Dayananda Sagar University Students: Preparing for a Data Analytics Internship Case Study Round

To prepare for a data analytics internship case study round, learn one repeatable framework, practise it against the clock, and rehearse speaking your reasoning aloud. Interviewers rarely want a perfect answer; they want to see you clarify the problem, structure the analysis, handle messy data sensibly and end with a recommendation. This guide gives you that framework, a 14-day practice schedule, and three mini cases on retail, edtech and mobility.

What a data analytics case study round actually tests

A case round gives you a business situation, often vague, and asks how you would use data to make progress. It may be live, in a 30 to 45 minute conversation, or take-home, with a small dataset and a deadline. Either way, the same four abilities are being checked:

• Problem framing: can you turn “sales are down” into a measurable question?
• Analytical structure: do you break the problem into parts instead of listing random metrics?
• Data judgement: do you notice missing data, bias, and the difference between correlation and cause?
• Communication: can you finish with a clear, defensible recommendation and say what you are unsure about?

Technical skills such as SQL, spreadsheets or Python matter, but in a case round they are usually a means. A candidate who writes a flawless query for the wrong question scores worse than one who asks two sharp questions first.

The seven-step case framework, from clarifying question to recommendation

Use these steps in order. Say each step’s name out loud in a live round so the interviewer can follow you.

1. Restate and clarify (2 to 3 minutes). Repeat the problem in one sentence. Ask about the business goal, the time period, the customer segment, and how success is measured.
2. Define the metric. Pick one primary metric (for example, weekly active buyers) and one or two guardrail metrics (for example, refund rate). Write the formula.
3. Build a hypothesis tree. Split the problem into branches: internal factors versus external ones, or acquisition, activation, retention and revenue. Each branch gets one or two testable hypotheses.
4. Choose the data and the cut. State which tables or fields you need, and how you would slice them: by time, geography, cohort, channel, product or device.
5. Check data quality. Ask about duplicates, nulls, tracking changes, outliers and sample size before trusting any trend.
6. Analyse and interpret. Describe what you would compute, what result would confirm or reject each hypothesis, and what you would do in each case.
7. Recommend and plan next steps. Give one clear action, its expected impact, the risks, and how you would measure whether it worked, such as an A/B test or a pre/post comparison.

Clarifying questions worth memorising

• What decision will this analysis feed, and who makes it?
• Is the change sudden or gradual, and did anything launch or break around that date?
• Is the metric down for everyone or only a segment?
• What is the baseline: last week, last year, or a target?
• What data exists, and how far back does it go?
• Are there constraints on budget, time or engineering effort?

Case round scorecard: rate your own answer

After every practice case, score yourself with this table. Anything below 3 in a row tells you where to spend tomorrow’s practice.

A 14-day timed practice schedule for case rounds

This plan fits the shape of a free two-week online internship and works equally well as a standalone prep sprint. Plan for about 90 minutes a day; adjust on exam-heavy days. The free two-week CareerSwami internship gives you 14 days of structured analytics work to practise alongside, and you receive an offer letter and a completion certificate after successful completion.

Three mini cases to practise with

All three cases below are illustrative. The numbers are invented for practice and are not drawn from any real company. Treat the suggested approach as one reasonable path, not the only answer.

Mini case 1: Retail, falling repeat purchases

Prompt: An online grocery retailer sees repeat purchase rate drop from 42% to 35% over two months. What do you do?

• Clarify: Is repeat rate measured within 30 days? Did a pricing, delivery or app change happen? Is it all cities or some?
• Hypotheses: new customers from a discount campaign are lower quality; delivery delays rose; stock-outs on staple items; a competitor launched.
• Cuts: cohort by acquisition month and channel, city, delivery time, order size, category availability.
• Data checks: was the definition of “repeat” changed, and are guest checkouts being counted as new users?
• Recommendation shape: if the drop sits in one acquisition cohort, adjust campaign targeting; if it follows late deliveries, prioritise logistics in affected cities. Track repeat rate by cohort for four weeks.

Mini case 2: Edtech, low course completion

Prompt: An edtech platform finds only 18% of learners finish its paid data course. Leadership wants to raise this. Where do you start?

• Clarify: What counts as completion? Is the course self-paced? Has completion changed over time?
• Metric: completion rate, with guardrails on learner satisfaction and refund requests.
• Hypotheses: drop-off at one hard module; learners are too busy after week two; poor onboarding; mismatched prerequisites.
• Cuts: funnel by module, time of enrolment, device, prior experience, engagement in week one.
• Recommendation shape: if one module shows a sharp cliff, redesign it and test reminders for learners who stall. Run an A/B test and compare completion at the learners’ own pace, not a fixed date.

Mini case 3: Mobility, rising ride cancellations

Prompt: A ride-hailing app notices driver-side cancellations rose from 8% to 13% in one city. Diagnose and recommend.

• Clarify: Is it after acceptance? Which hours? Did fares, incentives or the app flow change?
• Hypotheses: pickup distance grew because of supply shortage; low fares on short trips; a new incentive rewards completed trips only; traffic or weather.
• Cuts: hour of day, pickup distance, fare band, trip length, driver tenure, zone.
• Data checks: are system-triggered cancellations mixed with driver-initiated ones?
• Recommendation shape: if long pickups drive cancellations, adjust dispatch radius or add short-pickup bonuses in the worst zones, then compare cancellation rates against a similar control zone.

