Data Scientist
Role Overview
The Data Scientist will own the measurement science behind Audiohook's performance audio advertising platform. You'll design and run incrementality tests, build and maintain marketing mix models, and apply causal analysis to quantify how Audiohook drives outcomes for advertisers. This role combines hands-on modeling with the opportunity to shape how we prove value to customers, sharpen our bidding and optimization systems, and influence product direction. You'll collaborate closely with Engineering, Product, Sales, and Customer Success to ensure measurement isn't just statistically sound but operationally useful.
Key Responsibilities
Marketing Measurement & Causal Inference
Design and run incrementality experiments (geo, ghost bidding, holdout, PSA) that quantify Audiohook's lift for advertisers
Build, maintain, and evolve marketing mix models (MMM) and multi-touch attribution analyses across customer campaigns
Apply causal inference methods — difference-in-differences, synthetic controls, instrumental variables, propensity scoring — to questions that can't be answered with RCTs
Translate measurement results into clear narratives for advertisers, internal stakeholders, and the product team
Modeling & Analysis
Partner with Engineering on the data and modeling layer that powers bidding, pacing, and optimization decisions
Develop and validate predictive models that improve campaign performance and platform efficiency
Instrument experiments and analyses for reproducibility, monitoring, and ongoing measurement quality
Cross-Functional Collaboration
Partner with Sales and Customer Success on measurement studies for priority accounts and renewals
Partner with Product on roadmap inputs grounded in causal evidence, not just descriptive data
Present findings to advertisers, internal teams, and leadership in clear, decision-ready formats
Communicate clearly and proactively in a remote-first environment
Qualifications
Required
Bachelor's or Master's degree in Statistics, Economics, Data Science, Computer Science, or related quantitative field
3–5 years of applied data science experience with a focus on marketing measurement — incrementality, MMM, attribution, or causal analysis
Hands-on experience designing and analyzing experiments (A/B, geo, holdout) in a marketing or advertising context
Strong fluency in Python (pandas, statsmodels, scikit-learn, PyMC, or similar) and SQL
Solid grounding in statistical inference, regression, and causal methods
Ability to communicate technical results to non-technical audiences — advertisers, sales, leadership
Excellent attention to detail and intellectual honesty about model limitations
Preferred
Experience in adtech, digital advertising, or media measurement
Experience with Bayesian methods or Bayesian MMM frameworks (e.g., PyMC-Marketing, LightweightMMM, Robyn)
Experience working with large-scale ad event data (impressions, clicks, conversions) and modern data stacks (e.g., Iceberg, Snowflake, BigQuery)
Experience in a startup or high-growth company
Comfort using AI tools to accelerate exploratory analysis, code, and write-ups while maintaining methodological rigor
What We Offer
Fully remote work environment
Competitive salary and equity opportunities
Performance bonuses
Health, dental, and vision benefits
Other benefits such as daily lunch stipend, monthly wifi, cell phone and subscription reimbursement, and annual hardware stipend
Flexible PTO and remote-friendly culture
Bi-annual Corporate Offsites
Opportunity to help shape a function at a rapidly scaling tech company
Skills
As published by ashby · 7 questions · 5 written answers
Basics
Name, Email, Resume
Short answers (1)
- Please specify your desired salary.
Pick from a list (1)
- This position does not offer immigration sponsorship at this time (e.g., H-1B, TN, etc.). Will you now or in the future require sponsorship to work in the United States?
Written answers (5)
- Tell us about a specific incrementality test or causal analysis you designed and ran end-to-end. What method did you choose, why over the alternatives, what did the result actually show, and what changed as a result?
- Describe an MMM or attribution model you built or substantially improved. Walk us through the data, the model choice, the priors or assumptions you had to make, and the single decision you're least confident about in hindsight.
- Tell us about a time your analysis pointed to a conclusion that stakeholders — sales, leadership, an advertiser — didn't want to hear. What was the result, how did you handle the conversation, and what did you do when pressure came back to soften it?
- Describe a specific time a model or experiment produced results that surprised you or didn't match expectations. Walk us through your hypothesis process — and importantly, what you got wrong before you got it right.
- How do you use AI tools in your data science work today? Give us a specific example where AI changed your output — and a specific example where it produced something you had to throw out or significantly rework. What practices do you have to catch AI errors in statistical work specifically?
