Data Analytics Interview Questions for Analytics Companies
40 questions weighted to the interview style of analytics companies — 4 easy, 29 medium, 7 hard.
How analytics companies interview for Data Analytics
Analytics and data-services firms test practical skill hard: SQL, Excel or Python tasks, statistics, and business cases where you turn data into a recommendation. Many run a take-home or live case round in addition to technical interviews.
This set focuses on A/B Testing & Experiments (9), Tableau (5), Excel (4), Data Modelling & Warehousing (4).
HR round Technical interview Case or guesstimate round Technical round 1 Technical round 2 or case round
Solved — analytics companiesStart practicing
3 QUESTIONS FREESign up to unlock the full question bank and track your progress.Unlock every question and answer, then continue practising company-wise and package-wise.
✓ 3 questions free • Unlock the rest in seconds
01
Average delivery time improved by 4 minutes, but customer complaints about late deliveries went up. How do you investigate?
An average can improve while the tail gets worse. I would look at the distribution: median, 90th and 95th percentiles and the share of orders beyond the promised time. Then break it down by city, time slot and restaurant to see where late orders cluster. Also check whether promised times changed, whether complaint tracking changed and whether a few very fast orders pulled the mean down.
02
Write a Tableau table calculation for year-over-year sales growth.
// with YEAR(Order Date) in the view, Compute Using: Table (across)
(ZN(SUM([Sales])) - LOOKUP(ZN(SUM([Sales])), -1))
/ ABS(LOOKUP(ZN(SUM([Sales])), -1))
// the built-in Quick Table Calculation 'Year over Year Growth' generates similar logic
// format the field as a percentage
03
Write a DAX measure for a KPI card showing achievement against target and a colour for conditional formatting.
Achievement % =
DIVIDE([Total Sales], [Sales Target])
KPI Colour =
VAR a = [Achievement %]
RETURN
SWITCH(
TRUE(),
ISBLANK(a), "#9E9E9E",
a >= 1, "#2E7D32",
a >= 0.9, "#F9A825",
"#C62828"
)
-- apply KPI Colour via Format > Callout value > fx > Field value
04
What is the difference between extract filters and data source filters in Tableau?
Extract filters decide which rows go into the extract file at all, which reduces its size and speeds every view; filtered rows are simply not available. Data source filters apply to all worksheets using that data source, for live or extract, and are useful for security or scope, such as only the India business. Both act before context filters. Normal worksheet filters only affect one sheet.
05
What does apply do in pandas and why can it be slow?
apply runs a Python function on each row or column of a DataFrame, or each value of a Series. It is flexible but slow on large data because it loops in Python rather than using optimised C code. For example, df['total'] = df['qty'] * df['price'] is far faster than an apply with a lambda. Prefer vectorised operations, np.where, np.select, map with a dictionary, or string accessor methods first.
06
Tell me about a time you worked with a marketing team to improve a campaign using data.
Situation: a lead generation campaign had a high cost per lead and sales said leads were poor. Task: find what to change. Action: I joined ad data with CRM outcomes and found two audience segments produced most leads but almost no closed deals, while one smaller segment converted well. Result: we moved budget to that segment and changed the form questions; cost per qualified lead fell about 30% in six weeks.
07
What is DAX and how does it differ from Excel formulas?
-- A2 = start date, B2 = end date, Holidays = named range of holiday dates
=NETWORKDAYS(A2, B2, Holidays)
-- if the weekend is only Sunday (weekend code 11)
=NETWORKDAYS.INTL(A2, B2, 11, Holidays)
-- date that is 10 working days after A2
=WORKDAY(A2, 10, Holidays)
09
The customer table has near-duplicates like 'Rahul Sharma, 98xxxx1234' and 'Rahul Sharma, +91 98xxxx1234'. How do you deduplicate?
First normalise: trim and lower-case names, collapse spaces, and strip country codes and symbols from phones and emails. Then match on strong keys such as phone or email. For remaining records use fuzzy matching on names with a similarity score, restricted to the same city or pin code to limit comparisons. Review borderline scores manually, keep a mapping from old IDs to a master ID, and fix entry at the source.
I would explain that early results swing a lot and stopping on a good day inflates false positives, and that four days miss weekly patterns like weekend shopping. I would show how the lift has moved day by day and agree to the planned end date. If speed really matters, we can agree a sequential testing rule for future tests, but changing the rule mid-test after seeing data is not valid.
12
A stakeholder says the Power BI report shows different revenue than their Excel file. How do you troubleshoot?
I would first agree on definitions: gross or net, including tax or returns, order date or invoice date. Then compare one small slice, such as one month and one region, in both. I check filters on the report and page, relationships that drop unmatched rows, refresh time versus the Excel extract time, and duplicate rows. Once the gap is explained, I document the definition in the report so the question does not repeat.
