Webbo3 Data Analytics · Cohort 2 · Excel Track · Week 1, Live 1
What Is Data Analysis?
Before we touch a single cell, we answer the question that makes every later skill make sense: what is all this for?
By the end of this class you can
- Explain what data analysis is, in one sentence of your own
- Tell data, information, insight and a decision apart
- Name the types of data and where data comes from
- Match a business question to one of the four types of analytics
- Walk the full workflow from raw data to a decision
- Say which tool (Excel, SQL or Power BI) solves which problem
Two people sell phone accessories in Computer Village, Ikeja. Same street, same suppliers, same customers walking past. By December, one of them is restocking with confidence and the other is stuck with cartons of chargers nobody wants. The difference is not luck. It is what they did with their records.
The Story: Two Shops, One Street
Bayo restocks by feel. When a customer asks for something he has run out of, he remembers it. When the shop is busy, he assumes things are going well. At the end of each month he counts the cash, and if there is more than last month, he is happy.
Adaeze writes every sale in a notebook: the date, the item, how many, the price. Every Sunday she spends twenty minutes adding it up. In March she notices that power banks sell three times faster in the weeks NEPA is bad, so she stocks up when the outages start, not after. She notices that her Friday customers buy more expensive items, so she puts the good earphones at the front on Fridays. She notices that one supplier's chargers come back as faulty twice as often, so she stops buying from him.
Bayo has the same sales. The difference is that Adaeze asks questions of hers and lets the answers change what she does. That habit, done carefully and at scale, is data analysis.
1. What Is Data Analysis?
Definition
Data analysis is the process of examining data to answer a question, so that someone can make a better decision.
Three words in that sentence carry all the weight. Examining: you look carefully, and you check what you are looking at before you trust it. Question: analysis always starts with something someone wants to know; without one, you are just staring at numbers. Decision: the point is not the number, it is what somebody does differently because of it.
Real-Life Scenario
A doctor does not prescribe the moment you walk in. She asks questions, takes your temperature and blood pressure, maybe sends you for a test, compares the results with what normal looks like, and only then decides on a treatment. The measurements are the data. The diagnosis is the analysis. The prescription is the decision. A doctor who skipped the measurements would be guessing, and so is a business that skips the analysis.
Data, Information, Insight, Decision
Beginners often use these four words as if they mean the same thing. They are four steps up a ladder, and an analyst's job is to climb it. Here is the ladder on the TechPlug file you will open in the next live class:
| Step | What it is | TechPlug example |
|---|---|---|
| Data | Raw facts, one record at a time. On its own it tells you almost nothing | TP-0892, 2024-03-03, Kano, Thermocool AC 1.5HP, 3 units at N380,000 |
| Information | Data organised so it answers "what" or "how much" | Accessories made 217 sales in the first three months of 2024, worth N24,541,000 |
| Insight | Information compared or explained, so it says something you did not know | Accessories are 21% of all sales but only 2% of the money. Laptops are 19% of sales and 32% of the money |
| Decision | What someone does because of the insight | When the restock budget is tight, laptops are reordered first; a sales push on accessories can never move the total much |
Notice that the data row never changes. What changes is the question you ask of it. A spreadsheet full of rows is not analysis, in the same way a pile of bricks is not a house.
Write your own definition of data analysis in one sentence, without looking at the one above. Then check it: does it mention a question? Does it mention a decision? If not, rewrite it until it does. You will need this sentence for your Virtual 1 class.
Picture a ladder leaning on a wall. The bottom rung is a single receipt. The top rung is a manager changing an order. Every step up is you asking a better question of the same receipts.
2. Types of Data
Definition
Data comes in different shapes and different kinds. The shape decides which tool can read it. The kind decides what you are allowed to do with it: you can add up prices, but adding up branch names makes no sense.
Real-Life Scenario
Think of a market woman's records. Her sales book has neat columns: date, item, quantity, price. Her WhatsApp has voice notes from customers placing orders. Her phone gallery has photos of receipts. All three are data about her business. Only the first can be totalled at the end of the day without someone sitting down to listen, read and type.