Worked illustration: answering the retail case in 30 minutes

This is an illustrative walkthrough, not a record of any test or run. Here is how a timed answer could be split:

1. Minutes 0 to 4: restate (“repeat rate fell seven points in two months; I want to find which customers stopped returning and why”) and ask four clarifying questions.
2. Minutes 4 to 8: define repeat rate as customers with a second order within 30 days divided by customers with a first order that month; set average delivery time as a guardrail.
3. Minutes 8 to 14: draw the tree with branches for customer mix, service quality, product availability and competition.
4. Minutes 14 to 22: describe the cohort comparison and a small query sketch.

-- Illustrative sketch only
SELECT acquisition_month, channel,
       COUNT(DISTINCT CASE WHEN second_order_days <= 30 THEN customer_id END) * 1.0
       / COUNT(DISTINCT customer_id) AS repeat_rate
FROM customer_first_orders
GROUP BY acquisition_month, channel;

5. Minutes 22 to 28: interpret both outcomes: “If the weak cohort is the discount-driven one, the fall is mostly a mix effect and the product is fine. If every cohort fell after late deliveries rose, it is a service problem.”
6. Minutes 28 to 30: recommend one action and a way to measure it.

Dayananda Sagar University students: fitting case prep around internship and placement structures

DSU describes itself in its 2026-30 scheme document as a State Private University under Karnataka Act No. 20 of 2013. The same document shows that in the B.Tech CSE (Data Science) scheme, course 26DS4802 “Internship (Industry / Research / CoE based)” sits in semester VIII with 2 credits and 100 marks of continuous internal evaluation. That means your internship is assessed by ongoing evaluation, so keeping dated notes, outputs and mentor feedback from any internship, including a short online one, is sensible.

DSU’s wider rules matter too. An older scheme document says B.Tech programmes are governed by the DSU B.Tech Rules and Regulations 2021, and lists internship types as Rural, Industry and Research/Development for the 2023-27 batch. The pages reviewed gave no internship duration, documents required or rule on online internships, so do not assume a two-week online internship counts toward any credit requirement.

For preparation support, the Training and Placement Cell overview says the cell offers technical training and soft-skills development, which can complement case practice. Timing also matters: the 2025-26 B.Tech academic calendar places summer-term exams from 13.07.2026 to 31.07.2026 and odd semesters starting 10.08.2026. That calendar is for 2025-26, so check the current one before planning your 14 days.

Verification steps for DSU students planning internship case prep

1. Ask your department’s internship coordinator in writing whether a short online internship can be recorded, and what evidence (certificate, report, mentor letter) is needed.
2. Request the current DSU B.Tech Rules and Regulations and your batch’s scheme document, and look for internship duration and evaluation clauses.
3. Confirm which internship course code applies to your batch, since schemes differ by year and branch.
4. Email the Training and Placement Cell through the placements contact page and ask whether mock case rounds or analytics workshops are offered this term.
5. Check the latest academic calendar for your year so your practice days avoid exam weeks.

Mistakes that cost candidates marks in analytics case rounds

• Starting with tools. Saying “I would run a regression” before defining the problem signals recitation, not thinking.
• Listing every metric. Choose one primary metric and defend it.
• Ignoring segments. Averages hide the real story; always propose a cut.
• Treating correlation as cause. Say what experiment or comparison would separate the two.
• Ending without a decision. Even with uncertainty, commit to a recommended next step.

Take-home case studies: a time budget for the written version

For a take-home round with a 48-hour window, spend your effort deliberately:

• First hour: read the brief twice, write the questions you would have asked, and state your assumptions in the document.
• Next 25% of time: clean data and log every cleaning decision.
• Next 40%: explore, test hypotheses, and keep only the charts that support a point.
• Final 25%: write a one-page summary with the recommendation first, then the evidence, then limitations.

Keep the submission short. A reviewer reading thirty submissions will favour a clear two-page answer over a twelve-page notebook dump.

Frequently Asked Questions

How long should I spend on clarifying questions in a live case?

Two to four minutes is usually enough. Ask questions that could change your approach, such as the time window or the segment affected, and stop once you can restate the problem confidently.

Do I need Python for a data analytics case round?

Not always. Many rounds are discussion-based and test structure and judgement. When coding is involved, SQL and spreadsheets are common, so check the internship description and prepare for the tools it names.

What if I do not know the business domain in the case?

Say so briefly and ask the interviewer to explain the process, such as how a booking or enrolment flows. Then apply your framework. Interviewers value a candidate who learns the context quickly over one who bluffs.

How many practice cases are enough as a Dayananda Sagar University student?

Quality beats volume. Six to ten cases, each scored against the scorecard above and followed by a rewrite of the weakest part, will help more than thirty cases attempted casually.

Will a two-week online internship count for credit at DSU?

That depends on your department and batch rules, which this guide cannot confirm. The published scheme for one B.Tech programme lists an internship course in semester VIII, but ask your internship coordinator in writing before assuming anything counts. The CareerSwami internship itself offers a free offer letter and a free completion certificate after successful completion, with no academic credit or stipend claimed.

How do I practise speaking my reasoning if I have no partner?

Record yourself on your phone solving a case aloud, then replay it and score yourself on the table above. You will quickly hear filler words, skipped steps and missing recommendations.

What should the final minute of my answer sound like?

Give a short summary: the problem, your most likely cause, the single action you recommend, the risk, and how you would measure success. Keep it to four or five sentences.

Is it fine to say “I would need more data” at the end?

Yes, if you also say which data, why it would change the decision, and what you would recommend in the meantime. Used that way, it shows judgement rather than avoidance.

Related Data Analytics guides

• Lingaya's Vidyapeeth Students: First-Year Guide to a Light Data Analytics Internship
• West Bengal State University Students: Structuring a GitHub Repository for Analytics Internship Work
• The Global University Students: Finding Remote Data Analytics Internships from Arunachal Pradesh
• All Data Analytics internship guides