The only people keener than American presidents to enter an unwinnable war are marketing thought leaders.

The latest offensive comes from Andrew Tindall at System1 and Effie Worldwide, whose Creative Dividend report previewed at Cannes Lions in June 2025 and published in full earlier this year. The dataset is substantial: 1,265 campaigns across the US, Europe, UK, and Ireland, running from 2007 to 2023. The headline findings are striking. Creative quality alone accounts for 24.2% of business results. Add media support and that rises to 60.1%. In financial services and pharma it climbs above 90%.

System1's methodology matters here, because it shapes what the report can and can't see. They use emotional response testing, second-by-second facial coding and feeling surveys, to predict long-term brand growth. The Star Rating is the output: one to five, correlating across their database with market share trajectory. It's a well-validated tool. It's also built entirely around video advertising with characters, narrative, and emotional arc. Tindall acknowledged at Cannes that 35% of people shown Grand Prix-winning ads couldn't recall the brand. His own framing: emotion without branding is feeling without structure.

Byron Sharp read all of this and wasn't moved.

His piece in Marketing Week goes after the methodology. The Effie and IPA databases are made up entirely of voluntarily submitted case studies: campaigns that agencies thought had a good story. Failed campaigns don't submit. Business effects are self-reported. Big initiatives get big budgets, better talent, more media, more in-store support, more management attention, and better awards submission writers. The dataset conflates creative quality with every other advantage a well-resourced campaign enjoys, then hands the credit to the ad.

Sharp's analogy: study only the winning teams in a season, notice they all prayed before the match, conclude prayer causes victory. The losing teams prayed too. You just weren't looking at them.

He also generated a scatter chart with an R-squared of 0.6, using random data, matching the Creative Dividend's figure of 0.61. His conclusion: correlation dressed as causation, a failure of method that should be politely ignored by any thinking marketer.

There's a broader point here that goes beyond this particular report. Research tends to reflect the assumptions of whoever commissioned it. System1 measures emotional response to video. Effie collects case studies from agencies. The questions each organisation thinks to ask are shaped by the category they operate in, the clients they serve, and the conclusions that are useful to them. That's not cynicism. It's just how incentives work. If you're commissioning research, the work is in designing for the answer you didn't expect, not the one you're already committed to believing. Confirmation is cheap. Surprise is where the value is.

Sharp names the dog that didn't bark. In the same piece, he points to the launch of ChatGPT as a real-world example of extraordinary business growth that the dataset structurally excludes. No Effie submission. No System1 Star Rating. No emotionally-coded video. Just millions of people discovering, in real time, that something genuinely new existed. The product spread on its own meaning. Word of mouth did the rest. The instrument can't see it because the instrument wasn't built to look for it.

Before the creative camp gets too comfortable, Sharp has never actually said creativity doesn't matter. In How Brands Grow he wrote that advertising's first role is to cut through, hence creativity. Ehrenberg-Bass's own research found that changing creative quality is more likely to shift sales than changing spend. Rachel Kennedy, co-founder of the Institute, said good copy drives sales and creativity should be a top priority. Sharp's argument isn't that it doesn't matter. It's that the numbers being used to prove how much are built on shaky ground.

Which brings us to the actual problem. Both sides are arguing over a definition of creativity that's too small for what they're trying to describe.

Creative people are famously good at taking offence when their particular version of the craft isn't given sufficient weight. The irony is that the people making the loudest case for creative thinking have applied almost none of it to the definition of creativity itself. They've picked one expression of the thing, the short emotional film, built a measurement infrastructure around it, and started defending the infrastructure as though it's the thing.

It isn't.

Creativity isn't a format or a medium. A product that reframes what software can do is creative. A pricing decision that removes the barrier to trial is creative. A name that makes a category feel different is creative. These aren't metaphors. They're often the most consequential creative acts a business can make, and they're invisible to instruments built to score thirty seconds of video.

The measurement defines the territory. If you can only measure emotional response to film, film becomes your definition of creativity, and everything else, strategy, product, pricing, distribution, becomes someone else's problem. The bits that compound get left out of the model.

This isn't a case against emotional advertising. It's a case for applying the same creative ambition to the question of what creativity actually is.

The most useful definition, across all of these forms: creativity is the discovery or invention of meaning.

Under that definition, ChatGPT's growth isn't an anomaly. It's the argument in its clearest form. A new shared meaning about what computers can do for people spread at a pace no media plan could match. The reach was earned because the meaning was real.

The question worth asking before the next brief isn't whether the work will score well on an emotional engagement test. It's whether you're making something meaningful enough to travel on its own.