Meta's Andromeda update is the most significant change to how Facebook and Instagram ads are delivered since the platform launched. It is not a tweak to the bidding system or a new campaign type. It is a fundamental rebuild of the ad retrieval engine - the system that decides which ads are even considered before ranking happens. By October 2025, Andromeda was fully deployed across most objectives and placements. By Q1 2026, it was the default behaviour across every DTC and ecommerce Meta account. If you are running paid social for a DTC brand, you are running it under Andromeda now, whether you know it or not.
This post explains what Andromeda changed, what it means practically for DTC brands, and what the brands adapting well are doing differently.
What Andromeda actually is
Before Andromeda, Meta's ad delivery worked like this: advertisers selected audiences (demographics, interests, lookalike seeds), and the algorithm optimised delivery within those boundaries. The advertiser controlled who saw the ad. The algorithm controlled when and how often.
Andromeda inverted this. The system now reads the creative itself - the visual content, the language, the format, the context - and uses those signals to determine who should see it. Your creative has become your targeting. The audience parameters you set are still processed, but Andromeda's creative-signal layer increasingly overrides narrow audience constraints in favour of what the AI predicts will perform. The result is that two ads in the same campaign targeting the same audience can reach very different people depending on what the creative communicates.
The engineering reason for this is scale. The volume of ads on Meta increased by roughly 10,000x as AI-generated creative proliferated. The old retrieval system, which evaluated ads primarily against audience match, could not process that volume. Andromeda was built to handle it - and in doing so, it shifted the entire model from audience-first to creative-first.
What changed for DTC brands running Meta ads
Creative is now the targeting
The practical implication is the most significant shift in paid social strategy in a decade. A DTC brand whose paid media approach was built on audience segmentation - multiple ad sets targeting different interest segments, lookalike audiences layered by spend tier, manual exclusions to prevent overlap - is now working against the algorithm rather than with it. Andromeda's retrieval system needs signal volume to learn. Fragmenting that signal across dozens of tightly defined audience segments starves the system of the data it needs to optimise.
The approach that works under Andromeda: broad targeting, simplified campaign structure, and a diverse creative library that gives the algorithm multiple distinct signals to work with. A single campaign, one or two ad sets, broad or Advantage+ audience settings, and a library of creatives that represent genuinely different hooks, formats, and messages. The algorithm finds the audience. Your job is to give it enough varied creative to find multiple audiences at once.
Creative diversity is the primary lever
Under the old system, creative fatigue was manageable - a well-targeted ad to a defined audience could run longer before performance degraded because the same people were seeing it repeatedly. Under Andromeda, the algorithm needs creative variety to maintain delivery efficiency. A single hook in three minor variations is not diversity - the system reads the underlying creative signal and sees the same ad three times. Distinct ideas, distinct formats, distinct emotional angles are what constitute diversity in Andromeda's terms.
For DTC brands, this means the creative production model needs to change. The volume required is higher - typically 8 to 15 distinct creative concepts running simultaneously rather than 3 to 5. The emphasis shifts from polish to distinctiveness. A slightly rough UGC video that communicates a genuinely different angle outperforms a polished brand video that says the same thing as the last three ads. Andromeda's AI reads authenticity signals in creative and weights genuine content higher than highly produced assets that pattern-match to advertising.
UGC has become structurally more valuable
User-generated content performs disproportionately well under Andromeda for a specific reason: it contains genuine creative signals that the system can read clearly. A customer talking directly to camera about their experience with a product - authentic, specific, unstyled - gives Andromeda a rich set of signals about who this product is for and why someone would want it. A brand-produced studio creative gives the system a polished visual and a marketing headline. The former is more readable, more trust-signalling, and more distinct from the sea of AI-generated advertising the system now processes.
For DTC food and drink brands, this means building systematic UGC collection into the creative pipeline - through post-purchase Klaviyo flows that prompt customers to share, through creator partnerships that produce authentic content rather than polished brand deals, and through product seeding to micro-creators whose audiences overlap with the target customer. The shared media layer - genuine customer content - feeds paid performance under Andromeda more directly than it ever did under the previous targeting model.
Advantage+ Shopping is the campaign structure that fits
Meta's Advantage+ Shopping campaigns (ASC) were designed to work with Andromeda's architecture. ASC removes the manual structure - multiple ad sets, lookalike layers, manual bid controls - that fragments the signal Andromeda needs, and hands audience delivery entirely to the algorithm. Most DTC brands that have restructured to ASC under Andromeda report improved ROAS relative to manual campaign structures, particularly for prospecting. The trade-off is reduced advertiser control over who sees the ads, which requires trust in the algorithm's optimisation - trust that the data suggests is mostly warranted when the creative library is genuinely diverse.
ASC is not the right structure in every situation. Retargeting of known subscribers and purchasers, where the audience definition carries more value than Andromeda's creative-reading, can still benefit from manual structure. But for top-of-funnel prospecting - the primary use case for most DTC Meta spend - ASC running against a diverse creative library is the architecture most aligned to how Andromeda actually works.
