Andromeda is not just another minor tweak to the algorithm, it is a fundamental shift in how the platform functions. So, it’s important to understand if you’re going to excel at ads on Meta. To help you really “get it”, we have broken down everything you need to know about this AI-powered overhaul.

What is the Andromeda update?

In the simplest terms, Andromeda is Meta’s new ad retrieval engine. To understand its impact, we first need to look at how ads used to work.

Historically, Meta’s ad delivery followed a ranking-first approach. When a user opened an app, the system looked at the targeting parameters set by the advertiser, such as interests, age, and location, and then tried to rank all eligible ads within that specific pool.

The problem was scale. With the introduction and rise of AI-generated content and Advantage+ campaigns, the number of ad variations exploded. The old system was trying to sift through millions of ads for every single impression, creating a massive technical bottleneck for Meta itself.

Andromeda fixes this by introducing a sophisticated “retrieval” stage before ranking occurs. Andromeda can scan tens of millions of potential ads in milliseconds. It narrows this massive pool down to a shortlist of a few thousand highly relevant candidates. Only then does the traditional ranking and auction process begin.

Moving from targeting rules to targeting suggestions

Before Andromeda was fully realised, we saw the beginning of this shift with the introduction of “Advantage+ Audience.” This was the moment Meta began moving away from treating your targeting parameters as strict rules.

In the old days, if you told Meta to show your ad only to “Business Owners in Belfast” the system would treat that as a hard border. If a perfect customer was a fan of golf but had not been “tagged” with that interest yet, they were excluded. This actually worked well historically when it came to testing Meta’s targeting-first approach, across a range of segmented adsets.

With the move towards Andromeda, Meta started viewing your chosen interests and demographics merely as targeting suggestions. The system now says: “Thanks for the hint, but if I find someone outside of this group who is highly likely to convert, I am going to show them the ad anyway.”

This worked into the Andromeda update by essentially removing the “blinkers” from the AI. Instead of being confined to a small, manual pool of users, the retrieval engine can now look at the entire landscape. It uses your targeting as a “seed”, but it relies much more heavily on how people actually interact with your content to find the next customer.

The old way vs the Andromeda way

To help you visualise how much the landscape has shifted, we have compared the traditional ad delivery system with the new Andromeda architecture below.

The 3 things Meta’s Andromeda optimises for

Andromeda does not just look for the highest bidder. It uses a sophisticated balancing act to decide which ad earns a spot on a user’s feed. To win the auction, your ad needs to perform well across these three core metrics:

(Table info source: https://nestcommerce.co/resources/what-is-meta-andromeda/) 

Why did Meta build a new engine?

Meta’s engineering team did not build Andromeda just for the sake of it. The update was a direct response to several major shifts in advertising:

• The explosion of creative volume: Tools like Advantage+ Creative now allow advertisers to generate a lot of variations of a single ad. The old infrastructure simply could not process this volume efficiently.
• Privacy and data restrictions: With the decline of third-party cookies and the impact of iOS privacy updates, manual interest-based targeting has become less reliable (this one’s been a problem for quite a while now!) Meta needed a way to find audiences using the signals it still has access to: the content of the ads themselves.
• The need for real-time relevance: Consumer behaviour moves fast. Andromeda is designed to understand what a user is interested in right now, based on their very latest interactions, rather than relying on demographic data.

Creative is now your targeting

This is the most critical takeaway for any business owner or marketer. In an Andromeda-powered world, your creative assets (the images, videos, and copy) are the primary instructions you give to the algorithm.

Under the old system, you told Meta who to target using settings in Ads Manager. Under Andromeda, the AI “reads” your creative to figure out who the audience should be. If your video features a modern kitchen and someone meal-prepping, the AI identifies these visual and contextual signals. It then retrieves that ad for users who have recently shown an interest in healthy eating or home improvement, even if you did not select those interests in your targeting.

What this means for your strategy

The manual era of digital marketing is fading. Success now depends on how well you can signal intent through your content.

1. Broad targeting is the new gold standard: We are increasingly seeing that broad targeting (leaving interests and lookalikes blank and only setting age, gender, and location) outperforms hyper-segmented campaigns. This gives Andromeda the freedom to find your customers wherever they are. This, however, doesn’t mean you shouldn’t test other audience types in your account. 
2. The shortlist hurdle: If your ad does not clearly communicate what it is and who it is for in the first few seconds, it may fail the retrieval stage. If you do not make the shortlist, your ad never even gets to enter the auction, no matter how high your budget is.
3. Creative similarity penalties: Andromeda is smart enough to know when two ads are essentially the same. If you only change a single word in a headline or use the same image with a different filter, the system may treat them as duplicates. To win, you need genuine conceptual true variety. This means minor-element testing is no longer as effective.

How to adapt your Meta ads for 2026

To help you get the best ROI from this update, we recommend a shift in how you structure and manage your accounts. The focus has moved from “media buying” to “creative strategy.”

1. Simplify your account structure

Fragmentation is the enemy of AI. In the past, advertisers would create dozens of ad sets within a single campaign to test different small audiences. Andromeda thrives on data density. By consolidating your budget into fewer campaigns and ad sets, you give the algorithm the volume of data it needs to learn and optimise.

2. Move from tweaks to “angles”

Instead of testing minor variations, start testing entirely different angles or hooks (true variation). For example, if you are selling a skincare product, your angles might include:

• The problem/solution angle: Focusing on clearing up a specific skin issue.
• The social proof angle: Customer reviews and before-and-after shots.
• The lifestyle angle: A high-quality video showing the product as part of a morning routine.
• The educational angle: A “founder-led” video explaining the science behind the ingredients.

Andromeda will test these against each other and naturally find which message resonates with which sub-audience.

Examples of what we mean by creative variation

Because Andromeda uses your creative as the primary targeting tool, you cannot simply swap a headline and call it a “variation”. The system looks for distinct “Entity IDs”, meaning it wants to see content that is visually and contextually different. If you upload creative with only small tweaks, it will identify it as duplicate, so we need to focus in one true variation. Here’s an example of what we mean by true variation:

3. Focus on “thumb-stop” moments (aka Hook Rate)

Because the retrieval engine is making split-second decisions based on engagement signals, the first three seconds of your video are more important than ever. If users scroll past immediately, Andromeda will quickly stop retrieving that ad. You need a strong visual or verbal hook to prove your relevance to the system.

Key performance indicators in the Andromeda era

When reviewing your reports, you may need to look at different metrics to gauge whether the update is working for you.

The creative diversity calculation

To get an accurate picture of your diversity, you should look at a specific campaign or your top-performing ad sets over the last 30 days. The calculation is as follows: