Graduate Marketing Scientist - Austin, TX

Fospha•Austin, TX

5d•Onsite

About The Position

Fospha is seeking a Graduate Marketing Scientist to join their Marketing Science team in Austin, TX. Fospha provides marketing measurement products for e-commerce brands, including attribution, marketing mix modeling, incrementality testing, and brand impact measurement. The Marketing Science team focuses on the applied aspects of these products, designing and delivering incrementality tests and MMM engagements for clients. This is an entry-level, hands-on role where the successful candidate will work on live test and MMM delivery under supervision from the start, and will be the first point of contact for client data discrepancies. The role is ideal for someone who wants to learn causal measurement in a business where it is the core product.

Requirements

Strong foundation in maths and stats.
Clear communication skills.
Working proficiency in SQL.
Python, or a demonstrated ability to pick it up quickly.
Grounding in inferential statistics (hypothesis testing, uncertainty, statistical power, meaning of null results).
Some exposure to experimental design (randomisation, control groups, confounding, impact of poorly designed tests).
Strong AI fluency (using AI tools for problem-solving and knowledge gaps, QAing AI output).
Clear, concise written communication.
Composure in client-facing conversations, including when clients are unhappy.
Real attention to detail.
Discipline to log things consistently.
High agency.

Nice To Haves

Exposure to marketing, ecommerce, or advertising data.
Familiarity with Bayesian methods and/or modelling.
Experience with cloud data tooling.
Experience presenting to or supporting external stakeholders.

Responsibilities

Assemble and validate test data, including geo-level spend and conversion series, checking pre-period parity between treatment and control, and identifying coverage gaps that could invalidate a design.
Support test design under review, including market matching, control selection, power and minimum detectable effect sanity checks, and identifying contamination risks.
Run analysis and interpret results honestly, including pre-treatment fit diagnostics, lift estimates with intervals, and the implications of null results.
Qualify client data for MMM, assessing spend coverage, variation in spend, series length and granularity, collinearity between channels, and gaps that could bias results.
Assemble and validate model input datasets, and investigate discrepancies that arise.
Support model runs and interpret diagnostics such as fit, residuals, convergence, and the plausibility of channel contributions.
Contribute to output-extension work under review, such as forecasting or budget scenario work derived from existing MMM results.
Compare results across different methods (MMM, incrementality, platform-reported figures) and understand the reasons for discrepancies.
Act as first and second line support for client trust queries, investigating number changes and isolating causes using SQL.
Distinguish between bugs and methodology changes, considering factors like attribution window changes, model recalibration, data feed gaps, and platform reporting shifts.
Triage PSPs (Platform-Specific Problems) on model trust, resolve issues where possible, and escalate genuine model problems with a clear diagnosis.
Reconcile platform-reported figures against Fospha's measurements, understanding discrepancies in metrics like ROAS.
Log and tag incidents consistently to identify recurring failure patterns for potential automation.
Run templated explainer sessions under review, guiding clients through Fospha's measurement methodologies.
Draft documentation and presentations beyond core explainer content.
Feed recurring query themes back into source material for documentation and enablement.
Fact-check methodology claims in product marketing collateral.
Log and tag trust incidents consistently so recurring failure patterns become visible and can be automated away.
Surface themes from the query queue into workflow design.