Forwarded this email? Subscribe here for more.

Presented by

This week on Signal & Noise, we’re looking at one question from two angles: when AI makes everything look convincing, how do marketers know what works? One conversation is about the ads themselves, and the other is about the research behind the decisions.

First, “The Attention Ledger: Matt Krepsik and Vas Bakopoulos on Benchmarking Creative in a Crowded Market.” Matt, CEO of MediaRadar, and Vas, who leads industry research at MMA Global, join Rio and Brett House to unpack a benchmark covering 10 industries, six channels, and roughly two-thirds of U.S. ad spend. Ads are packing in more messages every year, whole categories are converging on the same formulas, and brand messaging is making a comeback over direct response. They also cover why viewers decide whether to keep scrolling in under a second and why measurement so often credits sales to people who were already going to buy. Vas sums up the conclusion this way: AI raises the floor, but human choice has to raise the ceiling.

Then, Lucas Longacre goes live from TMRE 2026 in Denver with Tracy Adams, Senior Director of Research and Insights at the ARF, and Keith Smith, Managing Director of the Marketing Science Institute, in “Decision-Ready or Not? Tracy Adams on How to Validate Synthetic Data.” In an ARF survey, about 92% of marketers say they’re confident in AI outputs, but only about a third validate them. Tracy and Keith walk through the ARF/MSI framework, which scales the level of validation to the stakes of the decision. They explain why synthetic samples tend to lose the outliers marketers most need to find and why critical thinking may be the most important skill in research right now. Watch or listen on your platform of choice!

What does it take to break through when every ad in your category is saying more and sounding the same?

On this episode of Signal & Noise, Rio Longacre and Brett House are joined by Matt Krepsik, CEO of MediaRadar, and Vas Bakopoulos, who leads industry research at MMA Global. Together they dig into a new creative benchmark covering 10 industries, six channels, 20 quarters, and roughly two-thirds of U.S. ad spend.

The data shows that ads are getting denser. Brand, price, product, and direct response messages are increasingly packed into a single spot, and that message complexity has climbed every year for about five years. Meanwhile, whole categories have settled into the same formulas. With AI making it possible to produce creative at enormous volume, the brands that stand out will be the ones that know what works and have the nerve to break convention.

The conversation covers what the benchmark reveals, where measurement goes wrong, and why AI makes human judgment more valuable.

Key takeaways:

  • Ads are saying more than ever. Message complexity in TV, video, print, and radio has risen year over year. QSRs rank highest at about 1.98, while automotive has simplified to about 1.07 as it shifts from features to lifestyle.

  • Brand is back. Brand messaging rose 6.1 points while direct response fell 6.4. MMA research shows brand effects keep paying off long after a campaign ends, with multipliers of about 2x to 5x and banking approaching 6x.

  • You have about one second. A neuroscience study of roughly 900 people found scroll decisions happen in half a second to a second, which is why Vas argues every brand needs a “first-second strategy.”

  • Defying gravity pays. Brands that break from their category’s conventions tend to stay relevant longer, and Matt argues that quality creative can offset a smaller share of voice.

  • Attribution often credits the already-converted. Matt’s club-promoter analogy shows how measurement, especially in retail media, often takes credit for people who were going to buy anyway. Only about 2% of the U.S. population is actively shopping for a car at any given time.

  • AI raises the floor. Humans raise the ceiling. AI brings scale and processing power. Marketers still have to supply context, taste, and judgment. In Matt’s words: “Intelligence should sharpen your judgment, but not replace your courage.”

NEW PODCAST THIS WEEK:

Featuring Tracy Adams, Senior Director of Research and Insights at the Advertising Research Foundation (ARF), and Keith Smith, Managing Director of the Marketing Science Institute (MSI)

How do you know whether synthetic data is good enough to bet a decision on, especially when the output sounds so convincing?

Recorded live at TMRE 2026 in Denver, this episode features Lucas Longacre in conversation with Tracy Adams and Keith Smith about one of the most urgent questions in insights today. In an ARF survey, about 36% of marketers say they use AI for synthetic data, and about 92% report confidence in AI outputs. Yet only about a third take the time to validate them. Tracy and Keith walk through the ARF/MSI framework, which matches the rigor of validation to the stakes of the decision. Light checks are fine for early exploration, but launch, segmentation, and forecasting decisions call for multiple lines of evidence.

The conversation explores why synthetic samples tend to collapse toward the average and lose the outliers and niche segments marketers most need to find. It also covers how AI’s tendency to agree with users creates new risks for researchers, and why vendor claims in this “Wild West” market deserve scrutiny. As Tracy puts it, “AI output is so plausible and convincing it sounds so good you immediately say okay sure that makes sense.” Tracy and Keith also argue that the bigger challenge may be organizational, because many teams are handed a chatbot with no framework for using it, and the productivity gains from new technology have historically taken about a decade to arrive.

Want to see more great no-BS content?
Visit https://www.signalandnoise.ai

If you found this Signal & Noise content insightful, forward it to a colleague!