By Christina Bell  | Head of Strategy , Webtopia

Most founders have never heard of creative Entity IDs. Meta Andromeda is an AI system that reads, categorises, and matches ad creative to the right people. Once you understand how it works and why creative diversity matters, the case for rethinking your paid media approach becomes obvious. Here is how it operates and what it means for your business in 2026.

There is something most DTC founders have never been told about how Meta ads actually work, and it sits at the heart of why some brands keep finding new customers while others keep reaching the same ones.

When Meta's system reads one of your ads, it does not just register that it is an ad for your brand. It categorises the creative. It assigns what Meta calls an Entity ID, based on the visual and conceptual signals it detects in the asset. Ads that look sufficiently similar to each other receive the same Entity ID. And ads with the same Entity ID tend to reach the same audience.

This is the core mechanism behind Meta Andromeda, the platform's AI content understanding system. It is also the reason why producing more of the same creative does not drive growth. You are not widening your reach. You are re-confirming the same signal to the same people.

We work across a broad range of DTC brands in fashion, beauty, wellness, home, and jewellery. The pattern we see in underperforming accounts is remarkably consistent: a small pool of visually similar assets running across multiple ad sets, with the expectation that audience targeting will do the heavy lifting. It will not. Not anymore.

What Is Meta Andromeda?

Meta Andromeda is an AI system that reads, categorises, and matches ad creative to the right people. Unlike the old model of paid media, where your targeting inputs were the primary mechanism for reaching a specific audience, Andromeda works by analysing the creative itself: the visual elements, format, copy signals, and conceptual framing of each ad. From that analysis, it determines who is most likely to engage and convert.

Meta has reported an eight percent increase in overall ad quality directly attributable to Andromeda's deployment. That improvement does not come from better audience data or smarter bidding. It comes from the system being better at reading what an ad is about and matching it to people who will respond to it.

The practical shift this creates is significant. Audience targeting settings still exist in Ads Manager, but their function has changed. They are now treated more as broad guidance than as precise instructions. The system reserves the right to expand beyond them when it believes doing so will improve performance. What actually directs delivery is the creative you put in front of it.

The old lever was audience targeting. The new lever is creative quality and diversity. Andromeda is why that shift is structural, not a trend.

How Entity IDs Affect Your Reach

The Entity ID concept is worth spending time on, because it changes how you should think about creative production at a practical level.

When Andromeda processes your creative, it assigns an identifier based on what it detects. If your current ad looks and feels like previous ads in your account, it receives the same identifier. That means it gets matched to the same audience. If your creative is genuinely distinct in concept, format, or visual approach, it receives a unique identifier, and Andromeda can use it to find audiences your existing campaigns have not yet reached.

This is why running three slight variations of the same product image across multiple ad sets limits your growth. It is not a targeting problem or a budget problem. It is a signal problem. The system has nothing new to work with, so it keeps reaching the same people.

Genuinely different creative generates genuinely different signals. A video with a founder speaking directly to camera reads differently to a user generated style review from a customer. A partnership ad from a creator your audience follows generates a different signal again. Each distinct concept opens up reach that recycled assets cannot access.

For a deeper look at what actually generates a new Entity ID in practice, and how to structure your creative for genuine visual divergence, read our Entity ID diversification guide.

The Wider AI System Behind Meta Ad Delivery

Meta Andromeda is one part of a broader suite of AI systems that influence how your ads perform. Understanding the full picture helps explain why creative quality has become so important across the platform, not just in one placement or campaign type.

GEM, the Generalised Engagement Model, focuses on predicting which content will drive meaningful engagement signals. It has produced a five percent increase in conversions on Reels. Lattice, Meta's multi-task learning system, is designed to simultaneously optimise for multiple objectives and has driven a six percent lift in ad conversions. Sequence Learning, which models the patterns in which people are exposed to ads over time, has added a further three percent uplift in conversions.

Each of these systems reads from the same creative input you provide. A brand running a narrow, repetitive creative set is giving all of these systems very little to work with. A brand running a genuinely varied programme of creative across formats, concepts, and angles is giving them substantially more signal to optimise against.

Meta's wider AI direction, including its 2026 acquisition of Manus AI and what it means for how campaigns are optimised, is covered in our breakdown of Meta's Manus AI integration.

The compounding effect of this is real. Independent research from Kantar, cited in Meta's own platform guidance, attributes 49 percent of brand impact across campaigns to creative quality. Nearly half of what determines whether a campaign moves the needle at a brand level comes down to the creative itself: not audience selection, not bidding strategy, not budget. This is consistent with what we see when we audit accounts that have plateaued despite healthy spend levels.

What the Performance Data Shows

Beyond the brand impact figure, the short-term performance numbers are equally clear. Meta's own data shows that ad sets including a genuinely diverse format mix, meaning static images alongside standard video and vertical video with audio, see an average 7.3 percent lower cost per acquisition compared to ad sets running a single format. Across a meaningful media budget, that difference is not marginal.

Meta's data also shows that 15 million ads were created using its generative AI tools in a single month, across four million advertisers actively using these features. The scale at which AI-assisted creative production is now being used suggests that the brands still treating creative as a manual, infrequent exercise are falling behind a very fast-moving standard.

For DTC brands at the five to thirty million revenue stage, where CAC pressure is often the central operational tension, redirecting creative investment towards greater volume and genuine diversity is one of the most direct responses available. The instinct to narrow audience targeting and tighten spend is understandable. But it works against the platform logic. Restricting the system's ability to explore limits learning, and limits performance.

