About
I am an independent expert and advisor in marketing measurement, causal inference, and damages quantification, with a PhD in Decision Sciences (Computational Statistics) and more than a decade of experience building, validating, and independently evaluating predictive and causal models.
My work sits at the intersection of causal inference, econometrics, and strategic decision enablement — the measurement architecture that organizations rely on to allocate investment, quantify impact, and defend analytical conclusions under scrutiny. I bring particular depth in marketing mix modeling, incrementality, attribution, A/B testing, and quasi-experimental methods, built from the ground up and made actionable for senior leadership across $500M+ in investment decisions.
More than a decade evaluating whether models deliver the reliability they claim — including more than four years auditing whether third-party marketing-measurement methods truly capture incremental impact — combined with early-career audit (Deloitte) and Sarbanes-Oxley controls (Rio Tinto Alcan) experience, is what I bring to independent validation, methodology review, and expert engagements. It is an evaluator's vantage point: assessing whether models and measurement are trustworthy, reproducible, and defensible.
Prior publications appear under Marie H. Roy and Marie-Hélène Roy.
Advisory & Expert Engagements
I work independently in two related capacities. Both rest on the same foundation: more than a decade evaluating whether models deliver the reliability they claim, including years spent auditing whether marketing-measurement methods truly capture incremental impact — grounded in consequential enterprise application rather than theory. As an independent party with no vendor stake, my role is to evaluate — not to sell a method or a tool.
Independent Measurement Advisory
For marketing and finance leaders who need a vendor-neutral assessment of whether their measurement is sound and their numbers are defensible:
- Independent method validation — whether an MMM, attribution, or incrementality method actually measures incremental impact, or only appears to.
- Measurement audit & assurance — independent review of a measurement stack for soundness, reproducibility, and defensibility.
- Vendor selection & methodology review — structured, vendor-neutral evaluation of incrementality and measurement providers.
- Measurement standards & neutral evaluation — defining what "incrementality" or "lift" means in practice, and serving as a neutral on measurement disagreements.
Expert Witness & Litigation Support
Expert consultation, report preparation, and testimony in disputes turning on marketing measurement, causal inference, statistical methodology, and damages:
- Marketing ROI, attribution, and incrementality-methodology disputes
- False advertising and performance-claim substantiation (Lanham Act, FTC, state consumer-protection)
- AdTech and measurement-vendor disputes; ad fraud and invalid-traffic assessment
- Commercial damages and lost profits involving marketing or sales data
- Expert rebuttal and methodology review of opposing statistical and causal models
Engagements. I am available for consulting and testifying engagements. Rates are provided on request.
Testimony. I have been retained as a testifying expert in commercial litigation and have authored an expert report to Rule 26(a)(2)(B) disclosure standards. Details of prior engagements are available on request.
Conflicts. Counsel are welcome to request a conflicts check before any substantive discussion; the names of the parties and, where known, related entities are sufficient.
Contact. mariehoffmann.ds@gmail.com · Full CV (PDF)
Areas of Expertise
Causal Inference & Experimentation
A/B testing design and validity, quasi-experimental methods, difference-in-differences, geo-lift testing, propensity score methods, identification strategy under real-world constraints, causal vs. correlational claim evaluation.
Marketing Measurement & Attribution
Marketing Mix Modeling (MMM), multi-touch attribution (MTA), incrementality and lift measurement, media ROI and elasticity modeling, unified measurement architecture, investment optimization.
Statistical Modeling & Econometrics
Bayesian and hierarchical models, time-series forecasting, market-response and elasticity modeling, nonparametric methods, model specification and performance evaluation, predictive and causal modeling frameworks.
Model Governance & Validation
Model risk evaluation, methodology soundness review, bias and fairness assessment, audit oversight and reproducibility standards, drift monitoring, and KPI governance — assessed from an independent evaluator's seat, with working familiarity with emerging frameworks (NIST AI RMF, EU AI Act, ISO 42001).
Career Summary
Marketing Data Science Manager
University of Phoenix
May 2026 – present
Leads the marketing mix modeling workstream end to end — development, validation, simulation, and optimization — with ownership of methodology decisions and executive alignment. Applies causal inference and experimentation (geo-lift, incrementality, A/B testing) across multi-stage conversion funnels, and builds calibrated rare-event propensity models with explicit invalid-traffic detection and temporal-drift diagnostics.
Senior Data Scientist, Measurement Science
Southern Glazer's Wine & Spirits
November 2025 – February 2026
KPI governance, measurement architecture, and analytics leadership coverage for digital product analytics across $1.5B+ quarterly influenceable revenue.
R&D Data Scientist — Personal Systems Quality
HP Inc.
January 2025 – November 2025
Large-scale predictive modeling for hardware reliability: rare-event and anomaly detection of component failures across high-volume device telemetry, with a tunable detection-model series giving explicit control of the false-positive / true-positive trade-off, and full-life telemetry trend-detection and validation tooling.
