Mathematics does not exclude “the magic”, but risks dissolving it into noise
Data can describe things we otherwise have no words for
Take a walk in the forest and look at a fern leaf. What you see is a fractal, infinite complexity organized in perfect, repeated structure. Look at a seashell and you will find phi. Nature is complex, chaotic and yet full of mathematics.
The same applies to marketing science. Byron Sharp demonstrates that brand markets at the macro level exhibit stable patterns. Buying behaviour, penetration rates, loyalty, all predictable with surprising precision across categories, countries and decades. But the greatest risk for the mathematician is to simplify reality to the point where something essential disappears.
Magic disappears in the average
What many in the industry best describe as brand love or brand magic, the deep, emotional connection between consumers and brands, disappears in aggregated data. It becomes statistical noise. It is flattened in the average of the large mass and disappears.
But noise is not the same as nothing. The models used to describe market patterns (such as the Dirichlet model) rest on one central assumption that is rarely challenged in everyday practice: that brand markets are stable and ergodic.
A system is ergodic if a snapshot of the entire population here and now tells you the same as following one individual over time. If that assumption holds, you can take static market data from many brands at one point in time and use it to linearly predict how a specific brand will develop over time, the patterns of the past are a reliable guide to the future.
Nike, when markets are not ergodic
An ergodic market assumes that the underlying preferences are stable, that the market will always move towards the same, predictable equilibrium. For Nike, reality broke this assumption on two levels:
Self-created structural shocks:
When Nike withdrew from wholesale distribution during Covid to accelerate DTC, they changed the market architecture. The empty shelf spaces were immediately taken over by new competitors, who thereby gained critical physical availability.
External macro shifts:
Cultural currents such as Guochao in China, where consumers actively choose away from international icons in favour of local brands, or an increasing saturation with large American brands in Europe, are not temporary statistical noise. They are permanent shifts.
In short, the sneaker market is not ergodic. Optimising based on historical macro KPIs in a system that has fundamentally changed direction leads to a strategic blind spot.
Magic lives in the overlooked in-between?
It is especially in the methodological in-between that the vital market dynamics live.
Cultural clusters:
Consumers are not evenly distributed. They cluster in social and cultural networks where they mirror and reinforce each other’s signals. It is these clusters that keep a brand’s social currency alive.
Temporal asymmetry:
Brand identity and emotional coverage take years to build, which makes them invisible in 12-month data.
Threshold effects:
Collapse or breakthrough in complex systems rarely comes gradually. They do not follow a linear curve. The system tips suddenly when a critical mass of consumers moves at the same time. What we have historically called brand magic is in reality emergent system dynamics that our traditional macro models are simply not designed to measure.
Two levels of the same system
Data and magic are not opposites. They simply describe different levels of a complex system. The macro formula is real. It captures something true about how markets look in a state of rest. But it is silent about the micro-mechanisms that produce the balance, and about what happens when that balance is broken.
This is an article intended to open a conversation, not close one. Comments, challenges,
and collaboration are welcome.