13
Write SQL to show each category's percentage share of total revenue.
SELECT category,
SUM(amount) AS revenue,
ROUND(100.0 * SUM(amount)
/ SUM(SUM(amount)) OVER (), 2) AS pct_share
FROM sales
GROUP BY category
ORDER BY revenue DESC;
-- SUM(SUM(amount)) OVER () is the grand total across all groups
14
What is columnar storage and why do analytical warehouses use it?
Columnar storage keeps all values of one column together on disk instead of storing full rows together. Analytical queries usually read a few columns across millions of rows, so the engine reads only those columns. Similar values sit side by side and compress very well, which cuts storage and scan cost. The trade-off is slower single-row inserts and updates, which is why OLTP databases stay row-based.
15
Sales data from different countries mixes currencies and some weights are in grams while others are in kilograms. How do you handle it?
Never add mixed units. Identify the unit for each row from a currency or unit column, or infer it from the source system. Convert to one standard, for example INR using a dated exchange rate table joined on transaction date, and kilograms using a conversion factor. Keep the original value and unit columns for audit. Document which rates were used, and flag rows where the unit could not be determined.
An A/A test splits users into two groups that both see the same version. Since nothing differs, you expect no significant difference except about 5% of the time by chance. It checks that randomisation, tracking and the analysis pipeline work correctly, and it reveals sample ratio mismatches or metric variance before real tests depend on the platform. Running many A/A tests also confirms your false positive rate is near 5%.
Use a star schema, remove unused columns, prefer measures, reduce visuals per page, limit high-cardinality fields and use aggregations or incremental refresh.
18
What are primary metrics and guardrail metrics in an A/B test?
The primary metric is the one decision metric the test is designed to move, such as checkout conversion. Guardrail metrics are ones that must not get worse, such as page load time, refund rate, unsubscribe rate or revenue per user. A change that lifts conversion but increases cancellations should not ship. Define both before launch, so the decision rule is clear and nobody picks the favourable metric afterwards.
19
Marketing wants to A/B test two email subject lines on a list of 2,000 subscribers. What do you advise?
With 1,000 per group and a typical click rate of 3%, only a very large difference, roughly 70% relative or more, would be detectable, so a small real lift would look like noise. I would suggest testing on a larger send, pooling several campaigns, or using open-to-click rate with a bigger effect. If they still run it, report the confidence interval and treat the result as directional.
20
What is CUPED and how does it speed up experiments?
CUPED is a variance reduction method that uses each user's pre-experiment behaviour, such as spend in the previous four weeks, as a covariate. You adjust the outcome by removing the part explained by past behaviour, which is unrelated to the treatment because assignment was random. The lower variance gives tighter confidence intervals, so you can detect the same effect with fewer users or a shorter test, often 20-50% faster.
21
Write Python code to extract UTM source, medium and campaign from a column of landing page URLs.
Look at the metric that reflects business value, such as revenue or margin per visitor, which combines both effects. Here revenue per visitor is roughly 1.03 times 0.95, about 2% lower, so the change may hurt. Check whether the drop is significant, which segments drive it, and whether new buyers are making smaller first orders that grow later. Without a strong reason, I would not ship it as is.
Use Go To Special and select Constants with Numbers to highlight hard-typed values inside formula areas. Show Formulas (Ctrl+`) to scan columns for inconsistencies, and use Trace Precedents and Dependents on key totals. Inquire or Spreadsheet Compare, if available, can flag inconsistent formulas. Then fix the overwritten cells, protect formula ranges and add check totals that compare with the source data each month.
25
Write Python code to test whether conversion differs between control and treatment.
from statsmodels.stats.proportion import proportions_ztest, proportion_confint
conversions = [1180, 1295] # control, treatment
visitors = [30000, 30100]
z, p = proportions_ztest(conversions, visitors)
print(f'z = {z:.2f}, p = {p:.4f}')
for c, n in zip(conversions, visitors):
low, high = proportion_confint(c, n, method='wilson')
print(f'rate {c/n:.3%} 95% CI {low:.3%} to {high:.3%}')
26
How do you present insights to non-technical stakeholders?
Lead with the answer, use one simple chart per point, translate numbers into business impact (₹, %), avoid jargon and end with a clear recommendation and next step.
27
Write a Tableau calculated field for profit ratio and a profit category.
// Profit Ratio (aggregate first, then divide)
SUM([Profit]) / SUM([Sales])
// Profit Category
IF SUM([Profit]) < 0 THEN 'Loss'
ELSEIF [Profit Ratio] < 0.1 THEN 'Low margin'
ELSE 'Healthy'
END
// avoid AVG([Profit]/[Sales]): it averages row ratios and gives a wrong total
28
Why are A/B tests tricky in two-sided marketplaces such as ride-hailing or food delivery?
