Meta’s Andromeda update – what does it really mean?
- Jacob Bennett
- Reading time: 6 minutes
Launched late last year, Meta’s Andromeda update completed its roll out in October 2025. This isn’t a world-altering change – audiences have mattered less in Meta for some time now, whilst creative has grown into its role as a key targeting and performance lever. However, alterations to the back end of how Meta operates does mean that new tactics are required to get the best out of the platform.
It has also led to an onslaught of social media gurus on Linkedin claiming to have the secret formula to success, often advising that you completely tear up your existing structure for their shiny new Andromeda-optimised structure. Ultimately, as always, what works best for each business will vary. Incremental testing is the key approach, not sudden and drastic change.
With that in mind, we’ve put a quick guide together to help you understand what the implications are for your business and investments into paid media.
Technically speaking, it’s the shiny new AI ads retrieval engine. In layman’s terms, it’s the new brain and decision maker on how ads are served across Facebook & Instagram.
The engine predicts at the individual level instead of the historical ‘bucket’ based approach. I.E YOU will be served ads, not a 25-34 female interested in fitness. By product of this, the platform can process far more ad creative and user signals than before. Meaning the old way of uploading lots of creatives with minimal changes between them (such as a CTA), is no longer the main play.
This means Meta is now far better at detecting the nuances in your existing creative, and finding the right users to match those nuances with.
Creative has cemented its place as the key signal. The stronger, richer and more varied your creative assets, the better the algorithm can learn who responds to what and drives your campaign goal. Limited experimentation will struggle to generate performance as the algorithm requires diversity to personalise.
In short, move away from iterative tweaks of your creatives to conceptual diversity.
Targeting / Audience will matter even less as Andromeda leans on broad audiences & AI based matching. The effectiveness of narrow interest, lookalike or custom audience targeting will likely continue to drop, although this should be assessed on an account-by-account basis.
Performance may be volatile as the new engine learns and gathers enough data at the creative level to optimise effectively.
Effectiveness of Advantage+ should improve as this retrieval engine is built for the creative/audience variety that this campaign type brings. In theory, it should mean Meta’s AI creative enhancements will be better, but the jury is very much out on this.
Ultimately, the key takeaway is that we’re now confirmed to be in a world where creative volume and diversity rules over meticulous audience segmentation and micro optimisation.
First up, Creatives:
Your approach to creative is now comfortably the most important contributory factor towards performance on the platform. And that is here to stay.
It has also led to an onslaught of social media gurus on Linkedin claiming to have the secret formula to success, often advising that you completely tear up your existing structure for their shiny new Andromeda-optimised structure. Ultimately, as always, what works best for each business will vary. Incremental testing is the key approach, not sudden and drastic change.
With that in mind, we’ve put a quick guide together to help you understand what the implications are for your business and investments into paid media.
What is Andromeda?
Technically speaking, it’s the shiny new AI ads retrieval engine. In layman’s terms, it’s the new brain and decision maker on how ads are served across Facebook & Instagram.
The engine predicts at the individual level instead of the historical ‘bucket’ based approach. I.E YOU will be served ads, not a 25-34 female interested in fitness. By product of this, the platform can process far more ad creative and user signals than before. Meaning the old way of uploading lots of creatives with minimal changes between them (such as a CTA), is no longer the main play.
This means Meta is now far better at detecting the nuances in your existing creative, and finding the right users to match those nuances with.
What does it mean for my paid media campaigns on Meta?
Creative has cemented its place as the key signal. The stronger, richer and more varied your creative assets, the better the algorithm can learn who responds to what and drives your campaign goal. Limited experimentation will struggle to generate performance as the algorithm requires diversity to personalise.
In short, move away from iterative tweaks of your creatives to conceptual diversity.
Targeting / Audience will matter even less as Andromeda leans on broad audiences & AI based matching. The effectiveness of narrow interest, lookalike or custom audience targeting will likely continue to drop, although this should be assessed on an account-by-account basis.
Performance may be volatile as the new engine learns and gathers enough data at the creative level to optimise effectively.
Effectiveness of Advantage+ should improve as this retrieval engine is built for the creative/audience variety that this campaign type brings. In theory, it should mean Meta’s AI creative enhancements will be better, but the jury is very much out on this.
Ultimately, the key takeaway is that we’re now confirmed to be in a world where creative volume and diversity rules over meticulous audience segmentation and micro optimisation.
What needs to change?
First up, Creatives:
- Strategy must move away from trying to ‘out target’ the system to instead trying to ‘out creative’ it.
- Upload more distinct creative angles, hooks, pain points & storylines.
- Mix formats across reels, carousels, UGC, demos, etc
- Plan regular refreshes to ensure audience signals stay fresh, with a clear testing plan.
- Simplified campaign structures are likely to return.
- We’ve seen a similar trend across Google over the last few years as automated campaign types and AI keyword matching have become the norm.
- Broad campaigns gives Meta the flexibility to lean on the creatives, which is what the algorithm wants.
- This is where Meta’s Value Rules come in. If some areas of the broad approach aren’t working (most likely identified by your own separate data), their Value Rules allow you to make adjustments by platform/demo/geo to attempt optimisation.
- For example, Meta may be pushing more female users than male users, despite your internal data saying that female users are much lower LTV.
- Value Rules can help reduce your bids on female users.
- New tools allow various methods of testing:
- Meta’s creative testing tool allows you to apportion part of your budget into specific creative tests, and then keep the winners. This can be done without losing the learnings/data on the winners.
- However, it may fit your budget/structure better to split the campaigns into Test vs Scale.
- Test campaigns are there to test new creatives. The winner of each test period is then added to the scale campaigns. The test campaign’s sole purpose is to help generate exposure for a new creative. Without a test vs scale split approach, you’ll find a data snowball effect in which select creatives hold the majority of volume because they hold the most volume.
- Scale campaigns hold the majority of spend and should be your performance drivers. If a creative isn’t working here, pause it and roll in winners from your test campaigns.
In Summary
The completion of Andromeda’s roll out signals the next stage in Meta’s ad engine journey, a journey that it has been on for some time (alongside Google). This isn’t a surprising change in the mechanics of how to get the best out of Meta, but the natural next step as it responds to a world where advancements in user privacy have restricted their options.Your approach to creative is now comfortably the most important contributory factor towards performance on the platform. And that is here to stay.