New Creative Tools Now Available in Meta Ads Manager

Meta has added several features to Ads Manager that give advertisers direct visibility into the creative signals their accounts are generating. These tools exist precisely because creative diversity is now a performance variable the platform expects advertisers to manage actively.

Creative fatigue reporting shows when an audience has seen the same ad enough times that costs are rising and performance is declining. This was previously difficult to track clearly. It is now a visible, measurable signal that should be part of any regular account review.

Creative similarity scoring shows, at the ad set level, how visually alike your active ads are to each other. High similarity scores are a reliable indicator that your Entity ID diversity is low, meaning Andromeda is likely struggling to access new audiences from your current creative pool.

The creative testing tool, available at the campaign level, allows up to five creative variants to be tested simultaneously using a dedicated portion of the campaign budget. The key advantage is that learnings are retained within the campaign once the test concludes, meaning performance compounds rather than resetting. It is not currently compatible with Advantage Plus Shopping campaigns, but for standard campaign structures it represents a significant improvement over previous testing workflows.

Ads Manager also now includes theme insights across 150 creative categories and ad format insights that show which formats are driving the strongest results for your specific account. These are not vanity metrics. They are the data layer that should be informing your creative planning cycle.

What Good Creative Strategy Looks Like in 2026

At Webtopia, our three priorities for 2026 are Creative Planning, AI Adoption and Testing, and Strategy and Analysis. Each is a direct response to how the platform now works, and all three are anchored in a five-stage creative process we have developed and refined across the DTC brands we manage.

The first stage is research. Before any creative is produced, we look at who is actually buying the brand's products, what is generating engagement in the category, and where the creative gap is relative to competitors. This is not a one-off exercise. It feeds into every subsequent decision and is revisited continuously as the account generates new data.

The second stage is planning. This is where the core message and angle for each creative concept is defined. Not the execution, but the strategic intent: what this creative needs to communicate, to whom, and why it will be distinct enough to generate a new signal in the system. The goal is a diverse range of planned concepts before a single asset is produced.

The third stage is development, where ideas become assets. This is where format decisions matter most. Building vertical video with audio in from the start rather than as an afterthought. Collecting multiple hook variants from creators rather than a single piece of content. Producing partnership ads that carry audience trust from the creator's platform to the brand. These are not optional additions to the creative mix. They are structural requirements for a programme that generates meaningful Entity ID diversity.

Partnership ads are one of the most efficient ways to generate new Entity IDs at scale, because creator content introduces structurally distinct visual language by default. Read our complete guide to partnership ads on Meta for how to source creators, structure approvals, and integrate them into existing campaigns.

The fourth stage is testing, using the tools now available in Ads Manager to understand what is actually working and why. Not just which ad had the best return on ad spend, but which creative concept is reaching new audiences, which format is driving the lowest cost per acquisition, and which hooks are generating the strongest early engagement signals.

The fifth stage is iteration. Using those learnings to improve the next round of creative in a structured, systematic way. This is the stage most brands underinvest in. Iteration is what turns a good creative result into a compounding programme that gets stronger over time.

Questions Worth Asking Your Agency

If you work with a paid media agency, a small number of direct questions will quickly reveal whether they have adapted their approach for how Meta now works.

Ask how many genuinely distinct creative concepts are active in your account at any given time, and how different they are from each other. The number of ads is less important than the actual conceptual and format diversity. If the account is running predominantly similar static images, the creative pool is too narrow for Andromeda to work effectively.

Ask whether they are monitoring creative fatigue and similarity scores in Ads Manager. If these are not part of the regular reporting conversation, they are missing signals that directly affect performance.

Ask how they approach creative testing and what happens to the results. If testing is a quarterly exercise rather than a continuous programme embedded in the campaign structure, the account is leaving a significant amount of learning on the table.

Ask what your format mix looks like across active ad sets, and what the plan is to get to a genuinely diverse creative set if it is not there yet. The 7.3 percent CPA reduction associated with format diversity is a clear return on that investment.

None of this is a criticism of agencies as a model. It is simply that the platform has changed substantially over the last two years, and the approach that drove strong results in 2022 and 2023 is not the same approach that drives strong results now. The question is whether the people running your paid media have updated their playbook accordingly.

The Bottom Line on Meta Andromeda

What Andromeda has done is make the creative layer the primary commercial lever in Meta advertising. Not targeting sophistication. Not bidding strategy. Not budget alone.

For founders, the right response is not to try to reverse-engineer the algorithm. It is to build a creative programme that is genuinely varied, genuinely tested, and genuinely integrated with your acquisition strategy as an ongoing function rather than a periodic exercise.

The brands performing best on Meta in 2026 are not all spending more than their competitors. Many of them are spending less, but feeding the algorithm a richer, more diverse creative signal. Andromeda rewards that. And the gap between the brands that understand this and those that do not is growing wider every month.

Work with Webtopia

Webtopia works with DTC brands across fashion, wellness, beauty, home, and jewellery to build creative-first Meta strategies that work with the platform's AI systems rather than against them. Our three priorities for 2026 are Creative Planning, AI Adoption and Testing, and Strategy and Analysis: the same three areas where the platform's performance gap is widest. If you want to understand where your current creative programme stands, get in touch with our team.