Why does Meta's Andromeda favour agencies like Karmina

For years, advertising on Meta was built around a relatively clear logic: whoever targeted best, usually achieved better results. Agencies specialising in paid media grew by mastering audiences, interests, lookalikes, exclusions, funnels, and complex campaign structures. The key differentiating value lay in finding the right audience and technically optimising campaigns better than the competition. Creativity was important, evidently, but it often acted as a secondary piece within a system where the real prominence was in targeting.

This scenario is undergoing a radical change with the arrival of Andromeda, Meta's new AI-powered ad recovery engine. And the change is much more profound than it appears. We're not just talking about a technical improvement on the previous algorithm, but a complete transformation in how Meta interprets relevance, decides which ads deserve to enter the auction, and determines which brands effectively gain distribution within Facebook and Instagram.

The most important consequence of this change is that Meta is relying less and less on manual targeting and more on the creative ability of brands and agencies to generate content that naturally fits user behaviour on the platform. The algorithm no longer solely prioritises who has the best-configured audience, but who is capable of producing a creative ecosystem strong enough to generate constant positive signals. And that particularly favours structures such as Karmina, where strategy, creativity, content, branding, video, social media, and performance work as a single integrated team.

Andromeda introduces a very different logic to what existed a few years ago. Previously, practically any advert could enter the auction if the campaign setup was correct and the bid was sufficiently competitive. Now, there's a much more demanding preliminary filter where Meta analyses whether that specific creative makes sense for that specific user at that particular moment. If the system considers the piece to be not relevant enough or not fitting within the user's browsing context, the advert directly doesn't enter into competition, regardless of the budget behind it.

This means that a campaign's performance is increasingly dependent on elements such as the initial hook, editing pace, visual style, narrative, video structure, copy, tone, format, and the content's ability to appear organic within the feed. Creativity ceases to be solely an aesthetic concern and becomes algorithmic infrastructure. Each piece feeds the system, generates signals, and helps Meta understand what type of users might react positively to that content.

The Andromeda era rewards agencies where creativity and performance literally work together.

And here lies one of the major problems with many traditional paid media structures. For years it was viable to work with fragmented models where one agency handled ads, another produced content, design was outsourced, video was made as needed, and strategic decisions were taken separately. That model is becoming less and less effective because Andromeda (presumably a platform or a project) requires speed, creative volume, iteration capability, and constant coordination between people who traditionally worked in isolation.

Here at Karmina, we've been seeing this change in accounts for some time, where the monthly investment in media no longer allows for working with slow dynamics or separate structures. A very clear example is Tiki-taka Play, where we discuss campaigns with monthly investments running into thousands of euros and where small variations in CTR, CPM or conversion can represent enormous differences in turnover. In this type of account, success does not depend solely on launching campaigns and waiting for results, but on the ability to react practically in real-time to how the algorithm is interpreting the creatives.

What completely changes performance is that the person analysing campaigns and optimising Meta Ads doesn’t work in isolation. They literally have, two chairs away, the designer, the video editor, or the content team who are capable of modifying creative assets immediately based on the campaign’s behaviour right next to them. This means that an optimisation meeting that starts at ten in the morning can result in twenty new adapted creative assets before the end of the day. The team detects patterns, identifies which hooks are performing best, which messages are losing momentum, which formats have the most retention, and which assets are entering the auction correctly according to the algorithm’s signals. And this information is quickly turned into new creative assets without going through lengthy processes, email chains, or weeks of external validations.

This is not merely an organisational improvement. It is a structural advantage in the Andromeda era. The algorithm needs constant creative diversity to learn correctly. It needs distinct advertisements, not simple minor visual adaptations. It needs different formats for different moments in the funnel, specific messages according to context, and pieces capable of continuously generating new signals. When an agency can produce, analyse and modify content in a coordinated way within the same team, the system learns faster and optimises better. When each modification depends on third parties, freelancers or fragmented processes, learning slows down and campaigns lose competitiveness.

Another very representative case of how we work this logic is Pans & Company. In these types of accounts, the line between organic content, creativity, and performance practically disappears. The community manager, the ad campaign analyst, and the designer work together, sharing the same creative space, simultaneously seeing which ads are performing well, which organic content generates the best engagement, and which pieces have the potential to quickly become performance campaigns.

Many times, a quick five or ten-minute meeting, in person or even online, is enough to redefine creative approaches, adapt hooks, modify visual structures, or generate new versions aimed at improving CTR and conversions. Working in the same environment allows decisions to be made not solely from a technical logic, but from a global vision of the brand's social behaviour. The designer understands what performance needs. Performance understands how content thinks. Community understands what language connects with the audience. And all of this happens practically in real time.

Major brands are no longer competing solely on investment, they are competing for relevance within the feed.

This type of dynamic is precisely what Meta is indirectly encouraging with Andromeda. Because the platform's real objective isn't simply to sell advertising. Meta's business depends on retaining user attention and preventing them from abandoning the feed. Although brands pay to appear, the platform has no interest in flooding Facebook or Instagram with ads that appear artificial, repetitive, or disconnected from the current visual codes of the internet.

Meta makes billions of dollars in profit. It doesn't need to sacrifice user experience to maximise short-term revenue. That's why Andromeda prioritises pieces that behave like content and not solely like advertisements. And that forces brands to build much more integrated creativity within real social dynamics.

It is also important here to understand the role of artificial intelligence. Meta is betting heavily on tools capable of generating automatic adverts, text, backgrounds, visual adaptations, and quick pieces for small businesses. And it makes sense. Many small businesses cannot afford complex creative structures and need economical, functional solutions to enter the advertising ecosystem.

But the big brands play in an entirely different league. When a company invests hundreds of thousands of euros a year in advertising, a ten per cent improvement in conversion, CTR or costs can represent a huge amount of money. And at that competitive level, it's no longer enough to automate parts or generate generic ads with AI. The important thing is to build content with personality, judgment, identity, and the ability to stand out in completely saturated feeds.

Our vision for artificial intelligence is precisely in that direction. We use AI to accelerate processes, optimise coordination, improve timings and expand productive capacity, but not to replace strategic thinking or turn brands into indistinguishable generic content. Because the more algorithms like Andromeda evolve, the more important that which appears human, authentic, and difficult to replicate automatically becomes.

This is probably why more and more projects are looking for structures capable of integrating creativity, branding, video, social media, and performance under a single working system. And it also explains some of the awards we've recently received at The Communicator Awards, where projects developed for brands such as Pans & Company, BrewDog o Glow White have been recognised precisely for that ability to unite brand building, social creativity and campaigns designed to work within the actual codes of current digital platforms.

Because the big conclusion of this new stage of Meta Ads is likely this: artificial intelligence is not eliminating the value of creativity and brand storytelling. It's doing the exact opposite. It's forcing creativity, content, branding, and performance to work as a single connected system. And that particularly benefits agencies capable of working in an integrated, rapid, and coordinated way around how real people consume content on the platforms.

AUTHOR

Karmina Team

We are a group of experts committed to helping brands grow through effective digital strategies. With a clear vision and a collaborative approach we offer customised solutions that respond to the specific needs of each client.

If you liked this content, share it!