Data analysts do not need a statistics degree, but they do need a handful of statistical ideas, a reliable way to check data quality, an understanding of how company data is organised, and the ability to explain a finding so people act on it. This guide covers those core skills, then what the job looks like day to day and how to prepare for interviews in 30 days.
It pairs with our tool guides: Excel, SQL and Power BI, and with our machine learning guide if you are heading towards data science.
Contents
- 1. Mean, median and spread
- 2. Five statistical ideas to use every day
- 3. Hypothesis testing and p-values
- 4. Data quality checklist
- 5. What is a data warehouse?
- 6. Star schema
- 7. Data storytelling for managers
- 8. Using Gen AI tools safely
- 9. What a data analyst does in a typical week
- 10. Data analyst vs business analyst
- 11. A 30-day interview preparation plan
- 12. Top 20 data analyst interview questions
1. Mean, median and spread
The mean is the average, the median is the middle value when sorted, and the mode is the most common value. When a few very large values exist, such as salaries or order values, the median is usually the better summary; report both when the difference matters.
Spread shows how far values sit from the average. A small standard deviation means most values are close to the mean; a large one means they are widely spread. Two groups can share an average and still behave very differently.
Exercise: take a dataset with hours studied and test scores, calculate the mean and standard deviation of each and the correlation between them, then write one sentence on what the result shows and one on what it does not prove.
2. Five statistical ideas to use every day
- Averages and medians: choose the median when outliers distort the mean.
- Percentages need a base: “sales doubled” from 2 orders to 4 is very different from 2,000 to 4,000. Show the base beside the rate.
- Trends vs seasonality: a rise may only be the usual season. Compare with the same month last year before calling it a trend.
- Correlation is not cause: two things moving together does not prove one drives the other; check for a third factor that explains both.
- Sample size: a result from 20 customers is weaker than one from 2,000. State how many rows a finding rests on.
3. Hypothesis testing and p-values
A hypothesis test starts from a default assumption, the null hypothesis, that there is no real effect: “the new landing page converts the same as the old one”. It then asks how likely the observed result would be if that were true.
The p-value is the probability of a result at least as extreme as yours, assuming the null hypothesis is true. A p-value of 0.03 means a gap this large would appear only about 3% of the time by chance if there were really no difference. It is not the probability that the null hypothesis is true, it does not measure how big or important the effect is, and a single p = 0.04 is evidence, not proof.
Example: enquiry rates were 4.0% on the old page and 5.2% on the new page, with 500 visitors each. A two-proportion test gives a p-value; if it is small, say below 0.05, the difference is unlikely to be pure chance, but check the sample size, because small samples give unstable results.
from scipy import stats
t_stat, p_value = stats.ttest_ind(group_a, group_b)Use t-tests for averages and chi-square or two-proportion tests for rates and categories. Decide the test and threshold before looking at the data, report the size of the effect and a confidence interval, and remember that running many tests on the same data raises the chance of a false positive.
4. Data quality checklist
Most reporting problems come from incomplete, duplicated or stale data rather than formulas. Run this checklist before a manager sees the numbers:
- Completeness: how many key fields (customer ID, date, amount) are blank? Use
=COUNTBLANK(range)in Excel orWHERE customer_id IS NULLin SQL. - Duplicates: is any key repeated?
SELECT order_id, COUNT(*) FROM orders GROUP BY order_id HAVING COUNT(*) > 1; - Valid ranges: negative amounts, future dates, percentages above 100.
- Consistency: one spelling per category (“Bangalore”, “bangalore”, “Bengaluru”), summary totals matching detail rows, one date format.
- Freshness: when was the data last updated, and are there gaps such as a missing week?
- Reconciliation: compare totals with an independent source such as the bank statement, ad platform or finance system, and explain differences before publishing.
Write a short note of what you checked, what was wrong and what you did; it protects your credibility when a number is questioned. Then automate the routine checks as queries or Power Query steps that run on every refresh.
5. What is a data warehouse?
A data warehouse is a central store of cleaned, organised data from many sources, such as sales, billing and the website. Companies use one for a single version of the truth (every report shares the same definitions), for speed (queries run on data prepared for analysis, not on the live system taking orders) and for history (old values are kept so you can see change over time). You do not need to build one to use one: learn to read its fact and dimension tables, join them with SQL and check totals against the source.
6. Star schema
A star schema organises reporting data into one central fact table recording events, such as sales with quantity and amount, surrounded by dimension tables describing them, such as products, customers and dates. Each dimension row has a key that the fact table points to. Reports load faster because the joins are simple, measures are defined once and reused, and new team members understand the model quickly. “Revenue by product category by month” becomes one join from the fact table to each dimension. Power BI models work best in this shape; see relationships and the star schema.
7. Data storytelling for managers
Analysts are judged less on their SQL than on whether people act on their findings. Use a simple structure:
- The question in one sentence: “Why did enquiries drop in October?”
