• Article: See Why Everyone Gets the Monty Hall Puzzle Wrong: A great article on the cognitive biases and blockers that make The Monty Hall problem so unintuitive.
• Article: The Hidden Pitfall in Marketing Experiments: Understanding Uncertainty: An article on why A/B test results often don't translate to real-world gains.

• In the DMs: Have you seen any examples of successfully incorporating media audience segment dimensions into these causal dags? I.e. splitting the same channel into new customers vs. returning customer traffic.

• Around the Web: A/B Testing Rigorously (without losing your job).

• DevOps for Data Science: The dev ops for data science book. Looks to be a great resource for those looking to put complex models into production.
• Decayed estimators for time series: Cool video for using “decayed” estimators for time series analysis. Similar idea how to Recast thinks about the problem though a very different implementation
• How to measure brand: Another article from our friend John James on measuring brand. A great framework to help you get started when it comes to the tough conversations on this tricky concept.
• A/B Testing Rigorously (without losing your job): A fun discussion of repeated significant testing errors and how to think about adjusting for them.

Question: Have you seen any examples of successfully incorporating media audience segment dimensions into these causal dags? I.e. splitting the same channel into new customers vs. returning customer traffic.

"Yes, absolutely! In our opinion, every statistical analysis should start with a causal DAG. So whatever question you have, you should start by thinking about the causal relationships."

- Michael Kaminsky, Recast Co-Founder

If someone forwarded you this email, you can subscribe here.

If you're already subscribed, why not help us spread the word by sharing it with a friend or co-worker! 🙏

• MMM E-Book: Everything you need to know about MMM in one place
• The MMM Checklist: The essentials for every MMM project you undertake
• Modern Analytics for Marketers: Our YouTube series on validation in MMM
• Why Recast?: MMM should be fast, accurate, and trustworthy. Recast was built from the ground up and reflects all of these values directly in our platform.