Lead Product Data Scientist
Company:ZOE
Location:United Kingdom
Seniority level:Lead, 7+ years
Experience:7+ years
Skills:
PythonSQLdbt
Requirements:
- 7+ years of experience in product analytics, data science, or experimentation-heavy roles
- Degree in a quantitative field (Statistics, Maths, CS, Engineering, Physics, Economics, or similar)
- Deep fluency in SQL and Python
- Hands-on experience with statistical modelling and applied ML, such as regression, classification, survival analysis, or time-to-event modelling
- Experience building and validating LTV, churn or retention models, and translating predictions into concrete product or lifecycle interventions
- Strong judgment around model complexity vs. business value—you know when a heuristic beats a black box
- Comfort with messy, real-world data and imperfect signals
- Ability to lead by influence, mentor others, and raise analytical standards
- Clear, structured communicator to both technical and non-technical audiences
- Thrive in fast-moving, low-process environments; aligned with our #ActFast value and comfortable acting on ~70% evidence
Responsibilities:
- Lead product experimentation by introducing advanced statistical testing methods and platform improvements that deliver clear, confident insights for quicker decisions
- Own and evolve core product metrics across activation, engagement, retention, and monetisation to identify risks and leverage points
- Use causal and inferential thinking (e.g., uplift modelling, regression, survival analysis) to move beyond “what happened” to “why”
- Develop lightweight ML models and segmentations that identify the specific levers driving long-term retention and growth
- Set the standard for analytical methods and best practices across the team
- Mentor analysts and lead by example - staying hands-on with data foundations (dbt/instrumentation) and showing the team how to turn raw data into influential narratives
- Apply a "so what?" filter to every project, ensuring complexity is only added when it sharpens a decision, and iterating quickly when reality proves a hypothesis wrong
