Master Thesis

Dema helps leading brands make smarter marketing investment decisions through advanced analytics and modern Marketing Mix Modeling (MMM). Our platform combines statistical rigor with usability, enabling marketing teams to understand what truly drives performance and growth.

We are now looking for a curious and motivated
Master's student
eager to explore state-of-the-art techniques in marketing measurement. You should be excited to work at the intersection of statistics, machine learning, and AI, and to test emerging modeling approaches that push beyond traditional methods.

About The Role
At Dema, our current MMM framework is built on Bayesian regression, delivering robust, interpretable insights across clients and markets. However, we believe there's significant potential to model even more complex, non-linear relationships between multiple regressors, especially in how marketing and organic channels interact over time.

The goal of this thesis is to investigate and evaluate transformer-based neural networks for marketing measurement. Introducing high-dimensional embeddings and attention mechanisms to model the dynamic relationships between marketing activities, organic signals, and sales outcomes, offering a fundamentally new way to understand marketing effectiveness.

Your work will focus on exploring whether such models can complement or outperform Bayesian MMMs, especially in terms of predictive accuracy, long-term effect modeling, and interpretability.

This project will be conducted from Dema's office in Stockholm. You'll have close collaboration with our data science team, access to real-world marketing data, and the opportunity to influence the next generation of Dema's modeling framework.

Your responsibilities

• Implement and test Transformer-based models inspired by the NNN framework
• Compare model performance with Dema's current Bayesian MMM across predictive and interpretability metrics
• Analyze how embeddings and attention mechanisms capture relationships between marketing, search, and sales data
• Collaborate with Dema's data scientists to discuss findings and propose model enhancements or hybrid approaches

What do we offer?

This master's thesis offers a unique opportunity to apply advanced machine learning in a real-world marketing analytics context. You will work hands-on with real marketing datasets, gain exposure to Dema's modeling infrastructure, and explore the balance between interpretability and complexity in modern marketing measurement.

You'll also have the chance to:
Collaborate directly with experienced data scientists and marketing analysts, experiment with state-of-the-art architectures and compare them with production-grade Bayesian models and gain valuable experience bridging academic research and commercial applications.

A suitable background

You are currently pursuing a Master's degree in Statistics, Data Science, Machine Learning, or a related field, with coursework and/or experience in:

• Programming in Python
• Statistical analysis and modeling
• Machine learning and deep learning architectures
• Understanding of time-series or causal modeling is a plus

Who you are as a person:

• Eagerness to learn and explore new modeling approaches
• Ability to review and adapt academic research for applied use cases
• Strong analytical and problem-solving mindset
• Self-driven, structured, and proactive in finding creative solutions

Dema's core values

At Dema, we value curiosity, collaboration, and accountability. We believe in building models that are not only powerful but also transparent and actionable. To thrive and succeed here, you should be motivated by learning, experimentation, and teamwork,always with an eye toward impact and practical results.