Jean-Noël Zeh
Chief Data Officer, Ipsos bva — 25 years in scientific direction and data science applied to research
I design data fusion, agent-based simulation and synthetic data models that turn raw data into reliable audience measurement and marketing decisions. Creator of the Mobimétrie methodology, the reference audience measurement for out-of-home advertising in France.
A scientific profile at the service of research
25 years dedicated to scientific direction and data science applied to market research.
An engineering graduate of École Centrale de Marseille (1986), I built my career around scientific direction and data science applied to market research. I started as a software manager building data processing and analytical solutions, before joining CSA Institute as scientific director, then Ipsos in 2018.
Since 2018, I have led Ipsos bva's Data Management Center (DMC) in France: a multidisciplinary team of data scientists, statisticians and modeling experts who design data-fusion methodologies, activation-ready segmentation and mobility modeling for major accounts and for out-of-home audience measurement.
The projects presented on this site — Mobimétrie in particular — are carried out as part of my role at Ipsos bva; this site presents my expertise and professional background.
- ✓Chief Data Officer, Ipsos bva France — leading the Data Management Center (DMC) since 2018
- ✓Engineer, École Centrale de Marseille (1986)
- ✓Scientific direction at CSA Institute before joining Ipsos
- ✓Multidisciplinary team: data scientists, cloud computing experts, graph theory, statisticians, mobility experts
- ✓International deployments in the UK and Norway; prospecting in Japan, Ireland, Switzerland and Germany
Fields of intervention
From methodological design to industrial deployment, expertise spanning the full data chain applied to market research.
Data fusion & Machine Learning
Combining survey data, passive measurement, open data and third-party data to build robust predictive models (driver analysis, segmentation).
Agent-based modeling (ABM)
Simulating synthetic populations of several million individuals reproducing real behaviors: trips, purchases, media exposure.
Synthetic data
Generating synthetic populations and trajectories calibrated on INSEE data to overcome sample-size limits and accelerate research.
AI-calibrated market sizing
Market-sizing models combining surveys, open data and algorithmic calibration, with explicit uncertainty quantification.
OOH / DOOH audience measurement
Designing the Mobimétrie methodology: estimating traffic and advertising exposure street by street, road segment by road segment, outdoor and indoor.
AI-driven segmentation & activation
From data collection to activation: synthetic data, image analysis, persona bots, ideation, CRM and digital activation.
Cloud & large-scale computing
Simulation pipelines on Google BigQuery and Vertex AI to process billions of trips and records.
Research production automation
Tools that automate deliverable generation (PowerPoint reporting) to speed up client delivery.
Training & knowledge transfer
Designing and delivering applied statistics training (data analysis, data integration) for research teams.
Mobimétrie
The reference audience measurement for out-of-home advertising in France.
Estimating advertising exposure, street by street
Designed and scientifically led within Ipsos bva's Data Management Center, Mobimétrie estimates, for each of France's 600,000+ billboards, the number of people exposed — by mode of transport, time slot and socio-demographic profile. The methodology combines agent-based modeling (ABM), data fusion and calibration against real-world data (Cerema, telecom operators, INSEE surveys).
Indoor extension 2025 – 2027
Since 2025, the methodology has been extended to indoor environments. Two proofs of concept validated the approach with less than 1% gap between targets and results: with Doohyoulike across 2,005 sites (convenience stores, gyms, campuses), and with ECN across 210 office buildings and 308 screens.
- Phase 1 · before summer 2026Simple environments — convenience stores, fitness, campuses, office buildings
- Phase 2 · H2 2026Complex environments — shopping malls
- Phase 3 · early 2027Transport — train stations, subway, airports
Sources: Mobimétrie / Ipsos bva op-ed, The Media Leader (April 8, 2026); Adwanted Research Awards 2026 submission (result pending).
Seminars, conferences and awards
Public speaking and professional distinctions around data science applied to market research.
Grand Prix Syntec des Études
Award for the Mobimétrie project — "Building a trip database for billboard audience measurement".
Ipsos client seminar
Presentation of the Mobimétrie methodology to advertisers and media agencies.
ESOMAR Congress
Talk: "Mobimétrie: creating the French OOH audience measurement currency" — data fusion, agent-based modeling and synthetic data.
Colloque des Sondages
"Fine-grained modeling of outdoor audiences" — academic presentation of the agent-based simulation methodology applied to audience measurement.
Ipsos client webinar — AI & Segmentation
"How AI boosts segmentation: from data collection to activation" — 252 registrants, 95 attendees.
IREP After-work
Panel discussion "Synthetic data and AI: what progress for market research?", chaired by the President of IREP's Scientific Council.
IREP Forum × UDECAM — "Commerce Next Generation"
"Unifying to measure better" — the first granular, standardized indoor audience measurement in France, serving next-generation commerce.
Adwanted Research Awards — submission
Entry submitted (Media & Audience category) for the Mobimétrie / Ipsos bva indoor audience measurement model; result pending.
In the media
Press coverage of the Mobimétrie / Ipsos bva work.
An approach grounded in the scientific literature
The methods used draw on established academic references in modeling, statistics and calibration.
- Agent-based modeling (ABM)Bonabeau, E. (2002). Agent-based modeling: Methods and techniques for simulating human systems. PNAS, 99(suppl 3), 7280–7287.
- Iterative Proportional Fitting (IPF)Deming, W.E. & Stephan, F.F. (1940). On a Least Squares Adjustment of a Sampled Frequency Table When the Expected Marginal Totals are Known. Annals of Mathematical Statistics, 11(4), 427–444.
- Synthetic population generationMüller, K. & Axhausen, K.W. (2011). Population synthesis for microsimulation: State of the art. Presented at the 90th Annual Meeting of the Transportation Research Board.
- Rank-correlation validationSpearman, C. (1904). The proof and measurement of association between two things. American Journal of Psychology, 15(1), 72–101.
- Constrained optimization (SLSQP)Kraft, D. (1988). A software package for sequential quadratic programming. DFVLR-FB 88-28, Institute for Flight Mechanics, Cologne.
- Independent measurement certificationMethodological audits by CESP (Centre d'Étude des Supports de Publicité), the French joint-industry body certifying audience research — cesp.org
How I can help
Consulting, modeling and training around AI and data science applied to market research.
Data & AI strategy consulting for research
Methodological scoping of projects combining surveys, passive data and artificial intelligence.
Agent-based modeling & synthetic data
Building calibrated virtual populations to simulate behaviors, markets and audiences.
AI-calibrated market sizing
Market-size estimation with explicit uncertainty quantification.
Audience measurement & catchment modeling
Traffic, exposure and catchment-area models for out-of-home media and points of sale.
Research production automation
Reporting and deliverable automation tools to speed up client delivery.
Training & knowledge transfer
Tailored applied-statistics and data-science training for research teams.
Let's talk about your project
For any question about my work, a speaking engagement, training or a collaboration, get in touch directly.
jnoel.zeh@gmail.com