Thinking Like a Data Analyst

The Analyst Mindset

Having the right tools is only part of being a data analyst. The bigger part is developing the analytical mindset -- the habit of approaching problems with structured curiosity, scepticism, and clarity.

The best data analysts are not just people who can write SQL or use Excel well. They are people who ask better questions, challenge assumptions, and communicate their findings clearly.

---

The Data Analysis Process

Good analysis follows a repeatable process. It might look slightly different across teams, but the core steps are almost always the same.

Step 1: Define the Question

Before touching any data, be clear about what you are trying to find out.

Bad question: "Show me the sales data." Good question: "Which product categories drove the most revenue growth in the Eastern region during Q3, compared to Q2?"

A specific, actionable question guides every decision you make in the analysis.

Step 2: Identify the Data You Need

Once you have a clear question, determine what data would help answer it:

• Which tables or datasets contain the relevant information?
• What time period do you need?
• What level of granularity (daily, monthly, per customer)?
• Is the data available? Is it reliable?

Step 3: Collect and Explore the Data

Pull the data using SQL, export from a tool, or load a CSV. Before jumping into analysis, explore the data first:

• How many rows? How many columns?
• What are the data types?
• Are there missing values?
• Do the values look reasonable? (A product price of 0 or a customer age of 500 are red flags)

This initial exploration is called Exploratory Data Analysis (EDA).

Step 4: Clean the Data

Real-world data is messy. Common problems you will encounter:

• Missing values: A customer record with no email address. Do you delete the row? Fill with a default? It depends on your analysis.
• Duplicates: The same transaction recorded twice. This distorts totals.
• Inconsistent formatting: "Lagos", "LAGOS", "lagos" are the same city but a computer treats them as different.
• Outliers: A single transaction for 50 million NGN when all others are under 100,000. Is this legitimate or a data entry error?

Data cleaning often takes 50-80% of an analyst's time. It is unglamorous but critical.

Step 5: Analyse the Data

Now the interesting work begins. Depending on your question, you might:

• Calculate summary statistics (mean, median, mode, standard deviation)
• Group data by category and compare
• Look at trends over time
• Correlate two variables
• Segment customers into groups

Step 6: Visualise the Results

Charts and graphs make patterns visible and insights accessible to non-analysts. Choose the right chart for your data:

Step 7: Communicate Your Findings

The final and often most underrated step. Your analysis only creates value if the right people understand it and act on it.

A good analyst can explain findings to a non-technical audience without dumbing them down. Structure your communication:

1. Lead with the most important insight (the "so what")
2. Show the evidence that supports it
3. Acknowledge limitations or caveats
4. End with a recommendation or next step

---

Avoiding Common Analytical Mistakes

Correlation is not causation

Just because two things change together does not mean one causes the other. Ice cream sales and drowning rates both increase in summer -- but ice cream does not cause drowning. Both are caused by hot weather.

Survivorship bias

Looking only at the data that survived a selection process. If you analyse only successful businesses to understand what makes companies succeed, you ignore all the companies that failed doing the same things.

Cherry-picking data

Selecting only the data that supports a predetermined conclusion. Good analysts present the full picture, even when it challenges their hypothesis.

Ignoring sample size

An average based on 5 responses is very different from one based on 5,000. Small samples produce unreliable results.

---

Asking Better Questions

The questions you ask determine the quality of your analysis. Practice reframing vague questions into specific, answerable ones:

---

Your First Analysis Framework

When you are given a new analytical task, work through these questions:

1. What decision will this analysis inform? (Understanding the end goal keeps you focused)
2. What data do I have access to? (Reality-check before committing)
3. What would a meaningful result look like? (Set the bar before you start)
4. What could make my findings wrong? (Anticipate objections and data quality issues)
5. Who needs to see this, and in what format? (Tailor your output to your audience)