Causal Roadmap

We have all heard this phrase:

Correlation doesn’t imply causation.

And we even know that the gold standard for measuring causation and not only correlation is A/B testing or RCTs (Randomized Controlled Trials).

Traditionally, the classic frequentist approach involves waiting until the end of an experiment, calculating a T-test or something similar, and then drawing conclusions.

Does this method always work?

No. Sometimes it just doesn’t.

There are countless reasons why:

1. It’s too expensive to launch an experiment.

2. We have already launched the feature.

3. Due to GDPR regulations.

4. We can’t control the randomization, such as with TV ads.

5. We want to stop our experiment earlier.

6. We want to maximize the impact during the experiment.

7. And many, many others.

So, what should we do, and what is actually used in practice?

I am sharing here Causal Roadmap, which you can use to learn about the most effective causal inference methods in practice.

I've broken it down into three main parts:

• AB testing

• Causal inference

• Marketing

Let’s dive into each of them.