Good Ads Aren't Luck. Now There's Proof.

In advertising, creative has always been the second-class citizen. Budgets were debated for hours. Audiences were segmented in detail. Bids were optimized daily. But the actual ad? "Put a nice visual and a clear CTA, let's go."

The result is visible across the industry: billions invested in optimizing distribution, while the content of the ads was treated as a variable you couldn't really quantify. It was a matter of taste, intuition, of "the creative team feels what works." And yes, intuition matters. But intuition without data is, at best, unvalidated experience.

A recent study changes that equation. Kroger, in partnership with Vidmob and MMA Global, demonstrated that predictive creative scoring can forecast e-commerce conversion with 81% accuracy. Not after the fact. Before launch.

What the Kroger-Vidmob data shows

A recent article on Marketing Dive details the study conducted by Kroger with Vidmob (a creative analytics platform) and MMA Global (an industry trade association). The team analyzed over 1,900 video and image assets from Kroger campaigns on Meta and Google DV360, evaluating variables including messaging, narrative structure, branding, and human interaction in each creative.

The results aren't subtle:

Creative assets aligned with predictive recommendations generated a 4x improvement in average conversion rates. Even more relevant: by simply reallocating media budget toward higher-scoring assets, campaigns can generate more than 2x conversions without additional spend.

Alex Collmer, Vidmob's founder, summarized the historical problem rather directly: "Historically, people made ads and then spent money blindly behind them, then looked to optimize after seeing performance." In other words, the industry operated for decades on the principle of "launch, see what works, adjust along the way."

The difference now? AI evaluates the creative before a single dollar goes to distribution. And it does so with a consistency no human team can replicate at scale.

Why creative has been treated as a minor variable

The reason is straightforward: we didn't have the tools to measure it properly. You could see CTR, but not why someone clicked. You could compare two ad variants through A/B testing, but that required large budgets and time.

Meanwhile, advertising platforms evolved massively on the distribution side. Google Ads, Meta Ads, Amazon Ads with their new AI Agent have poured billions into targeting algorithms and bidding systems. Creative remained the team's responsibility, but without a clear feedback loop connecting what you made to what actually sold.

What the Kroger research shows is that when you give AI enough data about creative performance, it can learn what works. Not at the level of "blue backgrounds outperform red backgrounds," but at the level of narrative structure, message type, and brand presentation.

That transforms creative from a subjective variable into a measurable one. Not just retroactively, but predictively. You don't guess what works. You verify.

What changes in advertising practice

Until now, the standard process looked roughly like this: brief, brainstorm, production, launch, analysis, optimization. Iteration came after you spent the budget. And if the first version didn't perform, that was the cost of "experimentation."

With predictive scoring, a new step appears between production and launch: evaluation. Before sending the ad live, you run it through a model that tells you "this has a 75% chance of converting, this one has 30%." You don't eliminate uncertainty entirely, but you reduce it significantly.

What this means in practice:

Content teams get objective feedback on creative, not just subjective opinions from the room. Budgets aren't wasted on market-testing variants that could have been filtered from the start. And creative assets that performed well become a dataset the AI learns patterns from.

Collmer observed that content creators are "super hungry for data because their paycheck is based on how their content performs." When you have a predictive feedback loop, you're no longer working on pure intuition. You have a concrete starting point.

This doesn't mean AI replaces creative judgment. It means it complements it. Just as ROAS doesn't tell the whole story about performance, a predictive score doesn't tell you how to make a good ad. It tells you whether what you made has potential, before you spend on distribution.

What you can apply now

You don't need to be Kroger to benefit from the principle behind this study. A few concrete directions:

Build an internal creative evaluation framework. Even without a sophisticated predictive model, you can systematically analyze: which message types work for which audiences? Which visual formats performed best? Collect performance data on each creative and look for patterns. Most teams don't do this because the data sits in separate platforms and nobody centralizes it.

Connect your creative team to performance data. The team making creative assets needs to see not just CTR, but conversion rates, cost per acquisition, LTV values. When a creative professional knows that variant A drove 40% more conversions than variant B, the next project starts from a higher baseline. Without this feedback, every project begins from zero.

Test scoring platforms. Tools like Vidmob, CreativeX, and Pencil offer accessible versions of predictive scoring. Not with Kroger's volume precision, but enough to filter weak variants before launch. Even basic scoring across a few hundred assets can reveal patterns the team intuitively sensed but couldn't articulate.

Invest more in the brief. A clear brief with precise messaging and audience objectives becomes crucial when you can measure creative impact. If the brief used to be "make something nice," now it's "make something that scores above our conversion threshold based on the data we have."

The difference between a campaign that converts and one that burns budget often comes down to the creative you put in front of the audience. That's not an opinion. It's what the data shows. And if AI gets the ad wrong and nobody's accountable, the fix isn't less AI. It's better feedback on the creative behind it.

The Kroger study isn't an isolated academic experiment. It's a confirmation of a trend visible across every advertising platform: creative is becoming measurable, predictable, and optimizable at the same scale as media buying. Teams that ignore this aren't just missing conversions. They're losing a competitive advantage that their competition is already adopting.

The tools exist. The data exists. What remains is the decision to use them.