Advertising research firm MediaScience has introduced a new AI-driven approach that could fundamentally change how marketers test and optimize ad creative.

The company announced Creative Twin™, a new methodology that uses artificial intelligence to recreate advertisements with near-perfect accuracy and then test the impact of individual creative elements within them. The breakthrough will be formally presented at the Audience x Science Conference hosted by the Advertising Research Foundation.

Developed using proprietary technology within MediaPET.ai—a spinoff initiative from MediaScience—the system enables researchers to generate AI replicas of advertisements that audiences reportedly cannot distinguish from the originals.

For marketers increasingly focused on creative effectiveness and personalization, the technology promises something long considered nearly impossible: isolating and measuring the precise impact of individual creative decisions within an advertisement.

From Creative Guesswork to Scientific Measurement

Creative optimization has traditionally relied on A/B testing or multiple ad versions, both of which can be expensive and time-consuming. Testing individual variables—such as a celebrity endorsement, visual style, or casting decision—often requires entirely new ad production.

Creative Twin aims to eliminate that barrier.

Once an advertisement is recreated as an AI-generated replica, researchers can systematically modify individual elements within the ad—such as talent, visuals, messaging, or background details—while keeping the rest of the creative identical.

The result is a controlled testing environment where marketers can measure exactly how each element influences audience response.

“This represents a fundamental shift in how advertising creative can be evaluated and optimized,” said Duane Varan. “For the first time, researchers can isolate and measure the contribution of individual creative elements within an advertisement.”

AI Replicas That Audiences Can’t Distinguish

To validate the methodology, MediaScience conducted controlled testing with 812 U.S. respondents in collaboration with the Ehrenberg-Bass Institute, one of the world’s most respected academic marketing research institutions.

Participants were shown both original advertisements and AI-generated replicas created using the Creative Twin technology.

According to the study, respondents were unable to distinguish between the original ads and the AI-generated versions, confirming that the replicas maintained the full production quality and realism of the original creative.

That fidelity is critical, because it allows researchers to modify ad components without introducing unintended differences that could skew results.

Testing the True Impact of Creative Elements

With the AI-generated “twin” in place, advertisers can test a wide range of creative variables.

For example, marketers can explore:

• The impact of different celebrities or influencers
• Alternative casting choices
• Visual design variations
• Messaging changes
• Background imagery or brand elements

Because each version of the ad remains visually identical except for the specific variable being tested, researchers can determine the incremental impact of each creative choice.

This level of control has historically been difficult to achieve without producing multiple expensive ad variations.

Personalized Advertising Without New Production

Beyond research applications, the technology could also reshape how brands personalize advertising at scale.

Addressable advertising—where ad creative is tailored to specific audience segments—often requires producing multiple versions of the same ad. Creative Twin enables marketers to generate these variations digitally without reshooting the ad.

One example highlighted by MediaScience involved a shampoo commercial.

In the original advertisement, the featured model had straight hair. Using Creative Twin, researchers generated an AI-modified version of the same ad where the model appeared with curly hair.

When shown to audiences with curly hair, the modified ad delivered significantly stronger results across several key marketing metrics, including:

• Brand recognition
• Brand attitude
• Brand preference (a proxy for purchase intent)

The results suggest that subtle creative adjustments—when matched to the right audience—can meaningfully improve advertising effectiveness.

Expanding Creative Testing Across Categories

The methodology also opens the door to highly targeted creative variations across different industries.

MediaScience outlined several potential applications:

• Pet food advertising: A commercial featuring a Labrador could be digitally adapted to show a poodle or French bulldog for owners of those breeds.
• Celebrity endorsements: Marketers could test whether replacing a celebrity with another talent—or removing the celebrity altogether—changes ad impact.
• Product personalization: Visuals and messaging could be adapted to align with specific audience demographics or preferences.

In each case, the modified ad retains the same production quality as the original version, ensuring that the test focuses solely on the element being evaluated.

Measuring the Economic Value of Creative Decisions

One of the most intriguing aspects of Creative Twin is its potential to quantify the financial impact of creative decisions.

Advertising production often involves major investments in talent, filming, and design, but marketers rarely have precise data on which elements deliver the greatest return.

Creative Twin allows researchers to test these elements individually, helping brands determine:

• Whether a celebrity endorsement justifies its cost
• Which visuals resonate most strongly with target audiences
• How casting or styling decisions influence brand perception
• Which creative components drive purchase intent

For marketing teams under pressure to prove ROI on advertising budgets, that level of insight could be transformative.

A New Era of AI-Driven Advertising Research

The introduction of Creative Twin reflects a broader trend toward AI-driven experimentation in marketing and advertising.

As generative AI tools become more sophisticated, marketers are gaining the ability to simulate and test creative variations without the traditional production costs associated with advertising development.

For MediaScience, the technology represents the next phase in advertising research—moving from observation to controlled experimentation.

If widely adopted, AI-powered creative replication could reshape how brands design, test, and personalize advertising campaigns in the years ahead.

Get in touch with our MarTech Experts.