In 30 seconds
You don't need a statistics or computer science degree to become a data analyst — you need three things, in this order: comfort with spreadsheets and basic statistics, working SQL for pulling data out of real databases, and Python (specifically Pandas) for cleaning and analysing data that's too messy or too large for a spreadsheet. A portfolio of 2-3 real analysis projects matters more to most employers than any single certificate.
Data analyst roles get lumped in with "data science" in a lot of career advice, which scares people off unnecessarily — data science jobs often do want a strong maths/stats background and sometimes a postgraduate degree. Data analysis is a different, more accessible role: taking data that already exists and turning it into decisions someone can act on. It's one of the more realistic tech career changes for someone starting from zero.
The skills that actually matter, in order
1. Spreadsheets and basic statistics — the foundation you probably already half-know
Before touching any code, a working data analyst needs to be fluent in the concepts that spreadsheets teach by default: averages vs medians, what a trend actually shows versus what it looks like, how to spot a misleading chart, and how to structure data so it's usable (one of the most common beginner mistakes is building spreadsheets that look fine to a human but are unusable by any analysis tool). If you're comfortable in Excel or Google Sheets already, you have more of this than you think.
2. SQL — how you actually get the data
If you haven't touched Python fundamentals yet, SQL is still the better starting point for this specific career. Real company data doesn't live in a spreadsheet someone hands you — it lives in a database, and SQL is how you query it. This is often the single most underrated skill in data analysis job postings: it's less exciting than Python or machine learning, but it's the skill that determines whether you can actually do the job day one, because every analysis starts with getting the right data out in the first place.
3. Python and Pandas — for everything a spreadsheet can't handle
Once a dataset is too large, too messy, or needs to be processed the same way repeatedly, spreadsheets stop being the right tool. This is where Python — specifically the Pandas library — takes over: cleaning inconsistent data, merging multiple sources, and automating an analysis you'd otherwise redo by hand every week.
| Tool | What it's actually for | When you reach for it |
|---|---|---|
| Spreadsheets | Quick exploration, small datasets, communicating results to non-technical people | First look at data, final presentation |
| SQL | Getting data out of a real database | Every single analysis that uses company data, not a downloaded CSV |
| Python / Pandas | Cleaning, transforming and analysing data too large or messy for spreadsheets | Repeated analysis, large datasets, anything that needs automating |
What actually gets you hired
A certificate tells an employer you studied the material. A portfolio tells them you can do the job.
For data analysis specifically, that means 2-3 projects where you took a real (or realistic) dataset and produced an actual conclusion — not just a chart, but an answer to a question a business would actually ask: "which product category is underperforming and why," "what's driving the drop in this metric," that kind of thing. This is exactly the gap a project-based course is built to close, versus a self-study certificate alone.
- Pick real or realistic datasets, not toy datasets everyone has already seen a hundred times.
- End with a conclusion, not just a chart. "Here's a visualisation" is step one. "Here's what it means and what I'd recommend" is what makes it a portfolio piece.
- Show your SQL, not just your Python. Most entry-level data analyst job descriptions list SQL before Python — don't let your portfolio skip it.
- Keep it readable. A portfolio piece a non-technical hiring manager can follow is worth more than an impressive-looking notebook nobody outside the field can parse.
Is this career still safe with AI doing more analysis work?
AI tools are genuinely useful for speeding up routine parts of data analysis — writing boilerplate code, suggesting a first-pass chart, summarising a dataset. What they don't replace is the judgement part: knowing which question is actually worth asking, spotting when a dataset is misleading, and translating a number into a decision a business can act on. That judgement is built through doing real analysis, not through a tool doing it for you — which is exactly why the project-based part of learning this skill matters more now, not less.