What happens when your competitors can produce just as many creatives, just as fast, using the same tools you do? Today, nearly 90% of advertisers now use generative AI in their creative workflow. The result is that production timelines that used to take days have collapsed to hours. Teams that once struggled to produce 10 variants a month are now generating 60, 80, even 200. By every measure of output, the industry has never been more productive.
And yet something strange is happening: as output goes up, distinctiveness is going down. When every team feeds similar inputs into similar tools, the output converges toward a visual and tonal middle ground that nobody specifically chose, but everyone ends up producing. As Taryn Crouthers, CEO of Spcshp, put it in Marketing Dive: “A lot of the output is trending toward the median. All of the content is merging to look very, very similar.”
This article is about why volume alone is not enough, and why the teams that invest in creative excellence, in the quality of their hypotheses, the distinctiveness of their ideas, and the discipline of learning from every test, will outperform those that simply produce more. Stuart Miller, Phiture’s Creative Director, shares his perspective throughout.
Why AI-generated creatives are starting to look the same
The promise of AI-generated creative is real: more variants, faster production, lower cost per asset. According to Digital Applied’s 2026 benchmarks, AI creative produces 12% higher click-through rates on Meta and saves teams roughly 20 hours per week. Creative is now responsible for approximately 70% of campaign performance outcomes, making it the single largest factor in whether a campaign performs.
But there’s a catch here. AI learns from what already exists, and when everyone is training on the same data and using the same tools, the results start blending together. Users scrolling through their feed see ten ads that all feel like they came from the same place, and in a sense, in a sense, they did. Coegi’s 2026 marketing outlook captures the tension well: “AI is excellent at accelerating production but can also amplify mediocrity if not guided by sharp human strategy.”
Meanwhile, creative fatigue cycles are compressing. On Meta, fatigue now sets in after roughly 10 days rather than six weeks, which means teams need to refresh constantly. Producing more just to keep the machine fed is a treadmill. The teams that win aren’t the ones producing the most variants. In fact, they’re the ones whose variants are different enough from each other and from the competition to actually break through.
From the practitioner: Stuart Miller, Creative Director at Phiture
The tools that creative practitioners have access to these days are frankly incredible. GenAI represents a gigantic leap forward, both in terms of reduction in effort to get a finished asset, and also the opening up of creative avenues that were, until just a few years ago, blocked off to anyone without an obscene budget. But it’s far from perfect. Without a clear, decisive approach, the models will simply default to what they’ve been trained on, which is pretty much the entirety of the open internet. This means that it’s very easy to get a nothing-soup made up of blended averages.
Why learning from creative tests matters more than running more of them
In our piece on why testing more creatives leads to better performance, we made the case that creative testing is a volume game. More experiments mean more winners. That argument still holds, but it’s worth clarifying here that volume without learning is just noise.
The real value of testing at scale is what you learn from the pattern of wins and losses over time. When a team runs 50 experiments and five win, the important question isn’t just “which five?” It’s “what did those five have in common that the other 45 didn’t?” Was it a specific emotional framing? A visual structure? A hook format? A value proposition? The answer to that question shapes the next round of hypotheses, and the round after that, and each cycle gets sharper because the last one taught you something.
This is what separates teams that test a lot from teams that get better over time. Each experiment teaches you something, and each round of testing builds on the last. Over months, the team can develop a growing understanding of what resonates with different audiences, on different surfaces, in different contexts. That understanding is the real asset. It can’t be replicated by a competitor who simply generates more variants without paying attention to why some work and others don’t.
Without this learning layer, volume is just a production metric. You’re spending more to say the same thing in slightly different ways, and the data you generate never accumulates into anything useful.
From the practitioner: Stuart Miller
I like to think that to make learnings ‘stick’, there are 2 fundamental conditions to be met. First, a test should be designed with its hypothesis structured so that learning is at least as important as the increase in conversion. Remember that we’re shooting for long-term success, not a flashy-but-isolated metric.Secondly, there needs to be intentionality around implementing learnings. Make a review of learnings a standard checkpoint in the sprint process, and regularly revisit your testing roadmap to determine how recent learnings either support the original plan, or necessitate its amendment.A creative team that does both of these things well is going to be very effective.
How to build a creative testing process that actually improves over time
Creative excellence doesn’t exactly mean spending three weeks polishing a single asset, but it also doesn’t mean generating 200 variants and letting the algorithm sort them out. The answer sits between the two, and it requires a specific set of disciplines. We recommend thinking about the following:
Start with better hypotheses. The quality of what you test matters as much as the quantity. A variant that probes a different user motivation is worth ten variants that change a background colour. Before producing anything, the question should be: what are we trying to learn from this test, and will the result teach us something we can use across channels? If the answer is no, the test isn’t worth running.
Design for difference, not iteration. The most common mistake in high-volume creative production is producing variants that are too similar to each other. Switching out a headline or changing an image doesn’t tell you much. In fact, the strongest tests explore fundamentally different approaches to the same problem: a utilitarian framing versus an emotional one, a product-focused visual versus a lifestyle-focused one, a direct hook versus an indirect one. Each variant should be a genuine creative direction, not a cosmetic tweak.
