Meta has quietly rolled out Andromeda, a new AI-driven ad delivery engine that changes how ads are matched to users across Facebook and Instagram. Instead of following audience targeting rules, Andromeda uses machine learning to evaluate millions of ad-user combinations and determine which ad is most likely to convert for each person – in real time.
Previously, Meta advertisers would create multiple campaigns with tightly defined audiences – by age, gender, location, interests, behaviours, and custom/lookalike audiences. That structure offered control but often led to overlapping delivery, fragmented budgets, and slow optimisation. Meta is now prioritising creative content, engagement data, and user behaviour signals over manual targeting rules. In other words, your creative now plays the biggest role in who sees your ads.
What Has Changed in Meta Ads Targeting?
Where the old system looked at a few hundred ads and matched them to users based on the targeting parameters advertisers set, Andromeda’s AI-driven retrieval engine now sifts through millions of ad candidates, analysing every creative element from your video pacing to the emotional tone, messaging, format, and visual style. Andromeda isn’t just looking at who should see your ad – it’s determining which specific ad each person should see based on their behaviour patterns.
For example, a brand story video might naturally reach users who respond to emotional storytelling, while a product demo could be shown to users actively searching for solutions. With Andromeda, the variety and relevance of your creative matter more than audience segmentation. Creative diversity becomes the new “targeting” lever: multiple creative concepts that speak to different audience motivations, contexts, and stages of awareness, as well as mixed formats (video, static images, carousels) – not just variations of the same ad.
If you give Andromeda only one creative concept, it has little to optimise from. What Andromeda wants is creative diversity and differentiation. The results seem promising, though. Meta reports that Andromeda has improved ad quality by 8% and boosted ad recall by 6%.
What This Means for PPC
This is definitely a mindset shift for advertisers. For years, we’ve spent many hours perfecting campaigns, ad sets, and audience groupings. Andromeda takes on a lot of that work. And instead of manual tweaking, advertisers need to trust the algorithm to learn. Andromeda takes time to optimise (often several days), but the results tend to improve over time as the system refines its understanding of audience behaviour.
That means:
- Less micro-targeting: Broad campaigns will perform better because Andromeda learns faster with larger data pools.
- More creative variety: Multiple ad formats and concepts give the system more material to work with.
- Better data signals: Accurate conversion tracking and server-side event data help the algorithm understand what success looks like.
Adapting to Andromeda
With less focus on micro-targeting and constant optimisation, Meta’s Andromeda update is an opportunity to simplify campaign structures, focus on creative, and let the algorithm do the heavy lifting. Here’s what we’re focusing on at Herdl:
- Prioritising broad targeting, creative variety, and strong data signals for new campaign setups.
- Auditing and restructuring legacy setups to reduce micro-targeting, consolidate ad sets, and simplify learning.
- Gradually shifting ad budgets to the new setup and monitoring performance (pausing legacy campaigns once the new structure is stable and delivering reliably).
- Scaling creative development (more formats, more messaging angles, and more variations) to give the algorithm enough material to optimise effectively.
- Auditing data infrastructure to make sure tracking, events, and conversions are set up accurately.
The Last Word
If you’d like to chat about adapting your advertising strategy for Meta’s Andromeda update, get in touch with the Herdl team.