Becoming a Data Analyst
What data analysts do between the spreadsheet and the decision, which skills matter most, and how to enter the field in Botswana.
What the job actually involves
Data analysis is the work of turning a question someone asked badly into a question that can be answered, then answering it in a form that changes a decision.
Most of the job is not modelling. It is finding the data, discovering it is wrong, working out how it is wrong, and deciding whether the answer survives that. Analysts who enjoy this part do well; those who wanted machine learning tend not to stay.
In Botswana the largest employers of analysts are the banks and insurers, the mobile operators, government ministries and the parastatals, the mining companies and their service contractors, and the development organisations. Monitoring and evaluation work in the development and public health sector is a distinctive feature of this market, and it hires analysts under titles that do not contain the word analyst.
A typical week
A representative week: a day writing SQL against systems that do not agree with each other, a day cleaning and checking, a day building or updating something someone will look at weekly, and the rest in conversations working out what was actually being asked.
Skills required
SQL is the skill. People entering the field consistently overweight Python and machine learning and underweight the ability to get correct numbers out of a database that was not designed for questions.
The second most valuable and least practised skill is written communication: saying what the number means, what it does not mean, and what you would do about it.
Typical career path
- 01Junior analystyears 0-2
You answer defined questions and maintain existing reporting. The job is accuracy and turnaround, and learning where the data is buried.
- 02Analystyears 2-4
You own a domain and the reporting around it. You start being asked what the business should look at, not just what it asked for.
- 03Senior analystyears 4-7
You work on questions with commercial consequences and are trusted to say when the data cannot answer them.
- 04Analytics lead, data scientist or analytics engineeryear 7 onward
Three genuine forks: leading analysts, moving into modelling, or moving into the pipelines and infrastructure that produce the data.
Education options
A strong fit, and the most common background among analysts here. Gives you the statistical grounding that is hard to acquire piecemeal. Does not teach SQL or the tooling, which you will still have to learn.
4 years
Works well because the core skill is learnable and testable. Build a portfolio of real analyses on public data. Fails for people who start with machine learning courses instead of databases.
9-18 months · Low
Certificates in specific tools help you past keyword screening and teach you the tool. They do not substitute for a portfolio showing you can reason about a business question.
2-6 months
Tools
SQL · Excel · Power BI · Tableau · Python with pandas
Portfolio requirements
An analytics portfolio should show a question, the work, and a recommendation. The recommendation is the part most candidates leave out and the part employers care about.
- Two or three analyses of real, public datasets, ideally from Statistics Botswana or the Bank of Botswana
- The messy parts included: what was wrong with the data and how you handled it
- A clear statement of what you would tell a decision-maker to do
Entry-level opportunities
Entry-level analyst roles here are commonly titled reporting analyst, MIS analyst, business analyst or monitoring and evaluation officer, and are concentrated in banking, telecommunications, insurance, government and the development sector. Job titles vary enough that filtering by title alone will hide most of the market.
Career progression
The clearest progressions are into analytics engineering, into data science, or into a commercial role where the analysis becomes ownership of the outcome. Analysts who move into product or commercial roles often do very well, because they arrive knowing what is actually true about the business.
Interview preparation
Expect a SQL test. It is the most reliable predictor employers have and it is used almost everywhere.
- Practise joins, aggregation and window functions until they are automatic
- Expect a case question: how would you find out whether X is working
- Say what the data cannot tell you; candidates who caveat well stand out
Recommended resources
- Statistics Botswana publication · freePublic Botswana data to build a portfolio on rather than another Titanic dataset.
Frequently asked
Do I need to know Python to become a data analyst?
Not to start. SQL and a spreadsheet will get you into most junior analyst roles here, and they are what interviews test. Python becomes valuable once the analyses you want to run outgrow a spreadsheet, and it is the natural next thing to learn after SQL rather than before it.
What is the difference between a data analyst and a data scientist?
An analyst answers questions about what happened and what to do about it, mostly with SQL, spreadsheets and visualisation. A data scientist builds models that predict or classify, usually with a stronger statistics and programming background. Analyst roles are far more numerous here, and many data scientists started as analysts.
Which industries hire data analysts in Botswana?
Banking, insurance and the mobile operators hire the most, alongside government ministries and the parastatals. The development and public health sector is a significant second source, hiring for monitoring and evaluation work, often under titles that do not say analyst. Mining companies and their service contractors are a third.
Can I become a data analyst without a degree?
Yes, though it is harder than in software engineering because more analyst employers are large institutions with formal screening. A portfolio of real analyses plus a demonstrable command of SQL is the route that works; expect to apply to more roles than a degree-holding candidate would.