How To Develop Analytical Intuition - Issue 254

Error-spotting and how to cross-check metrics when the data won’t match.

Before we dive into all things statistics, a quick reminder:

Just 1 week left until Data Council in Oakland!

This is your rare chance to meet the founder of Looker, test whether the Databricks VP of AI actually knows any math, ask Sigma how a spreadsheet could possibly cost that much, and most importantly, find out why on earth OpenAI is using Fivetran.

I mean, if OpenAI can’t build AI to automate their own ETL… why should we?

I’ll be having so much fun interrogating people. Come join me - use the code daj20 to save 20%.

---

Have you ever wondered why so many companies, when interviewing for analytical roles, ask sampling questions like:

• How many dentists are in the world?

• How many M&Ms can you fit in a jar?

• How many cars are there in the US?

• How many photos are taken at the Eiffel Tower in a day?

They’re not asking these questions to make fun of your misery - they’re testing your analytical intuition.

In addition to having SQL and Python skills, wearing the analyst hat requires critical thinking and the ability to arrive at the answer or spot an error without much context or research.

This publication is a follow-up to Mastering Critical Thinking: How to Improve Your Analytical Skills, where I introduced problem-solving, the lifecycle of analysis, and the types of analysis we do at work. Today, I will continue with analytical thinking with a focus on sampling and techniques for working with ranges to help you quickly spot a pattern, validate an assumption, or catch an error on the fly.

How to solve sampling questions

While SQL and Python can be taught relatively easily, analytical sense is much harder to develop. My first mentor, Andy Velushavi, used to say that some people aren’t simply wired for this kind of thinking and that they may not be destined for a career in analytics. He was very good at spotting this skill in analysts. But I’m convinced this skill can be trained and developed. And that’s exactly what I want to focus on today.

Since I brought up sampling questions above, let’s talk about how to answer them in an interview. Questions like:

• How many windows are in NY city?

• How many teachers are there in the world?

What interviewers are really looking for here is your ability to set reasonable ranges and break down sampling. This skill shows up everywhere in analytics - data audit, A/B testing, root cause analysis, deep dives, or business health metrics overview.

To answer these, start with an educated guess based on something related to the question — a proxy value that you do have some intuition about. Then, work your way toward a ballpark estimate using averages and scaling logic.

What’s most important is to get to a particular number. It doesn’t have to be perfect (it won’t be), but you need to land on an approximate value or rate and explain how you got there.

Here is an example of tackling a sample question - The Infamous 'How many Windows are in NYC?'​ Interview Question.

If you want a refresher on sampling questions, try Simple Random Sampling.

If you want to go deeper and study sampling, read this: Sampling and Estimating: How Many Jellybeans?

How to know if your data is accurate