𝗗𝗮𝘁𝗮 𝗜𝘀 𝗡𝗼𝘁 𝗔𝗻𝗮𝗹𝘆𝘀𝗶𝘀: 𝗖𝗼𝗻𝘁𝗲𝘅𝘁 𝗜𝘀 𝗪𝗵𝗮𝘁 𝗧𝘂𝗿𝗻𝘀 𝗡𝘂𝗺𝗯𝗲𝗿𝘀 𝗜𝗻𝘁𝗼 𝗜𝗻𝘀𝗶𝗴𝗵𝘁

Why understanding the business behind the data is one of the most important skills a data analyst can develop

There is something I have noticed repeatedly while working with data.

Many people can tell you what the numbers are.

Far fewer people can explain what those numbers actually mean.

And that difference is incredibly important.

You can have SQL.

You can have Python.

You can have Power BI.

You can build beautiful dashboards.

You can even write complex queries.

But if you don’t understand the context surrounding the data, you can still arrive at the wrong conclusion.

Data gives you the numbers. Context gives those numbers meaning.

𝗟𝗲𝘁 𝗠𝗲 𝗦𝗵𝗼𝘄 𝗬𝗼𝘂 𝗪𝗵𝗮𝘁 𝗜 𝗠𝗲𝗮𝗻

In a recent security incident analysis I worked on, I used SQL and Power BI to investigate suspicious and fraudulent transactions.The initial numbers were interesting.There were 11,979 transactions in the period analysed.

Out of these, 325 were classified as suspicious, while 71 were flagged as fraudulent.The average transaction amount was approximately ₦72,821.92, while the average income was approximately ₦697.64.

Those numbers give us the starting point.

But they don’t give us the complete story.

That is where many analyses stop.

And that is where I believe real analysis should begin.

𝗧𝗵𝗲 𝗗𝗮𝘁𝗮 𝗦𝗵𝗼𝘄𝗲𝗱 𝗨𝘀 𝗦𝗼𝗺𝗲𝘁𝗵𝗶𝗻𝗴 𝗜𝗻𝘁𝗲𝗿𝗲𝘀𝘁𝗶𝗻𝗴

Looking at the dashboard, North Central recorded the highest number of suspicious transactions, with 76.

South East followed with 71, while South West recorded 60.

At first glance, it is tempting to say:

“North Central has the highest fraud problem.”

But I would not make that conclusion yet.

Why?

Because we are looking at counts. We haven’t considered the context.

• What was the total transaction volume in each region?
• What percentage of transactions were suspicious?
• How many customers does each region represent?
• Were the suspicious transactions concentrated in one particular branch?
• Were specific transaction types responsible for most of the activity?

These questions completely change the analysis.

Imagine North Central processed 10,000 transactions and had 76 suspicious transactions.

Now imagine another region processed only 1,000 transactions but had 50 suspicious transactions.

Looking only at the counts, North Central looks worse.

But when we consider the rate, the story may be completely different.

This is why an analyst should never be too quick to interpret a number without understanding the denominator behind it.

𝗖𝗼𝘂𝗻𝘁𝘀 𝗗𝗼𝗻’𝘁 𝗔𝗹𝘄𝗮𝘆𝘀 𝗧𝗲𝗹𝗹 𝗬𝗼𝘂 𝗥𝗶𝘀𝗸

This principle goes beyond fraud analysis.

Suppose a hospital reports that one department recorded the highest number of adverse events. Does that automatically mean the department has the highest risk?

Not necessarily.

We need to know how many patients were treated there.

Suppose a company says one branch generated the highest number of customer complaints. Does that mean the branch provides the worst customer service?

Again, not necessarily.

Maybe that branch serves 60% of the company’s customers.

This is why context matters.

A number without its denominator can be misleading.

𝗧𝗵𝗲 𝗦𝗮𝗺𝗲 𝗧𝗵𝗶𝗻𝗴 𝗔𝗽𝗽𝗹𝗶𝗲𝘀 𝘁𝗼 𝗖𝘂𝘀𝘁𝗼𝗺𝗲𝗿 𝗦𝗲𝗴𝗺𝗲𝗻𝘁𝘀

The dashboard also showed that Retail customers recorded substantially more suspicious transactions than the other customer segments.

Again, it would be easy to conclude:

“Retail customers are the most vulnerable.”

But before making that statement, I would ask:

• How many Retail customers are there?
• How many transactions do Retail customers generate?
• What is the suspicious transaction rate?
• What is the fraud rate?
• What is the average transaction value?
• What channels are these customers using?

