Audience measurement is enhanced by data and artificial intelligence

Image

Reading time: 5 minutes

In a world where screens and media offerings are plentiful, audience practices are increasingly scattered and reflecting their diversity is becoming more complex. In order to continue to accurately reflect what audiences are watching and listening to, media audience measurement is constantly evolving. To do so, it incorporates the very technologies that are revolutionizing usage: data, artificial intelligence, and modeling algorithms.

Hybrid measurement is part of this ongoing movement, a generation of tools combining scientific rigor and technological power to offer a more complete and accurate reading of audience behavior. With a focus on clarity and education, Médiamétrie is publishing a white paper on hybrid measurement and artificial intelligence, written by Aurélie Vanheuverzwyn, Executive Director of Data and Methods, and Julien Rosanvallon, Deputy CEO of Marketing and Customer Experience.

Julien Rosanvallon explains: 

At a time when audience measurement is undergoing rapid change, this white paper summarizes the vision of Médiamétrie and that of European experts and leaders in the media sector. At a time of accelerating transformation in the sector, this perspective and global feedback are essential for thinking about the measurements of tomorrow.
 

Aurélie Vanheuverzwyn states:

Our goal is to shed light on the challenges and opportunities offered by data and artificial intelligence—but also on the myths—for the renewal of measurement methods. While their contributions are undeniable, this work shows that they play a complementary role to panels and cannot replace them.

Combining the strength of panels with the power of data

Hybrid measurement is based on the construction of a statistical model that combines different complementary data sources. The audience measurement thus created benefits from the scientific robustness of audience panels combined with the richness of big data from operator boxes, ad server logs, and analytics tools.

Aurélie Vanheuverzwyn explains:

Médiamétrie's panels—Médiamat and Internet Global in particular—form the essential foundation of the measurement: representative of the populations studied and independent, they guarantee the statistical reliability and comparability of the results and make it possible to identify and track an individual across all screens and over time

Big data, meanwhile, records all digital consumption activities, providing detailed, real-time information on connections, frequency, and comprehensiveness, which is essential in today's constantly connected media world.

Bringing these two complementary worlds together requires specialized expertise. It involves reconciling two opposing approaches: representativeness and comprehensiveness, the science of sampling and the science of counting.

It is this modeling work, carried out within Médiamétrie's Data Science Department, that gives hybrid measurement its full power.

Artificial intelligence: a driver of innovation and a matter of trust

Today, we cannot talk about data without exploring the prospects and opportunities offered by artificial intelligence for the design and evolution of hybrid measurements. These technologies pave the way for the automation and transformation of certain tasks. AI tools can facilitate the preparation of questionnaires, data collection and analysis, and reporting.

Julien Rosanvallon says:

Generative AI tools offer interesting prospects for reporting audience results. These tools are even simpler and more intuitive, capable of compiling several queries into a single analysis.

The development of learning methods has also contributed to the evolution of synthetic data generation methods. Although these new methods still raise many questions, they could in the future be used to supplement or enrich survey or panel data.

Certain hybrid approaches, which involve creating a virtual population used as a basis for reconciling different data sources, are based on this concept of synthetic data. This is particularly the case for the Virtual ID model used in the Origin cross-media advertising measurement project in the United Kingdom.

These technologies therefore open up a wide range of uses, but require vigilance: model transparency, processing traceability, and trust in algorithms remain essential to preserving the scientific legitimacy of measurements.

The challenge now is not only to use AI, but to integrate it judiciously, in a governed, explainable approach that is faithful to the principles of independence and rigor that underpin audience measurement.

More comprehensive, agile, and efficient measurements

Hybrid measurements are not limited to methodological prowess: they redefine the way we think about audience measurement. 
They offer a truly comprehensive view, reconciling linear and digital usage, tracking content from one screen to another, and reducing the time between consumption and publication of results.

Above all, they enable greater granularity—precision—which is essential at a time when audiences are becoming increasingly fragmented: we now measure formats, contexts, and audiences that traditional methods alone could not capture.

Julien Rosanvallon adds:

Advertising models have evolved considerably, with segmented advertising targeting individuals very precisely. Audience measurement must respond to these needs with increasing precision.

Finally, hybrid measurements make measurement more efficient: by using data already collected by digital environments, the hybrid method optimizes costs; achieving the same level of precision with panels would require much more significant investment. 
It's a virtuous circle: greater precision, greater responsiveness, and better control of resources.

Médiamétrie, a pioneer in convergence

In France, Médiamétrie is providing a concrete illustration of this transition with several hybrid systems already in operation. 
• Internet Global, which combines panel data and website and app visit measurements from analytics tools, offers a unified view of the entire web audience; 
• Watch Pub, the cross-video advertising measurement tool, which combines impression volumes from data from web servers hosting advertisements (ad servers) with linear TV audiences, makes it possible to measure the deduplicated coverage of campaigns;
• and segmented TV, which integrates data from ad servers to refine TV media planning.

These innovations are based on a powerful idea: hybridization is a methodological continuity. It does not replace but complements historical panel expertise by backing it up with the power of data, within a governed and controlled framework.

On the international scene, the same logic applies in Italy, Austria, Switzerland, Spain, Canada, and Sweden, with hybrid measurements. Projects are underway in Germany and the United Kingdom. In the United States, Nielsen combines its TV panel with massive viewing data in its measurement.

Everywhere, the dynamic is the same: more integrated, more fluid measurement that is more in tune with actual usage.

I find that the international experts we interviewed share a great deal of common ground in their views.

by Aurélie Vanheuverzwyn

Scientific rigour at the heart of innovation

This hybridization movement is based on a strong conviction: data alone is not enough. 
The integration of massive data flows can only produce reliable measurements if it is supported by rigorous statistical expertise. 
This is where the real added value of research institutes lies: in their ability to calibrate, adjust, verify, and above all, guarantee the comparability and transparency of results. Studies are always audited by third-party organizations.

There are multiple challenges: 
• Ensuring the quality of the data collected, 
• Avoiding duplication and bias associated with digital environments, 
• Preserving neutrality and independence from the actors being measured, 
• Maintaining consistency across panels, platforms, and cross-media environments.

It is a discipline at the crossroads of data science and measurement ethics, with challenges that are as much technological as they are educational. 
Technology, in this case, does not replace the method: it enhances it.

A discipline in constant motion

Hybridization demonstrates the sector's ability to transform itself along with its very purpose, to integrate new sources and new approaches without compromising the scientific rigor that underpins its legitimacy.

Hybrid measurement embodies the future of audience measurement: an alliance between science and data, between human experience and algorithmic power.

And in this movement, measurement remains, more than ever, an essential benchmark—rigorous, independent, and constantly evolving with the practices of its time.

Publication management : Médiamétrie Communications Team

Editor : Laure Osmanian Molinero