Tell the analyst apart from data scientists, engineers, and BI roles.
Goal: Tell the analyst apart from data scientists, engineers, and BI roles.
In her second week at Perch, Nadia sat in a meeting where two job titles flew around the room as if they meant the same thing. Someone asked the "data team" whether the company should expect a sales bump next quarter. Nadia felt a small panic: was that her job now?
It wasn't, and the line between the two roles turned out to be cleaner than the meeting made it sound.
A Data Analyst answers what happened and why. Nadia describes the past and the present clearly: "Repeat purchases dropped 12% last quarter, almost all of it among customers who bought a single desk and never came back." She's holding up a mirror to the business.
A Data Scientist leans toward what will happen and what to do about it. They build predictive models and machine learning: "Here's a model that scores each new customer on how likely they are to buy again in 90 days, so marketing can target the wobbly ones." They're holding up a forecast.
The toolkits overlap but tilt differently. Nadia works in spreadsheets, SQL, and visualization. A data scientist adds heavier statistics and real programming, usually Python or R. That extra layer of code and math is exactly why the scientist role is a harder first step for a career-changer, and why the analyst role is the one Nadia could actually start in.
The analyst explains the past so the business stops guessing. The scientist predicts the future so the business can place a bet.
A week later, Nadia tried to pull "revenue by product category" and hit a wall: the orders table had no category column at all. She'd done nothing wrong. The column simply didn't exist yet in the warehouse.
So she messaged Tom, Perch's data engineer.
A Data Engineer is a software-engineering role, not an analysis role. Tom builds and maintains the plumbing: the pipelines that pull data out of Perch's website, payment system, and ad platforms, and the warehouse where it all lands so people like Nadia can query it. When a column is missing or a number looks wrong at the source, Tom is who you ask. He works in tools like Airflow (which schedules the pipelines), Spark (which processes big batches of data), Kafka (which moves data in real time), and a lot of Python.
Nadia uses the plumbing. Tom builds it.
Picture a city's water system. Tom lays the pipes and keeps the reservoir clean and full. Nadia turns on the tap and decides what to do with the water. Both jobs matter, and neither could do the other's well on day one.
This distinction is the one career-changers get wrong most often, so it's worth saying plainly: data engineering is coding-heavy. It is not the no-code pivot some bootcamp ads imply. Knowing it exists helps Nadia understand where her data comes from, but it is not the door she's walking through.
Here's where the desk-category problem went next. Tom got the raw category data flowing into the warehouse, but it arrived messy: "Desk," "desks," and "DESK-001" all meant the same thing, and revenue was scattered across three tables that had to be stitched together before anyone could chart it. That stitching-and-cleaning-in-the-warehouse work is a job of its own now.
That job is the Analytics Engineer, a newer hybrid role that emerged around 2018. It sits squarely between the analyst and the engineer. They don't build the pipes (that's Tom) and they don't write the final business report (that's Nadia). They own the layer in the middle: transforming and modeling raw warehouse data into clean, trustworthy tables everyone else can rely on.
Their signature tool is dbt (data build tool), the tool that effectively created the role. With dbt, someone comfortable with SQL can define a transformation once, put it under version control, and document it, so "what counts as a repeat customer" lives in one agreed-upon place instead of being re-invented in every analyst's query.
For Nadia specifically, analytics engineering is a very common next step up from analyst. It builds on SQL she's already learning, adds version control and documentation, and pays well. Priya, her mentor, started exactly where Nadia is now.
Nadia opened a job board out of curiosity and counted six different "analyst" titles in one scroll: Data Analyst, BI Analyst, Reporting Analyst, Business Analyst, Product Analyst, Marketing Analyst. Her stomach sank a little. Six jobs to learn?
No. Mostly one job wearing six name tags, with two real exceptions to watch for.
A BI Analyst (business intelligence) leans toward building and maintaining the dashboards and recurring reports the business watches over time, usually in Tableau, Power BI, or Looker. If Marcus checks a "sign-ups this week" dashboard every Monday, a BI analyst probably built and maintains it.
A Business Analyst is the title to read carefully, because it can mean something quite different. It often leans toward processes and requirements rather than data, closer to a junior product or project role: gathering what a team needs, writing it up, coordinating. Some Business Analyst jobs barely touch a spreadsheet.
"Product Analyst" and "Marketing Analyst" are the easy ones. They just mean a Data Analyst pointed at a specific part of the business: product usage for one, ad and campaign performance for the other. Nadia, sitting with Marcus's marketing questions all day, is doing marketing-analyst work under a data-analyst title.
The thread tying almost all of them together: the core skills are shared. Spreadsheets, SQL, statistics, and visualization show up in nearly every one of these jobs. Learn those four and you've qualified yourself for a wide spread of titles at once.
So if "Data Analyst" and "Business Analyst" and "BI Analyst" all blur together, how does Nadia avoid applying to the wrong thing? Priya gave her one rule.
Don't trust the title at the top. Read the responsibilities underneath it.
Two postings can both say "Data Analyst" and describe different jobs. One lists SQL, build dashboards in Tableau, weekly KPI reporting — that's classic analyst work, the kind Nadia wants. Another lists gather stakeholder requirements, document processes, manage project timelines — that's really a business/project role with an analyst label, and Nadia would be miserable and underqualified in it for reasons that have nothing to do with data.
The tell is the verbs. Query, analyze, visualize, report point at data work. Gather, coordinate, document requirements point at process work. Build pipelines, maintain the warehouse point at engineering, and build models, predict, deploy point at data science.
Once Nadia could decode a posting this way, the job board stopped being scary. She wasn't facing six unknown careers. She was reading six descriptions and sorting them into "yes, that's me," "that's me in two years," and "that's a different job sharing my word."
Perch's VP of Operations, Dana, sets a goal for the year: more repeat buyers. One goal, and watch how it lands on four different desks.
Tom, the data engineer, makes sure the raw material exists. He builds a pipeline that pulls every order from the website and payment system into the warehouse each night, reliably, so nobody's working off stale or missing data. No pipeline, no analysis.
The analytics engineer turns that raw order data into a clean customer_orders table with one agreed definition of "repeat customer" (bought more than once), built in dbt, version-controlled, documented. Now everyone counts repeat customers the same way.
Nadia, the data analyst, queries that table to answer what happened and why: repeat rate is 18%, and it's much lower for customers who bought only a desk than for those who bought a sofa. She brings Marcus and Dana the finding in plain language, with one chart.
A data scientist (Perch hires one later) takes it further into what will happen: a model that flags, the day someone buys, how likely they are to come back, so marketing can nudge the at-risk ones early.
One goal. Four roles. Four different definitions of "done." Being able to look at any data project and name who owns which piece is exactly the map this topic hands you, and it's the map that tells a career-changer where the realistic first door is.
Open a real job board and search "Data Analyst." Pick two postings and ignore the titles completely. Read the responsibilities and sort each into one bucket: classic analyst (query, visualize, report), engineering (pipelines, warehouse), science (models, prediction), or business/process (requirements, coordination). Write down the verb that gave each one away. That's the exact move experienced candidates use to aim instead of spray.
Preparing your quiz…