- The answer first: “Enquiries fell 30% because the ad campaign paused on 12 October; organic enquiries were stable.”
- Two or three pieces of evidence: one chart per message, titled with the finding (“Paid enquiries stopped after 12 Oct”), the key line highlighted and the rest grey, compared with last month, target or last year.
- The limits: what the data cannot tell you, such as enquiries that came by phone.
- A recommendation with an owner and a check: “Restart the search campaign with the updated budget; review enquiries per day after two weeks.”
Writing a five-slide summary of one of your projects in this structure is also the best way to answer “Tell me about a project” in an interview.
8. Using Gen AI tools safely
Gen AI tools can draft a first SQL query, explain an unfamiliar formula or DAX measure, suggest a chart type, or turn findings into a short summary. They are useful only when you check the result against the data:
- Run a drafted query on a small sample and compare its total with a simple SUM.
- Confirm the column names and data types match your table.
- Ask whether a join could repeat rows and silently inflate totals.
- Keep the original data unchanged and work on a copy, and keep private data out of prompts.
Log what you asked, what the tool returned and what you changed; that record is useful in interviews.
9. What a data analyst does in a typical week
- Start of the week: requests from other teams (a sales number, a report that looks wrong, a manager’s new question). Part of the job is asking what decision each request supports.
- Midweek: pulling data with SQL, cleaning it and checking numbers against the source. Careful, detailed work.
- Reports and dashboards: updating regular reports, fixing broken refreshes and building new charts, checking totals before anyone else sees them.
- End of the week: sharing findings in a short meeting or written summary. Explaining a number clearly matters as much as calculating it.
The role suits you if you enjoy finding the reason behind a number, are patient with detail and can explain results in plain words.
10. Data analyst vs business analyst
- Main question: a data analyst asks “what does the data show?”; a business analyst asks “what does the business need?”
- Daily tools: data analysts use SQL, Excel, Python and dashboards; business analysts use requirement documents, process diagrams and meetings.
- Who they work with: data analysts often sit with reporting and finance teams; business analysts work across departments and with technical teams designing new systems.
If you enjoy numbers, puzzles and charts, start with data analytics; if you enjoy talking to people, mapping processes and writing clear requirements, look at business analysis. Many people move between the two later.
11. A 30-day interview preparation plan
- Week 1, SQL and Excel: practise SELECT, WHERE, GROUP BY and JOIN daily on a small database, plus pivot tables, lookups and cleaning a messy column.
- Week 2, Power BI or Tableau: build one dashboard from a public dataset and be ready to explain each chart and the top finding.
- Week 3, your projects: write two projects as short stories (question, data, what you did, result) and practise explaining each in two minutes.
- Week 4, mock interviews: expect “How would you find duplicate customers?”, “What does a LEFT JOIN return?” and “Tell me about a time your analysis changed a decision.”
On the day, bring your laptop with one dashboard open and be ready to walk through it. If you do not know something, say so and explain how you would find out.
12. Top 20 data analyst interview questions
Excel
- VLOOKUP vs XLOOKUP? VLOOKUP searches the first column and returns a value to its right; XLOOKUP looks in any direction and handles missing values better.
- What is a pivot table? A tool that groups and totals large data without formulas.
- How do you remove duplicates? Data, then Remove Duplicates, after deciding which columns define a duplicate.
- What does SUMIFS do? Adds values that meet more than one condition.
- How can AI help in Excel? It can suggest formulas and explain results, but you must check its answers.
SQL
- WHERE vs HAVING? WHERE filters rows before grouping; HAVING filters groups after GROUP BY.
- Types of joins? Inner, left, right, full and self joins.
- What is a primary key? A column that uniquely identifies each row.
- What is a subquery? A query inside another query.
- What is a window function? A calculation across related rows without collapsing them, such as ranking.
Statistics and data
- Mean, median or mode? Use the median when outliers distort the mean.
- What is an outlier? A value far from the others; check it before removing it.
- How do you handle missing values? Remove, fill with a sensible value or flag them, depending on the data.
- What is correlation? How two variables move together; it does not prove cause.
Python, Power BI and projects
- What is pandas used for? Cleaning, filtering and analysing tabular data in Python; a DataFrame is its table of rows and columns.
- What is DAX? Power BI’s formula language.
- Dashboard vs report? A dashboard gives a quick view of key numbers; a report goes deeper.
- Tell me about a project. The problem, the data, what you did and what you found.
- Why data analytics? Link your interest in solving problems with data to the role.
For more, see the Data Science questions and answers, the data analyst resume guide and data analyst salaries in Bangalore for freshers.
Learn data analytics with ScholarLogic
Statistics, data quality, dashboards and interview preparation are part of the 5-month Data Analytics course in Bangalore at ScholarLogic, BTM Layout, including mock interviews. Book a free demo class and a counsellor will share the fee and the next batch dates. To book, call 78923 93639.
