The First 90 days at your New Data Analyst Job
Estimated reading time: 3 minutes.

You’ve just started a new, exciting data analyst job, and you’re eager to impress your peers. The first 90 days at any job determines your long-term success. Here are simple tips on achieving success at your new role…

1. Understand schemas

Your organization’s database schema encodes the business processes and institutional knowledge that has made it so successful. As with any schema, it carries baggage from the many iterations over the course of years in operation – including technical debt. It’s super important to understand the data model and the relationships between the different database tables. Primary key and foreign key relationships tell you about the consistency of your data model, but from my experience a lot of database tables are missing these characteristics.

With the rise of high performance, columnar, parallel databases like Amazon Redshift, it is common to create highly normalized tables. Unlike yesteryear practices of creating denormalized tables and defining the schema upfront to speed up query performance, these newer faster databases have very fast join performance. Thus, “schema-on-read” is now an acceptable approach to data modeling.

2. Always be suspect

I’m always being cautiously optimistic when sharing an analysis with others. Yes, I’ve cleaned bad data, removed outliers, normalized numbers to compare apples with apples. Still bad data can creep up on you.

A single bad KPI will reduce the trust your stakeholders will have on the rest of your analysis.

Always be on the lookout for data that doesn’t make sense. Not suggesting you remove or omit the anomaly, but definitely dig deeper.

3. Learn Statistics

You don’t need to know Rocket Science level statistics to be effective!

It’s important to communicate not just data points but also a story around it. The spread or distribution of data, i.e., the statistics is exactly that – a summary of your data.

The very basic notions of statistics and probabilities will go a long way:

  • Basic statistical metrics: mean, variance, standard deviation, quantiles
  • Basic probability distributions: uniform, normal, exponential, binomial
  • Statistical plots: histogram, density, box-plot

4. Learn to tell a story

Dashboards and charts are all great, but to be truly effective, you’ve got to be able to tell a story around the whats, hows and whys of the data.

Why’s this trending up? What’s the reason for this anomaly? What can we do to mitigate a future incident? How can we setup alerting around major deviations around the mean?

Stories make your data come alive – people can put themselves into your narrative and visualize what the future will look like. Powerful.

Telling great stories with data will make you a better analyst.

5. Listen all the time.

In order to deliver a good analysis, you got to be able to listen to the ever-changing demands of the business environment. Most times, business requirements are never presented as “what was X last month?”

Make sure you understand business KPIs but don’t let that limit you in your analysis. The best data analysts are always looking at incoming data to find new trends and patterns.

Keep your ear to the ground and actively listen for potential changes to business requirements. Being able to provide fast iterations on your analysis will make you the point person for data related questions.

Good Luck, and let us know if we can be of any assistance!

Photo Credit: Unsplash