Do I want to analyze data for humans or machines?
Who your consumer is changes your job immensely
As you are considering the type of data role you want, you should think broadly about whether you would prefer to be analyzing data for people or for machines. This affects your day-to-day life in all kinds of ways and has a huge impact on the type of skills you’ll build in the role.
Analysis jobs with human clients are more common, for now. Examples of these include working as an analyst for a consulting firm or a government agency. In these roles, you’re tasked with drawing insight from data and packaging that insight in a way that it can affect change within the client’s organization. If the humans getting your insight aren’t somehow acting on it, you are worth very little to the organization. The skills you’ll build follow this. You’ll learn how to:
Simplify information as much as possible
Make judgments based on organizational context for when something needs to be statistically rigorous or when using data to support an existing story is okay
Manage people’s expectations about data and what it can do for their business
Use data to create logical arguments, with one insight leading to another
You will almost certainly have times where you feel you aren’t being rigorous enough but the client just wants any data to support their thinking. You may feel frustrated by the lack of truth-seeking and the use of data to validate existing heuristics. It will be your job to push back on this when necessary.
On the other hand, analyzing data for a computer rewards truth-seeking to a much higher degree. For example, if you’re an analyst at an ad-tech company and you’re charged with improving the targeting algorithm that serves users ads, your value to the company will be determined by how much better you can make the algorithm on a much more objective scale. You don’t need to spend time and energy convincing the algorithm to change. But you do need to spend a lot of time and energy making sure what you’re doing is right. As such, you’ll learn to
Use statistical methods to design experiments
Make sure you’re not overfitting data
Test and disprove hypotheses quickly
But you won’t learn as much about communicating your insights to people because you won’t need to. And for most people’s careers that is a more valuable skill than technical acumen.
So think a lot about the things that give you energy and the skills you want to develop and when you are deciding which roles to apply for, make sure to get a sense of whether you’ll be analyzing data for humans or machines.