matters more than audience definition.

3. Real-Time Bidding Adjustments (The Cost Play)

The old bidding system was relatively static. You set a cost cap or bid cap, and the algorithm tried to hit it. Andromeda introduces more granular, real-time adjustments. It's able to bid differently based on time of day, user behaviour signals, and predicted conversion likelihood.

What this means: your cost cap or bid cap is no longer a hard ceiling. On high-intent moments, the algorithm may spend slightly above cap. On low-intent moments, it'll spend below. The quarterly average should hit your target, but day-to-day variance will increase. This is good news if your products convert at different rates in different contexts—the algorithm can now hunt for those micro-moments.

Why This Actually Matters for Your Account

Most algorithm updates feel like noise. This one isn't. Here's why:

Creative velocity is now table stakes. If the algorithm is retraining faster and responding to trends quicker, static creative will fatigue faster. You need a system for continuous refresh. The agencies and brands managing creative pipelines—rotating 4+ new ads per week—will see Andromeda as an upgrade. Everyone else will see performance dips they can't quite explain.

Broad campaigns beat narrow ones. If you've built your account around tight audience stacks and interest combinations, Andromeda is a headwind. The algorithm is being designed to ignore those boundaries. Advantage+ and broad targeting will become even more efficient. This isn't a forever rule—it's just the Andromeda era of Meta.

Predictability is lower in the transition. For the next 6–8 weeks after Andromeda rolls out fully, expect performance volatility. The algorithm is retraining. Your account's signals are being reweighted. CPAs will move. ROAS will swing. This is normal. Most agencies and brands panic here and start optimising mid-transition, which makes it worse. The right move is usually patience.

What You Should Do Right Now

1. Audit your audience structure. If you're running detailed targeting stacks (10+ interest combinations, small lookalikes, heavily excluded audiences), you're betting against Andromeda. Not necessarily a losing bet, but a headwind. Consolidate where possible. Test broad versus detailed targeting on the same campaigns at equal spend and measure which wins.

2. Tighten your creative pipeline. If you're testing one new ad per week, you're moving too slow. The old algorithm would keep a winner alive for months. Andromeda will age creative faster. Move to 3–4 new concepts per week minimum, with a clear Day 5 evaluation process. The teams that were already doing this will barely notice the update. The teams that were relying on static creative will feel it immediately.

3. Expect the dip and plan for it. When Andromeda fully rolls out, performance will shift. CPAs might rise 10-15%. ROAS might drop temporarily. Budget doesn't change, but efficiency does during transition. Communicate this to leadership now, before it happens. Have a plan to hold spend steady and let the algorithm restabilise rather than panic-scaling or cutting spend. The brands that maintained their budgets through the iOS update (2021) ended up ahead. The ones that cut spend didn't recover as fast.

4. Monitor daily rather than weekly. For the next 60 days post-update, switch to daily performance monitoring on key accounts. Weekly review cycles are too long during algorithm transitions. You want to catch structural problems fast—a campaign that's scaling the wrong audience, for example—but not overoptimise in panic. Daily eyes, slow hands.

What Andromeda Means Long-Term

Algorithms trend toward automation and away from manual controls. Andromeda is Meta saying "we can optimise faster than humans can configure campaigns." This means:

Campaign structure becomes less important than it was. Bid strategy becomes less about your cap and more about your learning signal. Audience definition becomes less about precision and more about providing context.

What becomes more important: creative quality and velocity. Data. Conversion event accuracy. Keeping the algorithm fed with good learning signals instead of starving it with over-exclusion.

The best teams won't be the ones managing every campaign toggle. They'll be the ones producing creative at scale, maintaining clean conversion data, and giving the algorithm clear objectives to optimise toward.

The Bottom Line

Andromeda isn't a threat if you're already running clean, broad campaigns with strong creative velocity. If you're still optimising like it's 2020—tight audiences, heavy exclusions, static creative—the next 90 days will be uncomfortable.

Start your creative pipeline now. Simplify your audience. Monitor closely through the transition. You'll be ahead of the brands that wait until performance drops to react.

Book your Growth Diagnostic Call if you want a specific audit of how Andromeda will affect your account structure. We'll walk through your current setup and show you exactly where the algorithm shift will hit hardest.