Types by Shape
| Type | What it looks like | Example | Can Excel read it directly? |
|---|---|---|---|
| Structured | Rows and columns. Every row has the same fields | TechPlug Sales Data, a bank statement, a class register | Yes |
| Semi-structured | Labelled, but not in a neat table | An export of a WhatsApp chat, data from a payment app | With some work |
| Unstructured | No fixed fields at all | Voice notes, photos of receipts, product reviews | No. It has to be turned into rows first |
Types by Kind: Numbers and Categories
Inside a table, each column holds one kind of value. These are the kinds you will meet in every dataset in this course:
| Kind | What you can do with it | TechPlug columns |
|---|---|---|
| Numerical (quantitative): a measured or counted amount | Add, average, find the biggest and smallest | Quantity, Unit_Price |
| Categorical (qualitative): a label that puts a record in a group | Count, group, filter. Never add | Product_Name, Category, Staff_Name, Branch, Customer_Type, Payment_Method |
| Date and time: a moment | Sort in time order, group by month, measure the gap between two dates | Date |
| Identifier: a unique label for one record | Find a record, check for duplicates. Never add or average | Transaction_ID, Product_ID, Staff_ID |
A phone number, a bank account number and a Transaction_ID are written with digits, but they are labels, not amounts. Adding two phone numbers gives you nonsense. Ask: would it ever make sense to take the average of this column? If not, it is a label, whatever it looks like.
Two finer splits are worth knowing. Numerical data can be discrete (counted in whole steps: Quantity is 1, 2 or 3, never 2.5) or continuous (measured, with fractions: litres of fuel, temperature). Categorical data can be nominal (no order: Lagos is not "more" than Kano) or ordinal (a natural order: Low, Medium, High).
Where Data Comes From
| Source | Meaning | Examples for a Nigerian business |
|---|---|---|
| Internal | Created by the business itself | Till and POS records, Paystack or Flutterwave transactions, stock counts, staff attendance |
| External | Created by someone else | Inflation figures from the National Bureau of Statistics, exchange rates from the Central Bank, a competitor's published prices |
| Primary | Collected by you, for your question | A survey you send your customers, counting people who walk past the shop |
| Secondary | Collected by someone else, for their own reasons | A government report, a market research study you download |
TechPlug's Sales Data is internal, secondary to you (TechPlug's tills collected it for accounting, not for your question) and structured. Most business data you will analyse in your career looks like that.
For each TechPlug column below, say whether it is numerical, categorical, a date or an identifier: Transaction_ID, Date, Branch, Quantity, Unit_Price, Payment_Method. Then name one column a business might add that would be ordinal.
Check your answer
Identifier, date, categorical, numerical, numerical, categorical. An ordinal column could be a delivery speed (Standard, Express, Same Day) or a customer rating (Poor, Fair, Good, Excellent).
Picture the market woman's three piles: the neat sales book, the phone full of voice notes, the envelope of crumpled receipts. Only one pile is ready to add up. An analyst's first job is often turning the other two into the first.
3. The Four Types of Analytics
Definition
Every analysis answers one of four kinds of question. They go from looking back to looking forward, and each one is harder, and more valuable, than the one before it.
Real-Life Scenario
Your phone battery. "It is at 12%" is what happened. "It drained fast because the screen was on full brightness all afternoon" is why. "At this rate it dies by 6pm" is what will happen. "Turn on battery saver and lower the brightness now" is what you should do. Four questions, one battery.
The Four Types
- Descriptive: What happened?
- Diagnostic: Why did it happen?
- Predictive: What is likely to happen next?
- Prescriptive: What should we do about it?
Type 1: Descriptive Analytics
Descriptive analytics summarises what has already happened: totals, averages, counts, rankings. It is where every analysis starts and, honestly, most of what businesses ask for. Weeks 2 to 4 of this track are mostly descriptive analytics, done properly.
TechPlug example: "Which branch sold the most from January to March 2024?" Answer: Port Harcourt, with N148,290,000 from 122 sales. The smallest was Abuja, with N90,260,000.
Type 2: Diagnostic Analytics
Diagnostic analytics digs into a result to find out why it happened. You take a number that surprised someone and break it down: by branch, by product, by customer, by day, until the cause shows itself.
TechPlug example: February was the weakest month: N337,417,000, against N368,006,000 in January and N361,920,000 in March. A manager asks whether something went wrong in February. Break each month down by day: January N11,871,161 a day, February N11,635,069 a day, March N11,674,839 a day. February 2024 had 29 days, January and March 31. Per day, the three months are almost identical. February was not a weak month; it was a short one.
That difference matters. If you thought February was weak, you might start a February promotion next year, or blame the staff. Knowing it was the calendar, you change nothing and save the money.
Type 3: Predictive Analytics
Predictive analytics uses patterns in past data to estimate what is likely to happen. It never gives certainty, only a better guess than instinct. This track does not build forecasts, but you should recognise the question when a manager asks it.
