Bleustein-Blanchet Brought Two Things Back from New York

The Two Suitcases

In 1938, twelve years after founding Publicis above a butcher’s shop in the Faubourg-Montmartre, Marcel Bleustein crossed the Atlantic for what would become a defining encounter: in New York, he met George Gallup, the pioneer of scientific opinion polling. Bleustein had been making trips to the United States with growing frequency, drawn there by something French réclame did not yet practice at scale: the systematic study of the consumer. Polling, segmentation, testing, scientific measurement of effect. He had been bringing those methods home for years. By 1936, back in Paris, he had built an integrated multi-media offer across radio, cinema, and press, each calibrated to specific audiences. After the war, Publicis would become the first French agency to formally integrate opinion research into advertising practice.

But the New York he kept returning to was not only Gallup. It was also a city where advertising professionals went to see. They walked through stores, watched people choose, watched them hesitate, watched them give up. They went home and made notes. The methodology had two legs: a quantitative one that produced numbers, and an observational one that produced understanding. Bleustein absorbed both, and over his career he built each of them at Publicis. He never confused one for the other.

A century later, the industry has kept the first leg, and let the second one atrophy.

What the Data Always Sees

Before going further, a clarification. This is not a piece against data. Hard data does what no other layer can do: it scales, it measures, it disconfirms hypotheses, it allows comparisons across markets and time. A modern marketing function that gave up its analytical infrastructure would be making itself blind in a particularly ridiculous way. Bleustein-Blanchet would not approve. He spent decades building precisely that infrastructure for European advertising, and he did so against an industry that, at the time, considered it unnecessary.

Hard data captures the transaction. It captures the click, the dwell, the scroll, the purchase, the return. It captures, with remarkable fidelity, what happened. And in the right hands it gets you very close to what is likely to happen next, which is the engine of every contemporary marketing operation.

What the Data Never Sees

What hard data does not capture is the social fabric in which the transaction is embedded. The gestures, the rituals, the local conventions, the imitation patterns that produce the transaction long before the consumer becomes legible to a sensor. The data sees the dot. It does not see the shape of the constellation the dot belongs to.

Consider a small example. In a bar in the lower Salento, the standard summer order is an espresso poured cold over a glass of ice, no sugar, no instructions. Pensionati, teenagers, tourists who learn quickly, all drink it the same way. Nobody can explain why, and nobody needs to. It is what one does there. The transaction shows up cleanly in the bar’s POS. The gesture, the inheritance, the social grammar that produced the gesture, do not. A multinational with ambitions in Italian coffee can know everything about each of those customers as individuals, and still know nothing about the coffee they are actually drinking. The data is not wrong. It is answering a different question.

This is what Durkheim called a fait social: a behavior that exists before the individual, orients the individual, and cannot be reduced to the sum of individual preferences.¹ You do not detect it by improving the resolution of the sensor on the individual. You detect it by looking at people together, in their environment, doing what they do. The methodology is the one Bleustein’s American counterparts were practicing in the 1930s, and that the industry, almost imperceptibly, has stopped funding.

A Complementary Knowledge, Not an Alternative

It is important to be precise about the claim, because the obvious caricature is the wrong one. I am not arguing that observation should replace data. I am arguing that the two legs of Bleustein-Blanchet’s original synthesis need to be restored to balance. Hard data answers questions about what and how much, with a precision the twentieth century could not have imagined. Observational, ethnographic, social knowledge answers questions about why and with whom, with a depth that no data infrastructure can produce on its own. The two are not in competition. They are in complement. Treating them as substitutes is the conceptual error that produces a great deal of the marketing output we are all tired of: the kind that performs well on dashboards and lands as friction in actual lives.

The European advertising industry, in its formation between Paris and New York in the first half of the twentieth century, invented the integration of these two legs. The integration has drifted. The second leg has been quietly outsourced to platform signals which capture behavior without producing understanding. The next competitive advantage lies in rebuilding the observational layer with the seriousness brought to the analytical one a century ago.

