We Can Finally Make Ads Faster Than We Can Test Them
Fifteen years ago the bottleneck was production. Now it's figuring out what works.
This is the part of the AI conversation everyone’s dancing around. Yes, you can generate video in minutes now. Yes, the tools keep improving. But if you’re making 10 variations and testing them the same way you tested one hero spot five years ago, you haven’t solved anything. You’ve just moved the problem downstream.
This exact pattern played out with programmatic buying. Suddenly you could buy thousands of placements across hundreds of sites in milliseconds. Brands learned that buying faster just meant wasting money more efficiently if you didn’t know what worked.
Where the Metrics Broke
Over half of advertisers now use GenAI in digital video ad creation An AI video ad is making a splash. Is it the future of advertising? : NPR, and that percentage keeps climbing. Amazon’s new creative platform generates full video concepts from product info in minutes. Meta’s tools turn still images into video with automated branding.
But here’s the problem: when Debra Aho Williamson’s team surveyed consumers about AI-generated advertising, less than half felt positively about it An AI video ad is making a splash. Is it the future of advertising? : NPR. We can make it faster. People are skeptical of what we’re making.
The real issue is we’re testing with metrics built for a different era. When you made one expensive spot and ran it for months, impressions and viewability were good enough. Lumen Research found that only 20% of digital display ads are actually viewed by users Attention Metrics in Digital Advertising 2025 - the other 80% might as well not exist. Impressions tell you an ad loaded. Viewability tells you some pixels appeared for two seconds. Neither tells you if anyone paid attention.
Attention is 7x more effective at predicting awareness than viewability How to measure attention: the metric that matters most in digital advertising. That gap matters more now because if you’re testing 10 video variations, you need to know quickly which ones work so you can produce the next batch.
What Actually Drives Results
Adelaide analyzed campaigns using attention metrics and found 41% higher brand lift and 55% stronger lower-funnel impact on average 2025 Outcomes Guide: How Attention Metrics Increase Ad Impact. An airline identified attention thresholds and saw a 95% lift in unaided awareness for placements above the threshold. A retailer got 152% more add-to-cart actions from high-attention media.
But those case studies don’t tell you the interesting part: how many variations did they test to find the high-attention creative? How fast did they get results? How did they know which specific elements drove attention so they could apply that learning to the next round?
The brands winning aren’t making the most creative. They’re the ones who figured out the testing loop. Make variations, test for attention signals, identify what works, generate the next round based on learnings. Cycle time is everything.
CTV showed the highest attention levels - double online video effectiveness and triple display How AI is Revolutionizing TV and CTV Advertising in 2025. CTV creative used to cost more to produce, which limited testing. Not anymore. You can generate CTV-quality video from product images in minutes. The constraint isn’t production. It’s knowing which variation deserves budget.
The Infrastructure Nobody Talks About
Lumen and Ebiquity demonstrated a 0.98 correlation between attentive minutes per thousand impressions and incremental profit Attention Metrics in Digital Advertising 2025. That’s why platforms are rebuilding measurement around attention instead of impressions.
But implementing attention measurement properly requires infrastructure most brands don’t have yet. You need systems that track attention signals in real-time, ways to feed that data back into creative production, processes for turning attention insights into creative briefs, and teams that can operate at the new tempo.
The brands succeeding have reorganized around the testing loop rather than the production loop. Creative and media buying aren’t separate functions with quarterly check-ins. They’re continuous optimization where creative adjusts weekly based on attention data.
Amazon’s new tool is interesting not because it generates creative quickly - plenty do that now. It’s interesting because it explains its thinking at each step, helping advertisers understand what drives creative decisions Amazon Ads launches new agentic AI creative tool | Amazon Ads. That transparency helps you write better prompts for the next round.
Michael Vamosy at Defiant LA nailed it: AI “helps you accelerate your ideas faster and further” but “a digital camera is not going to make you Ansel Adams with Photoshop” AI and Optimizing Advertising in 2025: From Hype to Hard Results - WideOrbit. The technology accelerates. It doesn’t replace judgment about what works. It just lets you test your judgment faster.
Why Creative Orgs Can’t Keep Up
Most creative teams still organize around campaigns. Develop a concept, produce assets, launch, measure results, wait a quarter, repeat. That tempo made sense when production was slow and expensive.
Now production is fast and cheap. The winning tempo is: develop concepts, generate variations, test for attention, scale what works, generate new variations, repeat weekly. Traditional creative organizations can’t operate that way. The approval processes, review cycles, stakeholder alignment - it all assumes a quarterly rhythm.
Brands are struggling with this more than the technology itself. The tools work fine. The organizations can’t keep up with what the tools enable. You end up with creative teams that can generate 50 variations and review processes designed for three options.
Meta’s early testing of new AI creative tools showed up to 22% better return on ad spend Meta unveils generative AI tools for automated video, branding, and creative ads. The improvement likely came less from creative quality and more from being able to test more variations quickly. When you test three options, you’re limited by whoever picked those three. When you test 30, you’re more likely to find something that actually works.
Where the Constraint Moves Next
Right now the bottleneck is testing infrastructure. Brands can generate creative faster than they can test it properly. That’s solvable - attention measurement tools are improving and getting faster.
The next constraint will be strategic direction. When you can test hundreds of variations weekly, “creative strategy” shifts from “what should we make” to “what should we learn.” The brands that figure out how to run creative as a learning system rather than a production system will pull ahead.
This sounds theoretical until you think about what it means practically. Instead of asking “what video ads do we need for this campaign,” the question becomes “what hypotheses do we need to test about what drives attention with this audience in this format?” The creative becomes the experiment, not the final output.
Nearly half of buy-side decision-makers expected to focus more on attention metrics in 2024 Why media mix modeling, attention metrics may take the spotlight in 2025, up from about a third the year before. That adoption curve is steep. By next year it’s table stakes. Brands still testing creative the old way will be operating blind.
The Part Nobody Wants to Hear
The industry has been solving the production problem for 20 years. Print to digital meant faster production. Programmatic meant faster buying. Now AI means faster creative development. We kept optimizing production.
Turns out production wasn’t the problem. Testing was. Knowing what works was. Learning fast enough to stay ahead of creative fatigue was.
We can make ads faster now. The question is whether we can learn faster. Most brands can’t yet. Their testing infrastructure, org structures, approval processes - they’re built for a different tempo.
The brands that figure this out first will operate at a different speed than everyone else for a while. That window closes when attention measurement becomes standard and every platform offers it. But right now there’s an edge for brands willing to reorganize around the testing loop instead of the production loop.
Everyone else is just making more ads.