TechPlug example: "What will April bring in?" TechPlug averaged about N11,727,023 a day across January to March. If April is like those months, its 30 days would bring roughly N352,000,000. Three months is a short history, so this is a rough guide, and it gets better as the business keeps records for longer.
Type 4: Prescriptive Analytics
Prescriptive analytics recommends an action. It combines what happened, why, and what is likely next into "here is what we should do". This is where an analyst earns trust: a recommendation that turns out right.
TechPlug example: laptops bring in 32% of all the money from 19% of the sales. When the restock budget is tight, reorder laptops first, because a laptop missing from the shelf loses far more money than a missing mouse.
All Four Together
| Type | The question | Looks | Difficulty |
|---|---|---|---|
| Descriptive | What happened? | Back | Lowest |
| Diagnostic | Why did it happen? | Back, deeper | Low to medium |
| Predictive | What will happen? | Forward | Medium to high |
| Prescriptive | What should we do? | Forward, into action | Highest |
Jumping straight to prescriptive. "We should close Abuja" sounds bold, but without the descriptive and diagnostic steps underneath it, it is an opinion, not analysis. Maybe Abuja opened in February. Maybe its till was down for two weeks. Earn the recommendation by doing the first two steps properly.
Label each question as descriptive, diagnostic, predictive or prescriptive. (a) How many wholesale orders did Kano take? (b) Should we hire another sales rep for Abuja? (c) Why did Enugu sell less in the second half of the year? (d) How many laptops will we sell in December?
Check your answer
(a) descriptive, (b) prescriptive, (c) diagnostic, (d) predictive.
Picture the battery icon again: 12% (what happened), the bright screen (why), 6pm (what will happen), battery saver (what to do). Every business question is one of those four.
4. The Analytics Workflow
Definition
The analytics workflow is the order every analysis follows, from a raw file to a decision. Each stage depends on the one before it, so a mistake early on travels all the way to the end.
Real-Life Scenario
Making jollof rice. You buy the ingredients (raw data). You wash the rice, pick out stones, chop the peppers (clean). You cook (analyse). You dish it nicely onto a plate (visualise). Your guests eat and decide whether to come back next Sunday (decide). Skip the washing and it does not matter how well you cook: someone bites a stone.
The Five Stages (and the Question Before Them)
| Stage | What happens | Where you learn it |
|---|---|---|
| 0. The question | Agree what the person needs to know, and what they will do with the answer | Every week |
| 1. Raw data | Find the data and get it into your tool without damaging it | Week 1 |
| 2. Clean | Remove duplicates, deal with blanks, fix spelling, case and formats | Week 2 |
| 3. Analyse | Calculate, group, compare, label, look up | Weeks 3 and 4 |
| 4. Visualise | Turn the answer into a chart or a dashboard someone can read in seconds | Weeks 4 and 5 |
| 5. Decide | Say what the numbers mean in plain business language | Weeks 3 to 5 |
Overall Example: One Question, All Five Stages
A TechPlug manager asks: "Which category should the restock budget go to first?" Here is how that question moves through the workflow:
- Raw data: you open the Sales Data sheet: 1,070 rows, one per sale.
- Clean: you find twenty sales entered twice, and names typed with stray spaces, such as " Enugu" and "Enugu", which Excel treats as two different branches. You fix both first. (You will do exactly this in Week 2, following the Cleaning Process tab.)
- Analyse: you total the money by category. Laptops brought in N344,720,000; Accessories N24,541,000, despite having more sales.
- Visualise: a bar chart, categories sorted from most money to least, titled with the finding: "Laptops bring in 32% of all revenue".
- Decide: you write one sentence: "Restock laptops first; accessories sell often but cannot move the total."
If you skipped the cleaning stage in that example, the twenty duplicated sales would be counted twice, so some categories would look bigger than they really are. The analysis would be wrong before you had drawn a single chart, and nothing on screen would warn you.
Picture the jollof pot. Nobody praises the washing of the rice, but everyone notices the stone.
5. The Tools: Excel, SQL and Power BI
Definition
This programme teaches three tools. They overlap, but each is best at a different stage of the workflow and at a different size of data.
Real-Life Scenario
A kitchen knife, a blender and a serving tray. You can chop with all three if you are desperate, but each is built for one job. Analysts are the same: the skill is knowing which tool to reach for.
