Advertisers adapt by moving their effort from targeting to creative. Meta's Andromeda delivery system now relies on how people engage with creative as its main signal, so broad audiences of two to five million people work best and the ad itself does the targeting. The job is to build genuinely different creative for different personas, grounded in competitor and audience data rather than generic AI output, and then let Meta's automation find the buyers.
Key takeaways
- Interest targeting is now a loose starting point; the creative asset has become the main targeting tool.
- Small edits, such as changing a headline colour, make Meta's systems treat ads as identical and can trigger a creative limited status that raises acquisition costs.
- Meta asks for distinct creative styles built around different personas: varied people, real moments, behind-the-scenes stories and mixed formats.
- Generic AI tools like ChatGPT or Gemini make content fast but blind, because they work without market or competitor context.
- Grounded creative connects generation to performance data, competitor concepts and persona tensions, and scores ideas before launch.
Why has the strategic job moved back to the brand?
Artificial intelligence driven advertising platforms are moving digital marketing away from manual targeting toward creative strategy as the primary engine for performance. The digital marketing landscape in Southeast Asia and around the world has reached a major turning point. As we navigate the realities of 2026, the basic mechanics of digital advertising have been fundamentally changed by breakthroughs in algorithms and the rapid rollout of artificial intelligence.
For startup founders, small businesses, and growth teams, the days of generating performance through intense manual audience targeting and constant testing of minor copy changes are over. With the rise of the autonomous Andromeda delivery system from Meta and its major acquisition of the AI agent startup Manus, the strategic responsibility has moved completely away from the advertising platform and back to the brand.
Why is the algorithm now the audience?
To understand why this shift is necessary, leaders must recognise the decline of traditional media buying. The loss of reliable third party data forced Meta to rebuild its targeting system around its most powerful internal signal which is how users engage with specific creative content.
Interest targeting no longer strictly controls who sees an ad. Instead, it acts as a loose starting point. Broad targeting that defines an audience pool of two to five million users is now the most effective mathematical approach. This means the creative asset itself has become the main targeting tool.
This change brings a heavy penalty for campaigns that lack variety. When advertisers make only tiny changes, such as adjusting a single headline colour, the Meta systems view those assets as identical. This triggers a status called creative limited which causes acquisition costs to soar. Radical creative difference is now a mathematical requirement for keeping a business profitable over the long term.
How do you build real creative differentiation?
To avoid wasting your budget, Meta now requires the development of distinct creative styles built around different customer personas. True variety requires mixing standard product photos with energetic content from the real world. Including authentic moments, various people, and a lifestyle focus across different video and image formats is the secret to reaching new audiences and stopping ad fatigue.
Diversify your concepts and perspectives
Whether you are displaying clothing, a vacation spot, or a new food item, your ad visuals should provide unique viewpoints rather than small edits to the same image.
What to do: Feature different ways to use a product, real life moments, and a wide range of people across various ages and body types.
What to avoid: Do not rely on repetitive images with tiny changes, such as a slightly closer version of the same photo or a nearly identical product shot.
Show meaningful variety in your storytelling
Your campaigns should show the full range of your brand by moving past a single standard product shot. Highlighting different parts of your business, including the final product and the people who make it, helps you appeal to different customer interests.
What to do: Feature diverse products, different visual styles, and interesting behind the scenes stories like a chef cooking a meal or friends on a trip.
What to avoid: Do not use repetitive product shots with no conceptual change, like replacing a standard burger with a cheeseburger in the exact same photo.
Why isn't generic AI content enough?
Because brands need to create many different assets quickly, many have started using basic generative AI tools like ChatGPT or Gemini. However, the rise of these generic models has created a wave of boring and identical content that lacks strategic depth and brand personality.
Relying too much on unguided AI creates a situation that is fast but blind. The AI can write copy in seconds, but it works in a vacuum. It is disconnected from what is happening in the market or what competitors are doing. It provides the speed to grow but not the map to reach success.
What is grounded creative, and how does it work?
To survive this era of automation and generic content, advertisers must use a sophisticated method called generative grounded creative. This approach uses specialised AI models that are connected to historical performance data, competitor information, and real audience insights. Instead of guessing with a standalone model, a new category of tools acts as a navigator by connecting the creative process to hard market data.
How grounded AI works
Decoding Competitors: Computer vision and language processing can identify the high level concepts, visual styles, and psychological angles used by competitors rather than just looking at keywords.
Tension Mapping: Data is organised into specific customer personas paired with the emotional needs driving their behaviour. This stops brands from making wrong assumptions about demographics and helps generate content mockups known as creative expressions.
ROI Validation: Advanced scoring systems use math to check generated content and predict the return on investment before a launch. This prevents companies from spending money on ideas that will not work.
How do human strategy, grounded AI and Meta's automation fit together?
The modern marketing world is a collaborative space that requires the smooth integration of human strategy, grounded AI analysis, and the automated power of the Meta algorithms.
Because the technical side of media buying is now handled by tools like Manus, you must stand out before the media buy even happens. By using independent AI tools to find proven insights and create diverse briefs, founders can set a strong strategic foundation. Once these unique assets are ready, they can be given to the automated systems from Meta. This gives the algorithm the freedom it needs to find the best customers in places you might not expect.
How does SOMIN help brands stand out before the media buy?
In a world where Meta’s algorithms optimise delivery but cannot define differentiation, the real advantage lies in strategic intelligence before the media spend begins. This is where SOMIN closes the gap. Through its analytics SaaS ecosystem SODA, SOMONITOR (Brand Tracker, Content Library, Perspective Studies, SoInspire), and GWI brands can ground creative decisions in live market signals, competitor intelligence, persona-level tension mapping, and validated performance insights.
Instead of producing fast but blind AI content, teams can build data-backed creative expressions designed to feed automation with meaningful variety and strategic clarity. In the automation era, SOMIN ensures that brands do not just run ads efficiently, but enter the algorithm with insight, structure, and measurable advantage.
FAQ
Does interest targeting still work on Meta?
Only as a loose starting point. Meta rebuilt targeting around how users engage with creative, and broad targeting with an audience pool of two to five million users is now the most effective approach.
What does creative limited mean?
When ads differ only slightly, for example by a single headline colour, Meta's systems treat them as identical. That triggers a status called creative limited, which pushes acquisition costs up.
What counts as meaningful creative variety?
Different ways to use the product, real-life moments, a wide range of people, different visual styles and behind-the-scenes stories. Swapping a burger for a cheeseburger in the same photo, or using a slightly closer version of the same shot, does not count.
Can I just use ChatGPT or Gemini to make more ad variations?
You can make them quickly, but the article calls this fast but blind: generic models work without knowledge of the market or competitors, which produces similar content with little brand personality. Grounding generation in performance, competitor and audience data is what adds strategic depth.
What is generative grounded creative?
It is creative produced with specialised AI models connected to historical performance data, competitor information and real audience insights. It decodes competitor concepts, maps personas to the emotional needs driving them and scores generated content to predict return before launch.
Related reading
- How do you solve Meta's creative differentiation problem at scale?
- What is creative clickability, and can you predict it before spend?
- Why do ads in one category all look the same?
- Creative brief template: persona, tension, angle and story
- How to research competitor ads in the Meta Ad Library: a step-by-step guide
