Thinking Like An Analyst
Yesterday, I was stuck in traffic. Not unusual for those who live in Hampton Roads, but this particular instance of forced pausing allowed me to think about something. A few months ago I discussed why I was taking the CompTIA Data+ exam. Back then I was absolutely convinced that I needed as many certifications as possible to prove I could be a data analyst. I've come a long way since then, taking and passing the Data+ Certification even, and have since understood it's not the papers that make you an analyst, but how you look at data and use it to shape things. There are a lot of analysts out there who don't have certificates or masters degrees, and handle their data just fine. So what is the actual next step? I realized that in this I've made this transition. I've gone from “I need to learn all the SQL terms, data science concepts, the analysis techniques, all the EVERYTHING” to a desire to learn how and why it all gets put together.
In short: I want to learn how to think like an analyst. Let me show you what I mean with a recent project that taught me this lesson the hard way.
The Cart Before The Horse Approach
There’s data here
So basically I had a job to do. I was given a question and needed to perform a marketing analysis. But, I did the one thing that early data analysts do which is:
I ran with the data before finding out the question.
Lots of new data analysts (and even experienced ones) do this from time to time. I got the question from the client:
Client wants marketing analysis for targeting small businesses within 100-mile radius. Focus on businesses with <1000 employees. Interested in earnings/profit data and business categories to help with pricing strategy.
I took that literally, instead of digging down or looking at what she actually needed for the type of company she has and where she was in the process. This is what I did:
Had a list of questions generated but didn’t exactly prompt around finding out what the business needed
Attempted to find revenue/earnings from businesses in the area
Attempted to find business categories to market to
Jumped straight into a dashboard without the question to answer
Didn’t create any sort of wireframe or sketches for the dashboard. I do this with websites, basically sketching out what goes where and getting ideas together.
Got frustrated a lot
Ended up having to stop after 2 weeks of frustrating searching and actually do the website she required because I was taking so long to make the report that her customers started demanding a website
What Should Have Happened
What I should have done, was instead begin with finding out what state the business was in as a whole. Where were they in the entire process? My client had transitioned from brick & mortar full restaurant to online catering with an Etsy page selling spices. This was a unique case due to the upset of COVID and moving locations and demanded more research and better testing with a proper scope.
I should have dug deeper to clarify the actual question. What did they want? I heard them say “marketing analysis” but what does that mean for them exactly? Why did they need it? What information aside from just “people in the area” were they trying to get? And, most importantly, was the information they were looking for going to make them successful? My client now needed information about who was in the area that would want her services, who were any competitors, and where did her prices fall in the average from everyone else’s.
Sometimes Clients Don’t Know What They Need
Sometimes they do, sometimes they don’t. Thinking like an analyst means, I need to figure out what kind of data they actually need, which comes down to what kind of question they actually need to ask either me or themselves. And it’s alright if you don’t know.
From what I’ve seen so far, if you get into the data with a question, and it either doesn’t make sense or you realize it’s not going to provide them with much of anything, stop. Refine, readjust, go back to the drawing board.
How an Analyst Actually Thinks About This
Here's what I learned: real analysts don't start with data sources or dashboard wireframes. They start with conversations. Here is the process I learned that works better than just jumping in head first:
Business Context Discovery - "Instead of jumping into data, I should have asked: What kind of catering business is this? Are you B2B or B2C? What stage are you in - just starting online or trying to scale?"
Clarify the Real Question - "When you say target small businesses, do you mean businesses as customers for catering?" "What type of marketing are you planning? Direct outreach, advertising, referral programs?" "What do you hope to do differently after this analysis?"
Understand the Business Model - Show how this would have revealed the B2B vs B2C split
B2C: Individual customers buying cupcakes/cakes
B2B: Businesses booking catering services
Each requires completely different analysis approaches. I figured this, but I didn’t take it into account.
Define Success Metrics
I did a little bit of this thought, mostly what format does the client need results in. I was going to provide the spreadsheet, but also the dashboard in a .PDF format with a writeup of my findings. Maybe a recommendation going forward.
What does a successful analysis deliverable look like?
How will the client use this information?
What format does she need the results in?
Data Reality Check - "What data do you actually have access to? Customer emails? Past orders? Reviews?"
I asked some questions, but I didn’t really figure these: Before promising revenue data analysis I should have asked:
"What data sources do you already have access to?"
"Have you done any customer research before?"
Research what's actually available vs. what would be ideal
Using the Wrong Approach Can Lead to Frustrations
Finding the data was difficult because I didn’t know what I was looking for. I had an inkling, but I didn’t really know what I was going to be able to get. Not being able to find revenue really threw me off by several weeks, and honestly I should have figured as much. Businesses are not going to release their profits to just anyone, especially one small data analyst looking for information. Then, I searched Yelp for data, finding out that the API allows me only to pull up to 200 entries without paying $200 for it. Understandable I guess, but, disappointing regardless. Pulling that data into Power BI was very interesting, and yet not really usable.
The Dashboard Attempt
Without actually having the question to answer and without doing a wireframe for the dashboard, actually creating this was difficult. I didn’t have a plan of action, a plan of design, a plan of anything. It wasn’t going to work regardless of what I did and would have taken me longer to accomplish trying to work through it a little bit at a time.
The Bigger Picture
As I was working through the CompTia Data+ Certification, I began to want more than just learning phrases and what chart types to use with what.
This project taught me that the difference between a data person and a data analyst isn't the tools they know. It's whether they start with the problem or the data. All the SQL knowledge and Power BI skills in the world won't help if you're solving the wrong question. And starting with the wrong question can lead you down a path of wasted time and client trust-like when I had to abandon my analysis after two weeks to build her website instead.
The Data+ certification helped me understand the technical foundation, but learning to think like an analyst? That only comes from making mistakes, reflecting on them, and adjusting your approach. And that's the kind of learning you can't get from any exam.
