Skip to main content

Data Intuition

Premium

BizOps interviews test candidates’ abilities to shape data into business insights. Data intuition in BizOps interviews means being able to quickly identify the right data for answering the business question and noticing patterns and errors quickly.

What Is Data Intuition?

Data intuition refers to your ability to draw insights from data and use those insights to solve problems. When someone asks you a question, do you immediately know what data you need to answer it? When you look at a set of numbers, can you immediately find trends and develop testable hypotheses as to why the trends exist? What do these trends mean for the business and how do they fit into what we know about the business and the world?

This is a key skill for bizops.

Seeking the Right Data

There are many common metrics in the tech industry that you can study as a baseline. However, excelling here really means customizing common metrics to your specific use case.

For instance, most companies track weekly active users. However, if you're trying to answer a question having to do with revenue, then perhaps the data to look at is weekly active users contributing revenue.

Tip: In the interview, you'll get bonus points for describing how you would get custom data. For example, to answer questions around users of Instagram shopping, we would want to see weekly active users who contribute revenue. How would we define those users? Perhaps we select only users who click on ads or make purchases on Instagram shop.

Another great approach to thinking about what data to seek is to call out user journeys or conversion funnels. Oftentimes, the behavior we care about is at the end of a set of actions, so these frameworks can be extremely helpful for identifying the key step in the journey driving the outcome we're interested in.

Below are some common metrics to get you started answering any questions with data:

Consumer metrics

  • Unique Users (daily / weekly / monthly, depending on the frequency of usage that makes sense for the product)
  • Retention (again, depending on the frequency of usage that makes sense for the product)
  • Conversion Rate (open rate / click through rate / flow completion rate)

Monetization metrics

  • Revenue (total or per user)
  • Unit Profitability (profit / user or profit / item)
  • Customer Acquisition Cost
  • Customer Lifetime Value

We’ll talk about the full range of metrics that you should be thinking about from a data point-of-view in the metrics lesson.

Gut-checking Data

There are a few things to look for before getting your hands dirty.

Data cleanliness

Once you know what you're looking for, do a quick check for some common drawbacks of real-world data.

Quickly scan for data that is:

  • Incomplete (ex. role = “ “)
  • Noisy (ex. age = “-31”)
  • Inconsistent (ex. zip_code = “0000-000” in some places and “0000000” in others)

Other indicators of untrustworthy data include duplicates, missing values, or label errors.

Starting with high-quality, reliable data is extremely important, especially if you'll be manipulating data to come up with custom measures.

Trends

Once you're confident the data you need is clean and complete, it’s time to start looking for trends. You should first think about cyclic patterns that you expect to see in the data.

Any consumer product would see daily increases and decreases in volume because people won’t be using that product at night. Many products see annual seasonality related to holidays, companies’ budget cycles, etc.

These patterns help you figure out how to take out the noise of these cyclical trends and zoom in on trends that actually matter.

Many interviews will also test your ability to gut-check the data for any errors. For example, is the trend that you’re investigating really a data issue? Does the magnitude of the metric make sense? Performing these gut checks before digging too deeply into an investigation will save you a lot of time.

Finding Patterns

Finally, you can start looking for the "so what?" in the data. You should always think about patterns from three dimensions:

  • What could be causing this pattern? There are usually natural conversion funnels or levers that you can dig into to answer this question.
  • Why is this pattern happening? Whether it’s consumer behavior or macro-environment trends, have some hypotheses.
  • How does this trend connect to broader business health? For instance, a profitability decrease is not necessarily a bad thing if the business priority is growth.

Remember, data patterns are only as useful as the insights and decisions they inform.

Recap

Interviewers are interested in your data intuition because it falls to bizops to make recommendations, often based on messy or incomplete data. This'll be assessed during case questions ("generate recommendations based on this data") and/or through quantitative rounds or a take-home assignment.

To show you can intuit the value of (and problems with) data:

  • Leverage your knowledge of metrics, the customer journey, and/or conversion funnels to hone in on the right areas for your specific use case.
  • Call out any red flags you see in raw data including incomplete, noisy, or inconsistent data.
  • If you need custom data, articulate in detail how you'd pull it to answer your question.
  • Quickly evaluate any trends and patterns you find to help you down-select the right metrics and eliminate options that aren't actually data issues.
Next Lesson