What does a data analyst do
Started with an internship back in 2016, I found myself working as a full-time data analyst in big tech organizations. In the past, I have worked very closely with product, marketing, content, and dev teams, and the goal was to deliver business insights leveraging data. The data team is the go-to people whenever there is a problem to be solved or a question to be answered. I tend to see myself as a detective, trying to connect the dots in a forest of information. Detective work can sometimes be messy and time-sensitive, and some other times delicate.
For me, an analyst is the point of contact between information and people. Due to the versatile nature of available data, a system is built up to filter out priorities, just like a triage. That’s when the analysts jump in. They analyze the situation, deciding if the resources is available to finish the task. They either get their hands dirty right away, or decide it requires more sophisticated algorithm training from the data scientists.
Depending on the team and the company, the expectations might vary, but regardless the main responsibility is to draw actionable insights from data.
Actionable, for the stakeholders.
That would require a great understanding of the business, sharp data-skills, and also story-telling magic. They use the last one to picture users’ persona and experience with the product
The three pillars and the spectrum
DA jobs in the market tend to lie within the spectrum, one end being database and the other end being business. For smaller teams, full-stack analysts are usually sought, meaning a person covers all aspects of the data route. For larger teams, since the roles have been established and responsibilities are well designated, analysts are expected to perform more focal work, on market, user, or revenue, since many parts of the process have already been cover by other plays such as data scientists. The quality data infrastructure could also influence your workflow. You might find completely different roles under the umbrella term, data analyst.
There are three pillars to data analysis- Statistics, Computer Science, and domain knowledge.
Statistics is the foundation of data analysis work. From descriptive analysis to hypothesis testing, statistics is used to everywhere. My favourite is significant test because it’s usually overlooked.
For computer science and programming, the requirements are not as advanced as for a scientist, but it is always good to programming iterate since it does save tons of time. SQL is sometimes at the top of the list since you’d need this skill to grab all the data to begin with. Python and R are both powerful languages for analysis work; Python’s abilities cover beyond data analysis while R comes with more natural statistical background. Tableau and PowerBI are popular visualization tools that are nice to learn about.
Domain knowledge might be slightly tricky to accumulate. It is your understanding of the industry, the metric, as well as the users. This part connects with actionable insights and business goals, where only when the analyst understands the business can them make valuable suggestions based on data.
How did I get started and what has changed
Things were different in 2016.
There were not many specialized degrees yet, so most people came into the field from a related background. Computer science and statistics are among the popular ones, while psychology was reasonably farther down the line. I studied psychology with proper amount of training in significant tests and programming; however I didn’t realize what I learnt in school could be applied to the business world until my internship at LinkedIn. So naturally I started building up skillsets towards data analysis at that time, self taught more programming and visualization.
Now that it became an established field, there are certainly more resources online than ever.
That is also to say that expectations have been levelled up. Simply cleaning and analyzing the data is not enough. A larger weight has been put on the final deliverables. Not matter how fancy the analysis might be, the more important question is how your work drives the business.
People that serve other functions are also used to having support from data team, and it becomes important to understand what they truly need, instead of what they say they need, or what you can get. Back then anything you can draw from data seems like magic, but now mind-reading (communication) is the true art.
Everyone should become a data analyst
Everyone should become a data analyst, or at least has a data mindset.
This is not to encourage everyone to apply to DA jobs or attend Master’s programs in Data Science. You should only do it if you consider DA as a career option. You can be a data analyst without having the job title.
Analysis with data-based evidence is the gist of it.