Data Scientist

Graduate Data Scientist – Marketing Measurement & MMM Level: Graduate / Junior Location: UK Employment: Full-time - Contract About the role We're looking for a highly analytical graduate to join our team and help develop marketing measurement models for growing ecommerce brands. You'll work with real-world sales and advertising data to answer one of the most important questions ecommerce businesses face: Which marketing activity is actually driving incremental sales, and where should the business invest its next £1 of marketing budget? A major focus of the role will be developing Marketing Mix Models (MMM) for ecommerce businesses generating approximately £1m–£10m in annual sales. You'll work with data from platforms such as Shopify, Meta, Google Ads, TikTok and email/CRM platforms, using statistical models to estimate the incremental impact of marketing investment. This is a particularly good opportunity for someone interested in the intersection of statistics, econometrics, data science and commercial decision-making. We don't expect a graduate to arrive knowing how to build an MMM. We'll teach the marketing and MMM-specific components. We're looking for someone with excellent quantitative foundations who wants to become an expert in this area. What you'll be doing You'll help build and improve our marketing measurement methodology, including:

• Building statistical models in Python to explain and predict ecommerce sales
• Developing Marketing Mix Models across multiple ecommerce brands
• Cleaning and combining sales, advertising and commercial datasets
• Modelling advertising carryover/adstock and diminishing returns
• Analysing seasonality, promotions, pricing and other drivers of demand
• Developing marketing response and saturation curves
• Estimating incremental revenue and ROAS by marketing channel
• Quantifying uncertainty around model estimates
• Running model diagnostics and validation
• Comparing model estimates with marketing experiments and other attribution methods
• Building budget optimisation models to recommend marketing allocation
• Creating clear visualisations and explaining model results
• Presenting insights to colleagues and, increasingly, clients
• Helping develop a repeatable modelling framework that can be deployed across many ecommerce businesses You'll initially work within an established modelling framework rather than being expected to develop the methodology alone. As your experience develops, you'll take increasing responsibility for model design, validation and client recommendations. What we're looking for We're primarily looking for strong quantitative thinking rather than previous marketing experience. You'll probably have recently completed a degree or master's in an area such as:
• Statistics
• Econometrics / Economics
• Mathematics
• Data Science
• Physics
• Engineering
• Computer Science with a strong statistical component Other backgrounds are absolutely welcome if you can demonstrate the required quantitative skills. Essential skills Statistics You should have a good understanding of:
• Linear and multiple regression
• Probability and statistical distributions
• Statistical inference and uncertainty
• Hypothesis testing
• Multicollinearity
• Model selection and validation
• Regression diagnostics We're particularly interested in people who understand why a statistical model might produce a convincing but incorrect result. Python You should be comfortable analysing data independently using Python. Ideally you've worked with tools such as:
• pandas or Polars
• NumPy
• SciPy
• scikit-learn
• matplotlib / Plotly You should be comfortable taking a messy dataset through exploration, cleaning, modelling and validation. Basic SQL and Git knowledge is also desirable. Analytical thinking You should enjoy investigating questions where there isn't necessarily an obvious answer. For example: Meta advertising spend is highly correlated with sales. Does that mean Meta caused those sales? We want someone interested in investigating questions like this rather than simply reporting the correlation. Any of the following would be a significant advantage, but isn't required:
• Econometrics
• Causal inference
• Bayesian statistics
• PyMC or Stan
• Time-series modelling
• Experimental design / A/B testing
• Optimisation
• Marketing Mix Modelling
• Academic research involving observational data A dissertation or personal project demonstrating strong statistical reasoning would be particularly interesting. What you don't need You don't need previous ecommerce or digital marketing experience. We'll teach you about areas including:
• Ecommerce economics
• Meta and Google advertising
• Marketing attribution
• Incrementality
• CAC and ROAS
• Marketing Mix Modelling
• Adstock and carryover
• Saturation and diminishing returns
• Bayesian MMM
• Marketing experiments
• Budget optimisation What success looks like Within your first few months, we'd expect you to be able to take a client's ecommerce and marketing data, prepare it for modelling, run our MMM framework and diagnose common data or modelling problems. Over time, you'll progress from implementing models to understanding why models behave the way they do. Ultimately, we want you to be able to look at an MMM result and ask: “Do I actually believe this result, and what evidence would convince me that it's correct?” rather than simply accepting the model output. Why this role is interesting Marketing measurement is undergoing significant change. Traditional digital attribution often struggles to determine whether advertising actually caused a sale, while businesses increasingly need to understand the incremental return from their marketing investment. You'll have the opportunity to develop deep expertise in modern marketing measurement while working directly with real businesses and datasets. You'll combine statistics, software and commercial reasoning to build models that influence real marketing investment decisions.