About
I'm a marketing data scientist who likes to build. For 8+ years I've worked where marketing, data and engineering meet. Most recently I was Head of Marketing Analytics at Princess Polly, a global fashion e-commerce brand, where I went from Marketing Assistant to Head in 3.5 years.
My focus is causal measurement: finding out what really drives revenue, not what a dashboard claims. Where I can, I build the whole thing myself, from tracking and pipelines to Bayesian models and secure cloud apps with AI built in.
I moved to Hamburg in 2026. I have EU and UK work rights, and I'm learning German every day.
What I do
- Causal measurement
- Bayesian MMM, geo-lift and holdout tests, attribution, and budget allocation that finance and marketing both trust.
- Customer and marketing data
- Single customer views, CLTV, and clean tracking and pipelines that put marketing, finance and planning on the same numbers.
- Cloud and AI systems
- Secure, multi-client analytics products on Google Cloud, with LLMs built in where they help.
Selected work
Finding the campaign GA4 couldn't see
GA4 and our other platforms made one TikTok campaign type look unremarkable. My Python models said it could scale. We scaled it, and it became Princess Polly's highest-return campaign type across every channel.
To make sure the platform wasn't flattering itself, I set up discount codes for every campaign and creator. They confirmed the results independently.
30x+ in-platform ROAS and over 40% of total platform sales, confirmed by independent tracking
One customer, one record
Customers bought through the website, the app, stores, marketplaces, drop-ship partners and TikTok Shop, and each channel counted them differently. I built an in-house CDP layer that deduplicated them into one view.
The same data powered the company's first CLTV model and showed where first and second purchases happen.
New vs. repeat customer metrics 20–30% more accurate
A multi-tenant marketing analytics platform
I built a secure analytics platform on Google Cloud that serves several brands at once, with each client's data kept fully separate. It runs Bayesian MMM, Prophet revenue forecasts and a test tracker, with Claude built in for analysis.
A version of it now runs at Front Row in Hamburg, for two large global consumer goods companies.
Delivered in weeks against a 6–12 month scope
The live demo is private. Message me on LinkedIn for a walkthrough.
Experience
Front Row, Hamburg
2026 to now
AdTech & Performance Marketing
Deployed my measurement platform within weeks of joining; now rolling out to two large global consumer goods companies.
Princess Polly, Gold Coast
2021 to 2026
Head of Marketing Analytics
Causal measurement, attribution and budget allocation for ~$2M a month in ad spend. Grew analytics from one person to a team of three.
Earlier roles
2017 to 2021
Websites, web apps, tracking and automation for small businesses in Australia and Finland.
Tools I use
Python, SQL, PyMC and PyMC-Marketing, CausalImpact, XGBoost, Prophet, Snowflake, BigQuery, dbt, Google Cloud, AWS, Docker, React, GA4 and server-side tracking.
Beyond work
Away from data, you'll find me playing guitar, training karate, in the gym, or making art. I've lived in Finland, Australia and now Germany, and I travel whenever I can.
I'm open to data science and measurement leadership roles in Hamburg, hybrid, or fully remote across Europe. LinkedIn is the quickest way to reach me, and my CV is available on request.
linkedin.com/in/harry-peacock