In marketplaces, groups share supply, so the treatment can affect control. If treatment riders get a discount and book more, fewer drivers remain for control riders, which makes treatment look better than it would at full rollout. This violates the independence assumption. Teams use cluster randomisation by city or region, switchback tests that alternate treatment by time slots, or model the spillover when estimating impact.
29
A stakeholder wants a 40-column crosstab in Tableau that looks exactly like their Excel sheet. How do you handle it?
I would ask what they do with it. If they export and work on it in Excel, a crosstab view with a download option or a scheduled extract is fine, and Tableau should not fight Excel. If they actually scan it for exceptions, I would suggest a summary view with highlights and a drill-down to the detail table. Often both: a short visual summary on top and the detailed table on a second tab.
A factless fact table records that an event or relationship happened but has no numeric measures, only dimension keys. Examples are student attendance (student, class, date) or which products were on promotion in which store each day. You analyse it by counting rows. A promotion coverage table is useful for questions like which promoted products did not sell, by comparing it with the sales fact.
32
Write SQL to generate a simple date dimension for 2026.
-- PostgreSQL syntax
CREATE TABLE dim_date AS
SELECT
TO_CHAR(d, 'YYYYMMDD')::INT AS date_key,
d::DATE AS full_date,
EXTRACT(YEAR FROM d) AS year,
EXTRACT(QUARTER FROM d) AS quarter,
EXTRACT(MONTH FROM d) AS month_no,
TO_CHAR(d, 'Mon') AS month_name,
TO_CHAR(d, 'Dy') AS weekday,
EXTRACT(ISODOW FROM d) IN (6, 7) AS is_weekend
FROM GENERATE_SERIES(DATE '2026-01-01', DATE '2026-12-31', INTERVAL '1 day') AS d;
33
Your pandas script shows SettingWithCopyWarning and a new column does not update. What is happening and how do you fix it?
It warns that you may be modifying a temporary copy of a slice, so the change might not reach the original DataFrame, for example df[df.sales > 0]['flag'] = 1. Avoid it by using a single loc call, df.loc[df.sales > 0, 'flag'] = 1, or by calling .copy() when you intentionally create a new subset. Newer pandas versions with copy-on-write make behaviour more predictable, but clear loc assignments remain best practice.
34
What is the pyramid principle and how does it apply to presenting analysis?
The pyramid principle means starting with the answer, then giving the few key arguments that support it, then the detailed data under each argument. Analysts are tempted to present in the order they worked, from data to conclusion, but leaders want the conclusion first. Example: we should cut the free-delivery threshold, because it lifts order frequency, margin impact is small and competitors already did it, each backed by evidence.
35
What is an action title and why should you use one?
An action title states the insight of a chart or slide as a sentence instead of a label. Instead of Monthly Sales by Region, write South region drove 70 percent of the decline since July. Busy executives may read only the titles, so each one should carry the message and, ideally, the so-what. The chart below then proves the title. Read only your titles in order: they should tell the whole story.
36
Write Python code to bootstrap a 95% confidence interval for the difference in revenue per user between two groups.
A parameter is a single user-controlled value, such as a number, date or string, that is independent of the data. On its own it does nothing; you reference it in calculations, filters, reference lines or bins. Common uses are a Top N selector, a what-if growth rate, or a measure switcher where a CASE statement picks Sales, Profit or Quantity based on the parameter. Parameter actions let users set it by clicking a mark.
Partition the table by date so the engine can prune partitions outside the filtered week, and cluster or sort by frequently filtered columns such as store or customer. Make sure the filter is on the partition column directly, not wrapped in a function like CAST or DATE_TRUNC that blocks pruning. Select only needed columns, because storage is columnar. For dashboards, also consider a pre-aggregated daily summary table.
SUMIFS sums values meeting several criteria and is fast. SUMPRODUCT multiplies arrays and adds them, handling more complex conditions and weighted totals.
How are Data Analytics interviews at analytics companies different?
Analytics and data-services firms test practical skill hard: SQL, Excel or Python tasks, statistics, and business cases where you turn data into a recommendation. Many run a take-home or live case round in addition to technical interviews.
Which analytics companies are covered?
This page links to 11 analytics companies, including EXL, Mu Sigma, Fractal Analytics, LatentView Analytics, Tiger Analytics, Tredence, Sigmoid, Quantiphi. Each has its own page with rounds and a separate practice set.
What rounds should I expect?
Common rounds at these companies include: HR round; Technical interview; Case or guesstimate round; Technical round 1; Technical round 2 or case round. The exact process varies by company and role.