Why Your Facebook Ads Are Failing

Whether you spend $2k a month, or $2m, this is the strategy is becoming the go-to.

By Alex Murton | Read time: 5 mins

Hey, it’s Alex.

This week I want to share the Meta Andromeda strategy that is making the rounds on YouTube with many of the top global performance strategists.

This is a strategy that anyone can run - because now - creative and perspective, are the only true advantages.

Let’s go.


🧠 THE THEORY

Why Your Facebook Ads Are Failing: The Andromeda Update Explained

If your Facebook ads have been performing terribly lately, you're not alone.

There's a specific reason behind the widespread performance drops, - the Andromeda update.

This new algorithm change from Meta has fundamentally altered how Facebook sorts through millions of ads and decides which creatives to show to which users. ​

What Is the Andromeda Update?

The Andromeda update is Facebook's new AI-powered system for ad selection, specifically targeting the ad retrieval process.

While the article announcing this update was published in December 2024, it wasn't fully rolled out until July 2025 - which explains why many advertisers experienced a brutal July.​

Here's the scale we're dealing with: Facebook has to sift through tens of millions of ads and decide which 3-5 ads you'll see within your next few scrolls, all within 200 milliseconds of opening the app.

It's an insane amount of data processing that reinforces why Facebook is fundamentally a data platform, not just an ad platform.​

The Four-Step Ad Selection Process

Facebook's ad selection follows a systematic four-step process:​

  1. Retrieval: Filtering tens of millions of ads down to thousands (over 90% elimination)

  2. Light Ranking: Narrowing thousands down to hundreds

  3. Final Scoring: Ranking hundreds of ads based on "total predicted advertiser value"

  4. Auction: Direct competition between advertisers for the final impression

The Andromeda update specifically revolutionises step one - the retrieval process - which is the most data-intensive phase.​

There is a point for this new creation

Why Facebook Created Andromeda

The primary driver behind this update?

AI-generated creative content.

Facebook openly admits that the massive surge in AI-created ads forced them to develop an entirely new system. The volume became too much for their previous algorithm to handle efficiently.​

Facebook states that "the sheer volume of ad candidates is three orders of magnitude more than previous stages" due to increased advertiser automation with generative AI.

Essentially, marketers contributed to this problem by flooding the platform with AI-generated variations.​

Old System vs. New System

The Old Model:

  • Manual approach: Used smaller, individual rules for sorting ads.

  • Inefficient scaling: Looked through variations in-depth individually.

  • Basic computing: Ran on regular computers.

  • One-by-one analysis: Like reading every book title in an unorganised library..

The Andromeda Model:

  • AI-powered brain: One intelligent system handles all sorting.

  • Supercomputer processing: Uses Nvidia's Gracehopper super chips for higher computational power.

  • Group organisation: Clusters similar ads together for efficient processing.​

What This Means for Your Ad Strategy

This change can seem overwhelming - but there is a clear path to victory

1. Creative Diversity Is Now Critical

The biggest shift: stop testing minor variations and start creating truly diverse concepts.​

If you've been running five variations of the same video with different hooks, that approach is now counterproductive.

Facebook groups similar creatives together, so those variations won't be individually tested for performance differences.​

New approach: Focus 80% on new concepts, 20% on variations (this is completely flipped from the previous 80/20 rule).​

2. Creative Is Your New Targeting

With Andromeda's AI analysing every aspect of your creative - from colours and sentiment to demographics visible in images - your creative becomes your targeting mechanism. Facebook uses AI to examine:​

  • Visual elements (flowers, mountains, backgrounds)

  • Ad copy sentiment analysis

  • Demographics in images (skin tone, gender, age)

  • Every piece of visual and textual information.

3. Quality Over Quantity

When asked whether they'd rather test 300 creatives monthly or 20 high-quality, completely unique concepts, every knowledgeable advertiser chooses the 20. The algorithm now heavily favours:​

  • Personalised ads over mass-market content.

  • Specific avatar targeting over broad demographic appeals.

  • Unique concepts over endless variations.​

4. More Ads Per Ad Set

You can now safely load 25-50 creatives into a single ad set without breaking the algorithm, thanks to improved creative diversity handling.

Just ensure they're genuinely different concepts, not 50 variations of the same video or creative.​


New Concepts vs. Variations

New Concepts (What You Want More Of):

  • Format changes: VSSL to static image with long-form copy

  • Angle shifts: Different marketing angles targeting different customer motivations

  • Structure variations: Completely different creative layouts.​

Variations (Use Sparingly):

  • New hooks on existing video ads

  • Different thumbnails for the same video

  • Minor copy adjustments

  • Small creative tweaks.

It’s show time


The Bottom Line

The Andromeda update isn't revolutionary - it's the natural evolution of Facebook's long-standing direction toward more consolidation, less advertiser control, and greater emphasis on creative quality.​

Instead of spamming 500 variations of the same creative, successful advertisers are now creating diverse, unique concepts that stand out in Facebook's new AI-powered sorting system.

Key takeaway: If you focus on high-quality, diverse creatives and solid funnel optimisation while adapting to consolidated campaign structures, you'll not only survive the Andromeda update - you'll thrive while your competitors struggle to adapt.​

The advertisers who understand and embrace these changes will find themselves with a significant competitive advantage as the majority of the market continues to use outdated strategies that no longer work in this new AI-driven landscape.


 Do This

Step 1: Optimised for engagement

Create a campaign in Meta that is optimised for engagement, and set it for ad set budget optimisation.

Create 2 visuals (not variations - to complete different designs), and then have 3 different value sets on each visual.

This will result in 6 ads in total.

Create a campaign ad set in Meta, chuck your ads in, and go.

See which ones great the most engagement and the most spend.

This is meta serving them up, and creating the winners for you.

Step 2: Optimised for conversion

Create a campaign that is optimised for conversions, set for campaign budget optimisation.

Take the winners from step 1, and drop them into this campaign.

Scale these ads.

Kill the losers in Step 1.

Throw in new ads.

Repeat over and over again.

Problem > Solution > Product > Conversion

Gone are the days of creating a funnel within Meta structured by campaigns, ad sets and all of the above.

Meta will now do this for you.

You only need 2 groups (listed above).

Create ads that talk to the problem you are solving.

New other ads that talk to a solution - not yours - just “what a solution could be”.

Create ads that point to a product as a solution - yes you.

Then finally, have ads that focus on conversion - that is it.

Meta will serve the conversions ads to the right people in the right phase of the funnel.

Repeat. Repeat. Repeat.

💪 DONE FOR YOU


Nail your creative concepts

If you didn’t know - I run ecommerce agency - Studio Almond - and yes - we create our own ads.

When it comes to getting your creative competitive edge, you might need some help to stand out from the crowd.

Need a hand - get in touch.

📣 HAVE YOUR SAY

How did we do?

?What did you think of today's issue

Login or Subscribe to participate in polls.


Thanks for reading!

See you next week.

Alex Murton
Shopify for Lifestyle Brands
Co-Founder @ Studio Almond & Almond Labs.