At Jellyfish, we believe in the power of diverse perspectives and inclusive collaboration. We welcome individuals who excel in collaborative, varied teams and value the unique contributions that each person brings to the table.

Jellyfish is a global digital marketing agency; a unique fusion of tech enthusiasts, creative minds, and media and data experts all united to empower our clients along their digital journey. Our commitment to embracing diverse perspectives fuels our innovation and strategies that challenge the status quo, reinvent media activation, and craft influential stories for our global clients and their customers. Join us in shaping a future where business growth and personal fulfilment go hand in hand.

Job Description

Reporting to the Senior Martech Director within Jellyfish in Australia. Insights and marketing science operate as a single discipline containing two crafts. The Insights Analyst is responsible for performance interpretation and recommendations. You will be responsible for establishing causation. Both work closely with the Data Scientist.

The role covers both components of Marketing science; designing and running the measurement, and building and defending the models, weighted towards the former on appointment and towards the latter as the capability develops. The Director holds the most complex methodological decisions in the interim. The role sets technical direction for the analysts and data scientists working on measurement. Line management sits with the Director.

Your primary responsibilities will include:

Own the measurement framework across marketing mix modelling, multi-touch attribution, incrementality and experimentation, including how the methods relate to one another.

Determine which method is appropriate to a given question, and identify where the available data cannot support a reliable answer.

Reconcile conflicting results across methods and explain the causes of the discrepancy.

Produce a consolidated view of marketing performance for the client, with assumptions, limitations and confidence stated.

Lead the analysts and data scientists working on measurement: set the method, review the output, maintain the technical standard, and mentor on modelling technique and experimental design.

Design incrementality tests, geo experiments and holdouts, including the validity conditions for each design.

Build marketing mix models, taking increasing ownership of the modelling as the capability develops.

Build custom attribution models in BigQuery where platform- and rules-based attribution is insufficient.

Write advanced SQL against raw event data in BigQuery.

Specify models, test assumptions, quantify uncertainty and state the limitations of each result.

Translate findings into media recommendations that can be implemented within the current planning cycle, and contribute measurement evidence to client planning and strategy decisions.

Present and defend methodology to clients, supported by the Director on the most complex cases.

Conduct performance and audience analysis where it relates to a measurement question.

Build and document the datasets the measurement work depends on.

Monitor the effect of platform, consent and privacy change on the methods in use.

Technical Environment

Measurement framework. How marketing mix modelling, multi-touch attribution and experimentation relate to one another: the validity conditions for each, their respective blind spots, and the methods used to triangulate between them.

Platform. Google Cloud. BigQuery at an advanced level, including complex joins across event-level data, window functions, query optimisation and cost control, scheduled queries and materialised views. GA4 raw exports as a primary data source.

Statistics and data science. Regression and time series analysis, Bayesian inference, causal inference methods, and model diagnostics. Python or R for modelling.

Modern measurement. Marketing mix modelling, including open-source frameworks. Incrementality testing and geo-based experimental design. Holdout design. Attribution beyond platform defaults, including custom and algorithmic models built in BigQuery. Measurement under consent, signal loss and platform change.