Research strategy
Designing research systems, standards and studies that answer the right commercial question—not merely the one first asked.
Research Lead · PhD Candidate · Warsaw
I study how people decide — then build better ways to understand it.
I’m Maciej Kielar, a researcher working across marketing science, behavioural research, data science and machine learning.
Explore the work01 Practice
Rigour for the real world.
My work sits between the clarity of academic research and the pace of commercial decision-making.
Marketing science
Segmentation, MaxDiff, conjoint, experimentation and brand measurement made useful for real decisions.
Data systems
Reproducible R and Python workflows that turn fragile manual work into dependable analytical infrastructure.
Applied ML
Using language models, NLP and behavioural modelling where they genuinely improve the research process.
02 Selected work
Methods that move.
Selected themes from commercial work. Client details remain where they belong; the approach and impact are the story here.
A
Decision modelling
Ten thousand years of demand
A Monte Carlo model of demand, capacity and pricing for research services across Nordic markets. Count models (Poisson and negative binomial), PERT-distributed project sizes, correlated growth and the effect of AI automation on workload, bootstrapped from real project data—so pricing options could be tested against team capacity before anyone committed to one.
- Monte Carlo
- R
- Scenario analysis
10k
B
Applied AI
Translation that never leaves the machine
Report translation with the multilingual MADLAD-400 model running entirely on local hardware—converted with CTranslate2 and quantised to 4-bit, with no data sent to external services. Proof that LLM-based translation works on standard corporate hardware without compromising confidentiality.
- LLMs
- On-premise
- Quantisation
4-bit
C
Analytical automation
From twelve days to 42 seconds
In 2024, eye-tracking studies meant extracting zone-coded results from presentation decks and placing them into Excel by hand. I built a VBA tool that pulled the data out and populated the workbook automatically—faster, and without copy-paste errors. A past project; whether it is still in use, I can’t say.
- VBA
- Excel
- Eye-tracking
42 sec
03 Doctoral research
The Big Two · Agency & Communion
One idea.
Three lenses.
My PhD at ESCP Business School studies agency and communion across three empirical worlds. The thread is conceptual continuity; the method changes with the question.
Chapter 01
Literature & text
A systematic review, SciBERT classification and topic modelling map how the Big Two appear in consumer research.
Evidence synthesis · NLPChapter 02
Causal experiments
Experiments examine vicarious moral licensing and the drivers of sustainable consumption.
Experimental design · Causal inferenceChapter 03
Digital behaviour
Digital trace data from Dota 2 operationalise agency and communion through observed patterns of play.
Behavioural data · Multilevel models04 Profile
“The most valuable research does more than explain a result. It improves how the next question gets answered.”
Researcher by training.
Builder by instinct.
I have worked across academic and commercial research, from teaching microeconomics and mathematics to developing analytical tools and guiding market-level methodology. I’m at my best when a problem needs both conceptual precision and a practical system.
Today, I lead research for dentsu’s Norway market while completing my PhD at ESCP Business School and Université Paris 1 Panthéon-Sorbonne.
- 2026—Research Leaddentsu · Norway market
- 2025—26Research Expertdentsu
- 2024—25Quantitative ResearcherNielsenIQ
- 2020—26Doctoral StudentESCP Business School
- 2021—24Junior LecturerMathematics & Microeconomics · ESCP
Working with
R / Python / SQL / Causal inference / Psychometrics / Monte Carlo simulation / NLP & LLMs / Experimental design / Qualtrics / Data visualisation