Make every winning insight travel. A creative test result that stays inside one channel is a wasted learning. If a push notification hook drives high re-engagement, it should inform the next store page screenshot test. If an ad creative outperforms because of a specific emotional angle, that angle should appear in onboarding. At Phiture, our integrated growth approach is built around making sure these insights flow across paid, organic, CRM, and creative rather than dying in the channel where they were discovered.
Protect distinctiveness deliberately. As AI makes it easier for every competitor to produce polished creative at speed, the brands that stand out will be the ones that invest in originality. That means human creative direction sets the tone, chooses what to explore, all the while deciding what feels right for the brand, even as AI handles the production. Think of it this way: the machine can speak; the human decides what’s worth saying.
From the practitioner: Stuart Miller
The question ‘What is creative excellence?’ will get you a thousand different answers, but for me, creative excellence can be found in work that has a core component of surprise. Cultural relevance and a degree of craft are important elements, of course, but work that approaches its subject matter with curiosity can often result in outputs that are delightful and memorable.
AI can’t do that yet. That’s still human territory. Recognizing where people add value, and where machines add value—and then structuring your workflow around that— is the secret sauce.
How the role of creative teams is evolving in the age of AI
There’s a persistent fear that AI will make creative teams redundant. In reality, though, we actually see that the opposite is happening. We believe that as production becomes automated, the strategic and editorial role of the creative team becomes more important, not less.
Let’s break this down. First off, someone needs to first decide which user motivations to test. Then, somebody needs to look at last month’s results and identify the pattern that a dashboard won’t surface on its own. Thirdly, somebody also needs to recognise when the brand is drifting toward the same visual territory as every competitor and pull it back. Lastly, somebody also needs to understand what’s culturally interesting, emotionally resonant, or simply surprising enough to make a user pause their scroll.
It’s obvious AI can’t do any of that. But what it can do is handle the execution once those decisions are made: generating variants, resizing assets, producing iterations at speed. The creative team’s job is moving upstream, from making things to deciding what’s worth making and why.
From the practitioner: Stuart Miller
We touched on this in a previous article, but the creative team isn’t going anywhere. In some ways, the future remit of the creative team will be the same as it was in the years before AI: craft work that stands out and cleverly explains the value of a product to the right people at the right time.
However, the pace, the audiences, and the tools have changed, and thus where the team puts it energy has changed too. As more and more of the production phase moves to AI, we’ll likely see a stronger emphasis on the skill of creative strategy, as opposed to deep expertise in one particular discipline. That’s not to say those roles will disappear completely, and I think there is always going to be room for top-notch art directors and writers, but generally I think we’ll see the lines blur across the roles in a team. But I do foresee a growing expectation that creative teams’ primary focus will be guiding the work towards cultural relevance, while continually outsourcing more and more to AI.
AI helps us scale the work in order to check the volume box, but the team is
What’s next
The industry is at a turning point. AI has removed the production bottleneck, and every team now has access to the same tools and the same speed. The differentiator going forward won’t be who produces the most. It will be who learns the fastest from what they produce, who maintains the most distinctive creative voice, and who builds systems that turn every test into an insight that makes the next one sharper.
Creative excellence has always mattered. In an era where everyone can produce at volume, it matters more than ever.
At Phiture, creative excellence is a core part of how we approach growth. Our creative team works alongside ASO, paid, and CRM to make sure insights travel across channels, and tools like PressPlay help us test at the volume the math demands without sacrificing the quality of our thinking. If you’re interested in building a creative testing system that gets better over time, get in touch.
FAQ
Does AI-generated creative perform as well as human-made creative?
AI creative is closing the gap. On Meta, AI-generated ads show 12% higher click-through rates, though human-made creative still leads on conversion by about 8%. The most effective approach in 2026 is hybrid: AI handles production and variant generation, while humans guide strategy, hypotheses, and creative direction.
What is creative convergence?
Creative convergence is what happens when many teams use the same AI tools trained on similar data, producing output that starts looking and feeling the same. As more brands adopt generative AI for creative production, the risk of homogeneous, undifferentiated advertising increases.
Why does learning from tests matter more than running more tests?
Running more tests increases the chance of finding winners. But the real value comes from understanding why certain variants win and others don’t, and using that understanding to design better tests next time. Over months, this creates a growing body of knowledge about what resonates with different audiences, which is a competitive advantage that pure volume can’t replicate.
How should creative teams work differently in the age of AI?
The creative team’s role is shifting from production to strategy. Instead of spending time building assets, teams should focus on setting creative direction, designing better test hypotheses, identifying patterns across winning and losing variants, and protecting the brand’s distinctiveness. Simply put: AI can handle execution, but humans decide what’s worth executing.
What is creative excellence in mobile growth?
Creative excellence in mobile growth means producing creative that is not only high-performing but also distinctive, hypothesis-driven, and designed to generate insights that build on each other over time. It’s the difference between producing more variants and producing variants that each teach you something useful about your audience.
Table of Contents
