Note: The job is a remote job and is open to candidates in USA. HighLevel is an AI-powered business operating system that supports agencies, entrepreneurs, and SMBs in building and scaling their operations. The Staff Data Scientist will define the methodology for experimentation and causal inference to enable data-driven product decisions and business strategies across the company.
Responsibilities
- Define the end-to-end methodology every team follows - hypothesis → metrics → design → power → readout → decision - and make it the default
- Own the statistical approach (significance, multiple comparisons, sequential testing, variance reduction like CUPED) for small-sample, fast-paced contexts where classic A/B power is hard to reach
- Build the methods toolkit for our clustered, hierarchical data (user → sub-account/location → agency), where randomization and analysis units differ
- Apply rigorous causal inference (matching, diff-in-diff, instrumental variables, synthetic control, etc) when clean experiments aren't feasible - churn, onboarding, GTM - separating real signal from selection bias, seasonality, and mix effects
- Own the design discipline for running many experiments at once - layering, orthogonal experiments, holdouts, and guardrails that keep concurrent tests from contaminating each other
- Partner with AI/ML teams to design and evaluate experiments for AI features, including measurement for non-deterministic, fast-iterating systems
- Run the experiment review forum and hold the line on what counts as a real result
- Build the Experimentation curriculum and templates that level up PMs and analysts so good design scales beyond you
- Partner with Analytics Engineering on governed, experiment-ready data and consistent metric definitions
- Influence leadership and cross-functional partners on where to invest, translating statistical nuance into clear, decision-grade guidance
Skills
- 9+ years in data science, product analytics, or applied statistics, with deep hands-on experience designing and analyzing online controlled experiments at scale
- Strong applied statistics - frequentist foundations, Bayesian methods, power analysis, variance reduction, and the failure modes of A/B testing (peeking, multiple testing, network/cluster effects)
- Practical causal inference, with sound judgment about when a result is causal versus an artifact of how the data was generated
- Experience in small-sample, fast-paced, multi-product environments -you know when a decision needs a clean experiment and when it needs a fast, good-enough read
- Strong SQL and working proficiency in Python or R
- Cross-functional and senior-leadership influence - you raise others' experiment quality without direct authority
- Familiarity with a modern experimentation platform such as Statsig
- Experience building an experimentation practice or culture from the ground up
- Background in B2B SaaS, CRM, or product-led growth, and familiarity with the measurement challenges these motions create
- Multi-tenant or marketplace product experience (agency → sub-account → end-customer structures)
Company Overview