The World Economic Forum’s Future of Jobs Report 2020 ranks “data analysts and scientists” #1 for growing job demand, but the two roles are NOT interchangeable. Even before global pandemic struck, companies were struggling to make their data work for them. Now, thanks to the Great Reshuffle, your data analytics team (if you were fortunate enough to have one) may be taking some serious hits. The shortage of data experts can lead quickly to C-suite frustration. We hear the exasperated cries for help: “We need a data scientist!”
But do you? Really?
There are data architects, analysts, and scientists. Each one occupies a distinctly different lane. They do not careen off each other like bumper cars. (We data people never do anything willy-nilly.) Think of them as a problem-solving chain: First, the data architect designs and builds the data warehouse; she brings all the various streams of data into one pipeline and delivers it to a central location where it’s accessible to all. Then, the data analyst processes and interprets the data, turning it into actionable insights. If you reach a point where you have a business question you can’t answer using any existing modeling, then it may be time to hire a data scientist.
We hear the exasperated cries for help: “We need a data scientist!”
But do you? Really?
I am a data analyst who specializes in business intelligence. People like me make data sing. We can make it answer questions and tell compelling stories. But here’s an example of what a data scientist can do from back in my solar energy days: They can use geospatial satellite images to determine how much sunlight is hitting the ground in any one place. Harnessing predictive modeling and machine learning to translate raw imagery into quantifiable metrics. That’s data science. And it requires a ton of data. Here’s a data scientist telling you why you probably don’t need a data scientist. (TL; DR: You don’t have enough data.)
Data Analyst vs. Scientist: Difference Defined
A data analyst makes sense of your existing data. A data scientist invents new ways to collect and interpret data.
That’s the distinction in an overly simplified nutshell. I will expand on it to point out, as the people most immersed in your data, data analysts often run point on improving that data. We are the ones who identify where the data falls short, what might be missing, and/or which additional datapoints would be nice to have. Data analysts can then circle back with your data architect to make it so, or point you toward a data scientist. If they recommend onboarding a data scientist, they are usually the ones to orient the scientist to your data estate.
Our friendly andragogy experts at Coursera offer these bullet-pointed lists of what a data analyst and a data scientist each do.
Out-of-the-Box Data Analytics Tools
So, you probably don’t need a data scientist right out of the gate. But if your data analytics team has been depleted, or if you never had one to begin with, Microsoft Power BI offers some built-in, AI-fueled data visualization tools. You can experiment with these to get a sense of what your existing data may be able to tell you.
Start with Power BI Q&A. It uses natural language processing (NLP) to answer basic business questions asked in plain English. It’s great for business users’ miscellaneous requests, especially when you don’t need to understand the underlying data model. (e.g., How many of X did we sell last month? What was our total revenue for Q3?) Like a search engine, it attempts to autocomplete your request based on context and available/popular options. If you and the AI are struggling to speak the same language, despite Q&A’s NLP capabilities, try enabling Field Synonyms or Teach Q&A to help this tool understand how your team talks about its data.
When you are ready to move beyond asking and answering basic business questions to examining data relationships, Power BI offers two more visualization tools: Decomposition Tree and Key Influencers. Both help you see (literally) which factors are affecting your outcomes and by how much. The decomposition tree aids in root cause analysis, working backward from the problem you’ve identified to finding the underlying reason for it. The tutorials linked here give you a feel for the intuitive, drag-and-drop nature of both features.
These ready-made visualization tools can be a fertile playground for individual business users. Say Joe in Fulfillment has noticed a certain type of product is chronically backordered. Power BI enables him to follow his hunches down various rabbit holes and figure out what’s going on. This is the very definition of self-service data. It’s empowering and may lead to great insights. However, to really move the needle on enterprise-wide KPIs, you need to get more strategic. That starts with clearly defining the business questions you are trying to answer.
Ask a Data Analyst Expert
To assess the big-picture possibilities for your unique data estate, nothing beats a fresh pair of eyes. Of all the questions you could be asking, which ones will most help you reach your business goals? A professional data analyst can see and suggest parts of the data story you may be missing. From there, we help you identify data requirements.
In my experience, the biggest roadblock to a data-driven mindset is company culture. Business users at all levels of the organization get stuck in their ruts: “This is the report I pull every day.” I encourage them to take a step back and see what else they could know and how it might impact business. Together, we may find KPIs you’ve never even thought of.
Beyond helping you ask and answer the right questions, Mind Over Machines data analysts are an important resource for growing your data self-sufficiency. The best way to counteract the data analytics professionals shortage is to grow your in-house talent. (Remember Tally’s Get-Grow-Gather talent acquisition matrix.) Existing employees already know your business and your data. We build their data analysis skills through training and mentorship. CIO.com says both are essential components of assembling a highly effective analytics team.
If you give your data analysis effort the time, attention, and expertise it warrants, I’m betting you won’t need a data scientist, at least not in the immediate while you’re amassing mountains of quality data. But if you still can’t shake that craving for data science, allow me to introduce you to Automated Machine Learning. I’ve heard AutoML in Power BI referred to as machine learning on training wheels. If you want to devote your spare time this year to creating new machine learning models, more Power (BI) to you!
Fadi Zureick was a tactile kid growing up in Columbia, MD. Legos and K’Nex were the only toys for him. Between his fascination with how different objects interact and both his parents being engineers, Fadi had no choice but to earn a degree in mechanical engineering from the University of Maryland. “I love starting from scratch and building something. It’s super gratifying to see what develops over time.”
The transition from mechanical engineering to data analytics platform development came when Fadi was working in the solar energy industry. He was drawn to BI’s capacity for creativity, seeing the whole picture and finding ways to maximize operational efficiency. “A fresh pair of eyes is the most valuable asset I can bring to a client, helping them find answers to questions they haven’t even thought of yet.”
When he’s not making the data tell a story, Fadi is a bit of a Renaissance man, equally comfortable playing soccer and the piano. And now he’s added another skill to his repertoire: parenting. His teething 1-year-old daughter makes up for the sleep she steals with her astounding cuteness, palling around with Charlie the big yellow lab and using baby signs to satiate her endless appetite for reading: “More books, please!”