Landing a job as a (Growth) Analyst in 2024

A minimal set of skills to land a job as a Growth Analyst

Introduction

This post will be relevant to you if:

You would like to work as a growth analyst / growth data scientist or you are just curious of what growth analysts do

You are an analyst and would like to gain more skills and transition into a growth related position

You are a business leader or hiring manager and would like to know what to look for in order to hire a growth analyst.

For those of you that are not very familiar with the occupation of a growth analyst let me break it down for you.

A growth analyst uses data, statistical tools and domain expertise in order to:

• Support the business in picking the right KPIs and track progress against these KPIs

• Draw associations between different business metrics and study the business as a system

• Support the business in identifying and testing the effectiveness of growth levers against various KPIs efficiently

• Form and test hypotheses over changes in KPI norms (root cause analysis)

In some businesses the term ‘growth’ might refer to the stage of the business (start-up) and not necessarily on the responsibilities of the analyst. In that case, it might require a more general set of skills that would allow the analyst to contribute in data engineering or other data products.

Programming

In my opinion, you can be effective in a diverse set of environments with the mastery of just two languages:

• python

• SQL

With python you can:

• Perform statistical analysis, e.g. hypothesis testing, regression analysis and present findings with various charting tools

• Build machine learning models for predictions, recommendations and other applications

• Build APIs to serve applications, models or reports

• Schedule queries (Airflow)

With SQL you can:

• Run queries on a database and aggregate data for reporting

• Load data from a database to a python-based environment for further analysis

• Schedule, process and migrate data between tables (DBT)

Statistical Analysis

In terms of extracting knowledge out of data you can do the majority of the work focusing on two main categories:

• Causal Inference

• Statistics & Machine Learning

Causal Inference

I would argue that causal inference is a must-have tool for analysts. The more experience I gain and the more I study causal inference the more biases I am able to identify in my day-to-day work. Switching to thinking in terms of causal graphs was an aha-moment for me. Once you understand concepts like backdoors, synthetic control and difference-in-differences it’s really enlightening.

Statistics & ML

Here is a list of statistical analysis methods I use ordered by frequency of usage:

• Basic hypothesis testing, e.g t-tests, bootstrap tests

• Linear regression for analysis, experiments or even basic trend forecasting

• Logistic Regression

• ML regression or classification algorithms (xgboost is all you need) for data products, e.g. churn prediction, spam classification, valuation models

• In more rare occasions, word embeddings (word2vec), NLP (e.g. count vectorizer, naive bayes) and dimension reduction methods (e.g. PCA, t-SNE) for visualization.

Business Acumen

Last but not least, I cannot stress enough the importance of understanding how businesses grow, how they make money, how customers behave and why they are or aren’t willing to buy your product as well as the competitive environment the business operates in. Even exogenous shocks in the macro environment can change things in a big way (e.g. Covid, inflation, etc.).

A few domains of science that I personally found relevant are:

• Basic microeconomics: supply and demand curves, utility functions, the profit optimization problem, demand modeling and pricing optimization, game theory

• Basic finance: time value of money, compounding growth rates, net present value, modern portfolio theory, reading an income statement

• Business models: B2B vs B2C, marketplaces, network effects, subscriptions

• Business functions: Product, Marketing, Sales, Engineering, etc.

• KPIs: retention rates, RPU, RPPU, active users, LTV

• Growth related concepts: Growth models, product-market fit, unit economics, referrals, flywheels

Final Thoughts

In my experience, people trained in quantitative domains will have it easier in acquiring the necessary know-how for data science related jobs, whether this is math, statistics / operations research, engineering, physics or economics, though this shouldn’t be a blocker for someone motivated to do the hard work.