Principal Data Scientist — Marketing Measurement Lead
HP Inc.
May 2020 – December 2024
Advanced from Lead Data Scientist, Pricing & Advanced Analytics (May 2020 – January 2023) to Principal Data Scientist and worldwide Marketing Measurement Lead (January 2023 – December 2024). Led the design and execution of marketing measurement architecture integrating MMM, MTA, incrementality, and brand analytics across global business units, influencing $500M+ in investment decisions. Directed third-party model audits and established validation standards across 20+ deployed models.
Senior Data Scientist
Solsten
November 2019 – June 2020
Behavioral and psychometric segmentation linking user traits to marketing tactics; personalization impact quantified through causal modeling and adaptive experimentation.
Lead Data Scientist
Age of Learning
February 2018 – November 2019
Led the analysis of a randomized controlled efficacy study of an adaptive early-mathematics product, estimating causal learning gains with hierarchical mixed-effects models, power analysis, and IRT psychometrics; designed and deployed a live struggle-detection model with per-student prediction intervals; led methodology governance and technical training for the data-science team.
Doctoral Researcher
HEC Montréal / GERAD
2014 – 2018
Nonparametric (distribution-free) inference, prediction intervals, and high-dimensional and robust variable selection; ensemble methods for finite-mixture models.
Research Data Analyst
Tech3Lab, HEC Montréal
2016 – 2017
Longitudinal models and design of experiments for engagement measurement; finite-mixture models with mixed effects for EEG data from a heterogeneous population.
Early-Career Audit & Financial Controls
Audit & Assurance Intern (Public Sector)
Deloitte
Summer 2007
Public-sector external financial audit: financial-record and transaction testing, internal-controls evaluation, and working-paper preparation.
Sarbanes-Oxley Project Team
Rio Tinto Alcan
Summer 2006
SOX internal-controls testing over financial reporting — design and operating-effectiveness assessment, evidence collection, and control-deficiency documentation. Formal grounding in assurance and evidentiary review that informs current model-governance and validation work.
Publications & Research
Published under the author's former name, Marie H. Roy / Marie-Hélène Roy.
Peer-Reviewed
Roy, M.-H., & Larocque, D. (2020). Prediction intervals with random forests. Statistical Methods in Medical Research, 29(1), 205–229. (First published online 2019.) doi:10.1177/0962280219829885
Roy, M.-H., & Larocque, D. (2012). Robustness of random forests for regression. Journal of Nonparametric Statistics, 24(4), 993–1006. doi:10.1080/10485252.2012.715161
Owen, V. E., Roy, M.-H., Thai, K. P., Burnett, V., Jacobs, D., Keylor, E., & Baker, R. S. (2019). Detecting wheel-spinning and productive persistence in educational games. Proceedings of the Educational Data Mining (EDM) Conference.
Software & Implementation
The prediction-interval methods of Roy & Larocque (2020) are implemented in the CRAN package RFpredInterval (Alakuş, Larocque & Labbe), described in Alakuş, C., Larocque, D., & Labbe, A. (2022). RFpredInterval: An R package for prediction intervals with random forests and boosted forests. The R Journal, 14(1), 300–320. doi:10.32614/RJ-2022-012
Conference Presentations
Shapiro, S., & Roy, M.-H. (2024). Bridging the gap of MMM and MTA in a cookieless world. I-COM Global Summit, Málaga, Spain.
Roy, M.-H. (2018). Adapting ensemble predictive modeling for educational video games. IDEAS SoCal AI & Data Science Conference, Los Angeles.
Roy, M.-H., Larocque, D., & Dupuis, D. (2015). Robust variable selection with a multiple-step bootstrap procedure. Joint Statistical Meetings (JSM), Seattle.
Roy, M.-H. (2013). A study of random forests using robust aggregation methods and splitting criterion. Joint Statistical Meetings (JSM), Montréal.
Roy, M.-H. (2012). Robustness of random forests for regression. International Conference on Robust Statistics (ICORS), Burlington.
Education & Teaching
Ph.D., Decision Sciences (Quantitative Methods / Computational Statistics)
Université de Montréal (program administered by HEC Montréal), 2019
Dissertation: Three Essays on Nonparametric Prediction Intervals and Robust Variable Selection
NSERC Alexander Graham Bell Canada Graduate Scholarship (CGS-D), Doctoral
M.Sc., Business Intelligence (Statistics & Data Mining)
HEC Montréal, 2011
With Great Distinction (GPA 4.09 / 4.3); Honour Roll. Thesis: Robustness of Random Forests for Regression
B.B.A., Financial Markets
HEC Montréal, 2008
University Teaching
Lecturer, Statistics
HEC Montréal, 2013 – 2014
Instructor of record for an undergraduate probability and statistics course, delivered to three sections (in French).
Certifications
AIGP — AI Governance Professional (In Progress)
International Association of Privacy Professionals (IAPP)
Contact
linkedin.com/in/roymh · CV (PDF)
Dallas–Fort Worth, TX