Once again, context determines the meaning of the data.

𝗧𝗵𝗶𝘀 𝗜𝘀 𝗪𝗵𝗲𝗿𝗲 𝗦𝗤𝗟 𝗕𝗲𝗰𝗼𝗺𝗲𝘀 𝗔𝗻𝗮𝗹𝘆𝘁𝗶𝗰𝗮𝗹

One thing I always tell aspiring data analysts is this:

Don’t learn SQL just to write queries. Learn SQL to answer questions.

For example, I used SQL to investigate which customer segments were associated with fraudulent transactions.

SELECT
    dc.segment,
    COUNT(ft.transactionid) AS fraud_count
FROM fact_temp ft
JOIN dim_customers dc
    ON ft.customerkey = dc.customerkey
WHERE fraudflag = 'Yes'
GROUP BY dc.segment
ORDER BY fraud_count DESC;

Technically, this is not an extremely complicated query.

Get Tolulope Emuleomo’s stories in your inbox

Join Medium for free to get updates from this writer.

Remember me for faster sign in

But the query itself is not the most important part.

The question behind it is.

Which customer segments are most affected by fraudulent transactions?

And once we get that answer, we should ask another question.

Why?

That leads to another.

Is the segment genuinely more exposed to fraud, or does it simply have more transactions?

Then another.

What should the business investigate next?

This is where data analysis becomes valuable.

𝗔 𝗗𝗮𝘀𝗵𝗯𝗼𝗮𝗿𝗱 𝗦𝗵𝗼𝘂𝗹𝗱 𝗡𝗼𝘁 𝗝𝘂𝘀𝘁 𝗟𝗼𝗼𝗸 𝗚𝗼𝗼𝗱

I love a good dashboard.

But a dashboard is not successful simply because it looks beautiful.

A dashboard should help someone understand a problem.

It should help the business ask better questions.

It should make patterns easier to identify.

It should support decision-making.

In this particular analysis, the dashboard helped answer questions around:

• Regions with high suspicious activity

• Transaction types associated with suspicious activity

• Channels with high suspicious activity

• Customer segments affected

• States with suspicious transactions

• Fraudulent transaction counts

That is much more useful than simply displaying numbers.

𝗧𝗵𝗲 𝗥𝗲𝗮𝗹 𝗩𝗮𝗹𝘂𝗲 𝗢𝗳 𝗔 𝗗𝗮𝘁𝗮 𝗔𝗻𝗮𝗹𝘆𝘀𝘁

I believe this is one of the biggest differences between someone who knows data tools and someone who can genuinely analyse data.

Knowing SQL is valuable.

Knowing Power BI is valuable.

Knowing Python is valuable.

Knowing statistics is valuable.

But those skills become much more powerful when you understand the business problem behind the data.

A good analyst doesn’t just say:

“Here are the numbers.”

They say:

“Here is what happened.”

Then:

“Here is what may be driving it.”

Then:

“Here is why it matters.”

And finally:

“Here is what we should investigate or do next.”

That is the journey from data to decision.

𝗗𝗮𝘁𝗮 + 𝗖𝗼𝗻𝘁𝗲𝘅𝘁 = 𝗔𝗻𝗮𝗹𝘆𝘀𝗶𝘀

This is the lesson I want aspiring data analysts to take away.

Data tells you what happened.

Context helps you understand why it happened.

Analysis connects the two.

And ultimately:

Insight should lead to action.

Don’t just ask:

“What does the data say?”

Ask:

“What does the data mean?”

Ask:

“What else do I need to know before making a conclusion?”

Ask:

“What business decision could this insight influence?”

That is how you move from simply working with data to actually becoming a data analyst. Because the goal of analytics isn’t to produce more numbers. The goal is to make better decisions with the numbers we already have.

Don’t just analyse the data.

Understand the story behind it.

About the Author

Tolulope Emuleomo — #NaijaDataProfessor

Senior Data & AI Solutions Consultant | Analytics Engineer | Healthcare Intelligence | Enterprise Decision Intelligence | Machine Learning

I help professionals and organizations turn data into actionable business intelligence through analytics, AI, data engineering, and practical training.

📧 dataprofessor290@gmail.com

Linkedin: Tolulop Emuleomo

#DataAnalytics #DataAnalyst #SQL #PowerBI #BusinessIntelligence #DataScience #FraudAnalytics #DataStorytelling #NaijaDataProfessor