The Bridge Between Them

The operational question, of course, is the harder one. Once you accept that hard data and social knowledge are complementary rather than substitutable, how do they actually talk to each other? The space between a Tableau dashboard and an ethnographic notebook is wider than it looks, and most attempts to bridge it default to one side or the other.

The most rigorous attempt I have encountered, in two decades of working in this field, to articulate that bridge is Data Humanism, the manifesto Giorgia Lupi published in 2017 and the practice she developed at Accurat, where I worked on her team.² Her argument, in its operative form, is that data is a qualitative, imperfect, contextual artifact, and that the analyses and visualizations we build from it should preserve those qualities, not flatten them away. Data, in her formulation, is not the cold counter-language to human experience. It is one of the languages in which human experience is written, and the work of the analyst is to read it in that register.

This is a position with serious operational consequences. It means treating a dataset as a portrait, not a measurement. It means designing instruments that capture specificity, not only aggregate. It means accepting that the most informative thing a number can tell you is sometimes the texture of how it was produced, not the number itself. And it means, finally, that hard data and anthropological knowledge are not two separate methodologies waiting to be reconciled at the end of the project, but two registers of the same craft, designed in dialogue from the start.

Restoring observation does not mean abandoning data. It means making data answer better questions.

The Subject I’m Talking About

To do this work, the industry needs to be precise about whose practices it claims to be reading. The “human at the center” of contemporary marketing rhetoric is, on inspection, a very thin construct: a single, portable, optimizable subject, accessible through self-report. Four assumptions, four limitations.

The subject of social practice is plural. A node in a network before it is an autonomous unit. A person buying laundry detergent is rarely buying for themselves alone, and the decision is shaped by a household, a generation, a class trajectory.

The subject is situated. A profile is not portable from Lecce to Berlin without losing the context that defines it. The same person, displaced, behaves differently, because the surrounding practices are different.

The subject is contagious. Behaviors propagate through horizontal imitation long before they yield to vertical persuasion. Gabriel Tarde wrote a careful sociology of imitation in 1890 that contemporary platform research is, often unknowingly, rediscovering.³

The subject is opaque to itself. People do not have reliable introspective access to the reasons for their own behavior, which means everything they say in a focus group is interesting, but not for the reasons researchers tend to think.

Plural, situated, contagious, opaque. Four adjectives that operate as a diagnostic grid for reading current marketing practice. Each points to an assumption that is routinely violated, and to a cost that is rarely named.

Bleustein-Blanchet, in his American crossings, did not choose between the two methodologies he encountered in New York. He brought both home. The most useful thing this generation of practitioners can do is the same.

Postscript. I am writing this from inside the industry, not above it. The thesis of this piece is not that the advertising business has failed. It is that one of the two legs of its original method has been quietly forgotten, and that restoring it is good strategy long before it is good ethics. The next vantage point in communication will not be technological. It will be anthropological. Whoever rebuilds the observational layer first will not have to choose between data and people. They will simply remember that, once, those were the same craft.

Émile Durkheim, The Rules of Sociological Method (1895), Chapter 1: “What is a social fact?” The opening definition is still the operative one for any practitioner who wants to think about consumption beyond the level of individual preference. ↩
Giorgia Lupi, “Data Humanism: The Revolutionary Future of Data Visualization,” PRINT Magazine, January 2017. Subtitled by the author “A Visual Manifesto,” the essay articulates a small set of principles that have since been adopted, often without attribution, across the better data-visualization studios in Europe and North America. Lupi’s earlier project Dear Data (with Stefanie Posavec, 2014–2015) is the most accessible practical demonstration of the method. ↩
Gabriel Tarde, The Laws of Imitation (1890). Long overshadowed by Durkheim in the canon of sociology, Tarde has been quietly rediscovered over the past two decades, in part through the work of Bruno Latour, as a precursor to network-based theories of social diffusion. ↩