Predictive Modeling Director

The Predictive Modeling Director is responsible for designing and implementing advanced predictive models that unlock insights and drive marketing performance optimization for the NAOU Marketing team. In this role, you will serve as the lead data scientist for marketing, translating complex business questions into analytical models and tools that guide decision-making. You will focus on building and refining models that answer critical questions such as 'What is the optimal marketing mix?', 'Who are our high-value customers and how do we retain them?', and 'Where should we invest the next marketing dollar for maximum impact?'. Working closely with cross-functional partners, you will ensure that the models address real needs and that their outputs are understood and applied. You will work with a team of modelers/analysts, and collaborate with the broader Data & Analytics community as well as external vendors or agencies as needed, to deliver best-in-class modeling solutions. Ultimately, your work will enable data-driven planning, smarter targeting, and higher ROI by providing a predictive lens on our marketing strategies.

What You'll Do for Us

Lead Development of Predictive Models: Design and develop statistical and machine learning models to tackle key marketing challenges. This includes taking ownership of our marketing mix modeling (MMM) efforts - updating and enhancing econometric models that measure the impact of different marketing inputs on sales and other outcomes. You will also spearhead other predictive modeling initiatives, such as customer lifetime value models, churn/retention models, segmentation and clustering analyses, and demand forecasting for marketing planning. Starting from business hypotheses or questions, manage the full modeling process: data gathering and preprocessing, variable selection, model building, validation, and iteration. Ensure models are robust, explainable, and actionable, providing not just predictions but insights into drivers (e.g., which media channels are most effective at driving incremental sales).

Implement Modeling Solutions for Optimization: Translate model outputs into practical applications that marketers can use. For example, develop tools or frameworks that leverage model results to simulate scenarios (like a 'what-if' tool for adjusting media spend across channels) and recommend optimal allocations. Work with our technology partners to automate or integrate models into dashboards or planning systems, enabling real-time or regular access to model insights for stakeholders. Ensure that modeling solutions are user-friendly and can be run/updated with appropriate frequency (e.g., MMM updated quarterly) to stay relevant. Additionally, oversee any external modeling vendors or consultants (such as those providing third-party MMM services or software) to ensure their work aligns with our objectives and quality standards.

Collaborate with Stakeholders to Scope & Answer Business Questions: Engage directly with Marketing, Human Sciences and IMX teams to understand their needs and frame the

problems that modeling can help solve. For instance, partner with Media and IMX directors to define the scope of an MMM study (which brands, what time period, which metrics) or a pricing elasticity analysis. Regularly meet with brand managers, connection planners, and others to gather input on what decisions they are trying to inform (e.g., 'How much should we shift from TV to digital?' or 'Which consumer segments should we prioritize for a new campaign?'). Ensure each model or analysis you lead is grounded in a clear use-case and that you and your team clearly communicate the assumptions and limitations. After delivering model results, work with those stakeholders to interpret the findings and brainstorm how to apply them in marketing strategies. Your role is as much about asking the right questions as it is about crunching numbers, ensuring modeling efforts remain business-centric.

Ensure Data Accuracy & Modeling Best Practices: Manage the data inputs and statistical rigor for all modeling projects. Work closely with Data Engineering or IT teams to access and prepare the necessary data (e.g., historical spend and sales data, customer-level data from CRM, media impressions, promotional calendars). Perform thorough data cleaning and exploratory analysis to validate that the data makes sense before modeling. When building models, follow best practices for avoiding bias and overfitting - for example, using out-of-sample validation, significance testing, and business sense checks. Calibrate models using back-testing or holdout samples to verify they accurately predict outcomes. Document model methodologies and assumptions, and maintain a library of models and code for reproducibility. Continuously monitor model performance over time and refresh models as new data comes in or as market conditions change (for instance, if a new media channel emerges, incorporate it into the mix model).

Communicate Insights & Recommendations: Although very technical in nature, this role requires translating model results into clear, non-technical insights and recommendations. After completing an analysis, synthesize the key findings: e.g., 'TV ads are providing diminishing returns beyond X GRPs, while digital video still shows growth in ROI,' or 'Segment A has 20% higher lifetime value than average, suggesting we increase investment in loyalty for this group.' Create compelling presentations and visualizations to tell the story from the data, highlighting how the insights can improve marketing outcomes. Present these insights to marketing leaders and working teams, adapting your communication to the audience (detailed for analytic peers, high-level and outcome-focused for executives). Often, you will provide decision support by clearly outlining options informed by the model (e.g., 'Based on our model, Plan B would likely yield $Y more revenue than Plan A, albeit with higher risk in segment Z'). Aim to establish trust in the models so that marketing partners come to rely on them for planning and optimization cycles.