Hi! My name is Randall Lewis, and I am a causal measurement expert. I design and build systems to guide business decisions. In today’s digital economy, this requires inventing new methodologies and integrating them into products to facilitate causal machine learning analyses at scale.
I have more than 13 years of experience developing advanced techniques that measure the causal impact of advertising, estimate demand curves to calculate optimal prices, compute insightful AB-testing platform analytics, and quantify the value of entertainment media in order to help both humans and automated decision systems make better choices.
I am currently the Chief Scientist at Nanigans. In that role, I lead the scientific vision of an advanced toolkit for large-scale causal modeling, aiming to make causal modeling practical for measurement and reporting throughout the world of data science. I partner with product and engineering to prototype, develop, and deliver foundational technologies and practical solutions to customers. The Noumena Causal Modeling tools include the Timeline Query Language (TQL) and Causmos, a causal modeling library for creating and enhancing causal models starting from first principles: Which records denote the Outcome, Treatment, Intervention, and Opportunity? I use the technology to quickly explore a wide space of candidate models for AB-testing, ad-attribution, and customer-lifetime-value applications.
The toolkit empowers data scientists to tune and evaluate causal models throughout the steps of development and management: source data ingestion and preparation, training/test dataset creation and featurization, estimators/estimation frameworks, optimization algorithm hyperparameter tuning, goodness-of-fit and out-of-sample metrics, offline scoring and replay metrics, and heterogeneous insights.
Contact Information: Randall Lewis
- Email: randall [AT] econinformatics.com
- CV: Randall Lewis, PhD
- LinkedIn: Exogenous Variation
- Facebook: Exogenous Variation
Formal Biographical Sketch:
Randall Lewis is a world leader in the science of incrementality. He designs and builds causal measurement and prediction systems to drive business impact. He has developed advanced techniques to measure the causal effect of advertising, estimate demand curves to calculate optimal prices, create advanced AB-testing platform analytics, and quantify the value of entertainment media in order to help both humans and automated decision systems make better choices.
His research has been published in top journals and conferences in the fields of economics, computation, and marketing and has been recognized by “best paper” awards in the top 3 academic journals in marketing. He is also the co-inventor of 10 patents as well as Google’s award-winning Conversion Lift “Ghost Ads” methodology.
Randall attended MIT as a Presidential Fellow where he earned his PhD in economics with a focus on econometrics, experimentation, and industrial organization. Thereafter, he worked as a Director of Economics at Netflix and a Senior Economic Research Scientist at Netflix, Google, and Yahoo. In these roles, he served as a scientific advisor and individual contributor for a wide range of business operations: product, pricing, promotion, and placement. He consulted with scientists and engineers on methodological innovations for decision-oriented projects requiring expertise in causal inference, econometrics, machine learning, and statistics. He partnered with management, business stakeholders, and engineering to advance innovative solutions (e.g., incrementality, counterfactual reasoning, causal machine learning), perform technical research (e.g., estimated demand, analyzed experiments), develop decision tools (e.g., counterfactual pricing calculators), and share insights with the CEO and other executive staff for business-critical operations, routinely guiding decisions affecting >$100M in revenue.
In his current role as Chief Scientist at Nanigans, he leads the scientific vision of an advanced toolkit for large-scale causal modeling, aiming to make causal modeling practical for measurement and reporting throughout the world of data science. The toolkit empowers data scientists to tune and evaluate models throughout the distinctive steps of causal model development and management, accelerating the delivery of business impact through causal insights or deployment of decision models to production.
His work has required him to tackle empirical economic research at scale. Large-scale data can be instrumental for measuring statistically small but economically valuable quantities. The diffuse effects of advertising exemplify this work, which he calls “econinformatics.” The economic, conceptual, and technical skills required by this new class of problems has led to a new subdiscipline: tech economics. Randall’s efforts include promoting the diffusion and development of this expertise within the field of economics and industry.