What no longer works as well
The strategies that built DTC brands on Meta through 2020 to 2023 are significantly less effective under Andromeda. Narrow interest targeting produces fragmented signal and limits the algorithm's ability to learn. Multiple lookalike audiences stacked at different percentage thresholds produce diminishing returns because the system is now doing that optimisation itself. Manual placements that exclude certain contexts reduce the data volume Andromeda needs to make accurate delivery decisions. Retargeting as a primary conversion strategy has weakened alongside the deprecation of third-party cookies.
The DTC brands struggling most under Andromeda are those still trying to micromanage targeting parameters rather than investing in creative quality and variety. The platform has changed what it rewards. The brands adapting are the ones that accepted the shift and focused on what they can control: the creative itself.
Andromeda and the Shopify stack
Andromeda can bring the click, but it cannot fix a poor post-click experience. The algorithm optimises delivery to people who are likely to take the desired action - but that prediction is based on historical conversion data from the pixel and the Conversions API. A DTC brand with a well-built Shopify Plus store that converts cleanly and feeds accurate purchase events back to Meta gives Andromeda better signal to optimise against. A brand with a slow store, a broken mobile checkout, or a subscription mechanic that adds friction gives the algorithm less useful data and gets worse delivery as a result.
The Conversions API (CAPI) integration between Shopify and Meta is more important under Andromeda than it was under the previous model, because server-side event data is one of the primary signals the system uses for optimisation. A brand relying entirely on browser-side pixel data is sending incomplete conversion signals. Proper CAPI setup - either through Shopify's native Meta integration or through a dedicated server-side setup - is a prerequisite for Andromeda performing well for a DTC store.
First-party data from Klaviyo - email subscriber lists, purchase history, subscription status - is also more valuable under Andromeda than before, because uploading high-quality customer data as custom audiences gives the algorithm a strong conversion signal to build on. A brand whose owned data feeds into Meta's audience infrastructure gives Andromeda a better starting point for creative-signal matching than one running entirely cold. See our post on owned, earned and paid media for how these data assets work together.
What to do differently now
For DTC brands running Meta ads under Andromeda, the practical changes that make the most difference:
Consolidate campaign structure. Move toward fewer campaigns with broader audience settings rather than many campaigns with narrow targeting. Give ASC a genuine test with a meaningful budget allocation before concluding it does not work for your account.
Build a diverse creative library. Aim for 8 to 15 genuinely distinct creative concepts running simultaneously - not variations on the same hook. Different problems, different formats, different emotional angles, different customer voices. Refresh the library regularly rather than running the same creative until it exhausts itself.
Invest in UGC systematically. Build post-purchase Klaviyo flows that prompt customers to share content. Identify 10 to 20 micro-creators in the product category for seeding. The authentic creative this generates is the input that Andromeda's creative-reading system values most.
Fix the post-click experience. Andromeda optimises toward conversion - it cannot perform well if the Shopify store, the product page, or the checkout is creating friction that prevents the conversion it is trying to deliver. The CRO work on the store is as important to Meta performance as the creative work on the ads.
Implement CAPI properly. Server-side conversion event data is one of the primary signals Andromeda uses for optimisation. Make sure Shopify's native Meta integration is active and sending purchase events, or work with a developer to set up a proper server-side implementation.
If you want to understand how Andromeda is affecting your specific Meta account - and what the creative and campaign structure changes look like in practice for a DTC brand at your stage - get in touch.
Frequently asked questions
What is Meta's Andromeda update?
Andromeda is Meta's new ad retrieval system, the engine that decides which ads are considered before ranking happens. It replaced the previous audience-based model with a creative-signal-based model - the algorithm now reads the content of your ad creative to determine who should see it, rather than relying primarily on the audience parameters advertisers set. It was announced in December 2024, rolled out through 2025, and was fully deployed across most objectives and placements by October 2025.
How does Andromeda affect DTC brand Meta ads?
Andromeda means creative is now the primary targeting mechanism. Narrow audience segmentation, multiple lookalike layers, and complex manual campaign structures are less effective because they fragment the signal the algorithm needs to learn. Broad targeting, simplified campaign structure (Advantage+ Shopping), and a diverse library of distinct creative concepts is the approach that works best under the new system. DTC brands that have not restructured their Meta approach since 2024 are likely underperforming relative to what is achievable under Andromeda.
What creative works best under Andromeda?
Creative that is genuinely distinct - different hooks, different formats, different emotional angles - and creative that contains authentic signals the algorithm can read clearly. UGC performs disproportionately well because it communicates genuine purchase motivation in a way polished brand creative often does not. Volume matters: 8 to 15 distinct creative concepts running simultaneously gives the algorithm more signals to match with different audience segments than 3 to 5 variations of the same theme.
Should DTC brands use Advantage+ Shopping under Andromeda?
For top-of-funnel prospecting, yes - ASC is designed to work with Andromeda's architecture and removes the manual structure that fragments delivery signals. Most DTC brands testing ASC with a genuine budget allocation and a diverse creative library report improved prospecting ROAS relative to manual campaign structures. For retargeting of known purchasers and subscribers, manual structure can still be appropriate. The two are not mutually exclusive.