The Three Tools
| Tool | What it is | Best at | Limits |
|---|---|---|---|
| Excel | A spreadsheet: data in a grid you can see and touch | Small and medium data, quick answers, cleaning by hand, one-off reports, sharing a file | One sheet holds at most 1,048,576 rows; it gets slow well before that; easy to break by accident |
| SQL | A language for asking questions of a database | Millions of rows, data that lives in a company's systems, answers that must be repeatable | You do not see the data as you work; you write questions and read answers |
| Power BI | A tool for building interactive dashboards | Connecting several tables, refreshing automatically, dashboards that many people use online | Needs clean, well-shaped data to work well |
Which Tool for Which Stage?
| Stage | Excel | SQL | Power BI |
|---|---|---|---|
| Raw data | Open a file | Pull from a database | Connect to files and databases |
| Clean | Yes, by hand and with formulas | Yes, at scale | Yes, with Power Query |
| Analyse | Yes: formulas, PivotTables | Yes: totals and groups over huge tables | Yes: measures |
| Visualise | Yes: charts, a simple dashboard | No | Yes, its main strength |
| Decide | You, with any of them | You | You |
Notice the last row. No tool decides. The tool makes the answer faster to get; the analyst makes it mean something. That is why this track starts with Excel: everything you learn here (cleaning, conditions, grouping, lookups, clear charts) is the same thinking you will use in SQL and Power BI, in a tool where you can see every cell.
Which tool would you reach for first? (a) A 200-row list of your class's quiz scores. (b) 30 million mobile money transactions in a bank's database. (c) A sales dashboard 40 managers check every morning.
Check your answer
(a) Excel. (b) SQL. (c) Power BI.
Picture the knife, the blender and the tray on one counter. Excel is the knife you will use every single day.
Practical Exercises
Exercise 1: Very Easy
Name the four types of analytics and write the question each one answers.
Exercise 2: Simple Application
Classify every one of the twelve Sales Data columns as numerical, categorical, date or identifier. (The column list is on the Start Here page.)
Exercise 3: Combination
Write one TechPlug question for each type of analytics. Make them questions a real manager might ask.
Exercise 4: A Small Real-World Problem
A provisions shop owner tells you, "My profit fell last month and I don't know why." Write the five workflow stages for her problem: what raw data you would ask for, what you would expect to clean, what you would calculate, what one chart you would show her, and what kind of decision it could lead to.
Checkpoint
Q1. What is data analysis, in one sentence?
Examining data to answer a question so that someone can make a better decision.
Q2. What is the difference between information and insight?
Information says what or how much ("Accessories made 217 sales"). Insight compares or explains it so it tells you something new ("Accessories are 21% of sales but only 2% of the money").
Q3. Is Transaction_ID numerical data? Why or why not?
No. It is an identifier. It labels one sale; adding or averaging Transaction_IDs means nothing. It is also why a Transaction_ID that appears twice is a warning: the same sale was entered twice.
Q4. "Why was February the weakest month?" Which type of analytics is that, and what did the answer turn out to be?
Diagnostic. February 2024 had 29 days; per day, it sold almost exactly what January and March did.
Q5. Why does cleaning come before analysing?
Because every calculation uses the data as it is. Dirty data gives wrong answers that look right, and the mistake carries through to the chart and the decision.
Q6. Which tool is best for 30 million rows in a company database?
SQL.
Tick Before You Move On
- ☐ I can define data analysis in one sentence that mentions a question and a decision
- ☐ I can tell data, information, insight and a decision apart
- ☐ I can classify a column as numerical, categorical, date or identifier
- ☐ I can name internal, external, primary and secondary sources
- ☐ I can match a question to descriptive, diagnostic, predictive or prescriptive
- ☐ I can name the five workflow stages in order
- ☐ I know which of Excel, SQL and Power BI fits which job
Quick Recap
Data analysis is examining data to answer a question so someone can decide better. Data becomes information, information becomes insight, and insight earns a decision. Columns are numerical, categorical, dates or identifiers, and that decides what you can do with them. Analytics answers four questions: what happened, why, what next, and what to do. Every analysis follows the same path: raw data, clean, analyse, visualise, decide. Excel, SQL and Power BI are three tools for the same thinking.
Final Scenario
Back on the street in Computer Village, Bayo asks Adaeze how she always seems to have the right stock.
She does not show him a spreadsheet. She tells him a story: "In March I noticed power banks sold three times faster during outages (descriptive). The outages were why people bought them (diagnostic). NEPA always gets worse in the hot months (predictive). So I stock up in February (prescriptive)."
Every sentence she said climbed the ladder. That is what you will be able to do with any business's data by the end of this track.
A question at the start, a decision at the end, and careful work in between.
Next, Week 1, Virtual 1: review the workflow in your own words and connect it to a real Nigerian business. Then Week 1, Live 2: we open Excel for the first time.