Why Companies Use Take-Home Assignments
About 60% of analyst roles at mid-size companies include a take-home, because an interview question cannot show whether you can clean a real table. A recruiter I worked with said her team discards roughly 40% of submissions for broken files or charts with no labels. The assignment is your chance to prove you can deliver the kind of work the job actually requires, so match the output to the team's tools, usually SQL, Excel, or Tableau.
Open the email or PDF and list every deliverable: a cleaned dataset, an analysis file, and a presentation. A common mistake is starting in Excel and missing the 'one page of recommendations' line, which this will break your score even if the numbers are right.
Most take-homes expect 3 to 5 hours of work over 3 to 5 days. I recommend you spend 30% cleaning, 40% analyzing, and 30% presenting. Do not burn the whole window on a single fancy chart.
Understand the Dataset First
Before any calculation, learn what each table and column means. Real assignments often ship an Orders table with Order ID, Customer ID, Order Date, Region, Product, Quantity, and Revenue, plus a Customers table with signup date and churn flag. Sketch the join keys on paper so you do not aggregate the wrong grain.
In Excel or Python, count nulls, check date formats, and list distinct values for Region and Product. If Revenue has 200 blank rows, note it. A hiring manager told me he marks down candidates who report a total without mentioning data quality.
Decide whether you report per Order ID, per Customer ID, or per day. Mixing grains is the classic error. If the brief asks for monthly revenue, group by Order Date truncated to month, not by individual order.
Turn the vague brief into sharp questions, for example 'Which region lost the most revenue year over year?' Keep the list visible so your analysis stays on target and your final slide answers them directly.
Clean the Data Before You Analyze
Dirty input is where most candidates lose points, not the analysis itself. Spend the first block removing duplicates, fixing types, and documenting every change. The grader wants to see a repeatable process, so keep your cleaning steps in a script or a clearly labeled Excel tab.
A practical rule I use on every assignment: if a column is more than 15% null, do not silently drop it, surface it to the reviewer instead. In one real submission the Revenue column had 8% blanks, all concentrated in a single Region, and that gap was the actual story the manager cared about. The grader rewards candidates who notice the pattern, not those who hand over a suspiciously perfect table.
In Excel use Data, Remove Duplicates on Order ID. In SQL use SELECT DISTINCT or a ROW_NUMBER() window to keep one row per order. Document the count removed, reviewers like seeing 'dropped 142 duplicate rows'.
Convert Order Date from text to a real date and strip currency symbols from Revenue. This will break any time trend if left as text, because the tool sorts '10/01' before '2/01' alphabetically.
Decide a rule: drop rows under 5% missing, or impute. State the rule in your notes. ATS-screened resumes and take-homes both reward candidates who show they thought about missing data instead of ignoring it.
Run the Core Analysis
Now answer your three business questions with concrete numbers. Use SQL for aggregation if the brief mentions a database, or pivot tables if it is Excel-based. Either way, compute the headline metric, the trend, and one segment breakdown, then sanity-check against a handwritten total.
Keep a short notes tab that lists every metric next to its source column and formula. A hiring manager I spoke with said she trusts submissions that show their work, because numbers with no trail are easy to fake. I keep one row per metric, name, formula, and the sheet it lives in, so any reviewer can reproduce the result in under two minutes during the follow-up call.
Run
SELECT Region, SUM(Revenue) AS total_rev
FROM Orders
GROUP BY Region
ORDER BY total_rev DESCThe top region should match a quick Excel SUMIFS so you trust the result before presenting it.
Group Revenue by month of Order Date and calculate the percent change. In my last take-home, December to January showed a 30% drop in repeat purchases, the finding that landed me the interview.
Slice Revenue by Product Category to find the weakest line. A common mistake is only reporting the company total and missing the segment where the real problem lives, which a hiring manager will probe in the follow-up.
I recommend you round reported percentages to whole numbers and label every figure with its source column. Avoid presenting 14.382% when 'about 14%' is clearer, managers read take-homes on a phone between meetings.
Build a Short, Clear Presentation
The deliverable is judged by a busy manager, not a statistician. One title slide, one data slide, and one recommendations slide is the right length. Lead with the answer, then show the chart, then say what to do. A recruiter said she skips submissions longer than five slides unless the role is senior.
Make slide one title 'West region drove a 30% revenue drop in Q1'. The reader gets the point before seeing the chart. This is the single habit that separates strong take-homes from weak ones.
Show a bar chart of Revenue by Region and a line of the monthly trend. Label axes, add a source line, and remove chart junk. See our guide on choosing the right Excel chart for line, bar, and scatter use.
Give actionable next steps: investigate the West discount policy, re-engage lapsed customers, and review the weakest Product Category. Tie each to the number you found so the advice is not generic.
Common Mistakes That Sink Take-Homes
Time management is part of the score. I recommend you cap cleaning at 90 minutes even if the data is ugly, because an unfinished analysis section hurts more than a slightly imperfect table. A recruiter mentioned she weighs the depth of insight over polish, so protect the hours that produce the headline finding and do not polish the title slide at the expense of the numbers.
A common mistake is delivering clean totals while hiding that 200 rows were dropped. Always attach a short 'cleaning log'. This will break trust if the reviewer finds the gap before you mention it.
Do not train a machine learning model to answer 'which region declined'. A grouped SUM and a chart answer it faster and more clearly. Reviewers penalize complexity that hides the result.
Attach the query or workbook, not just the slide. A hiring manager wants to reuse your method. Avoid sending a PDF with numbers and no backing file, it reads as copied.
What Happens in the Follow-Up Interview
After you submit, expect a 30 to 45 minute call where the panel walks through your file line by line. They are not testing new skills, they are checking the work is yours and that you can defend each choice. Practice explaining why you dropped duplicates and how you computed the trend.
Be ready to say 'I grouped by month because the brief asked for trend' in one sentence. A recruiter told me the candidates who freeze here are the ones who outsourced the work.
Ask what the team does with take-home insights after hire. It shows you think past the assignment. Strong questions signal you already picture the job, which nudges the offer.
In my experience, candidates who present their own take-home out loud for five minutes the night before interview day sound far more confident. You will hit this exact moment, so rehearse the walkthrough once.
Next Steps After You Submit
Save the cleaned dataset and slides in a portfolio folder, because the same analysis makes a great portfolio project later. If the role is through a certificate program, link your work to that credential. US data-analyst salaries for people who clear the take-home loop commonly land between $72,000 and $95,000, so the few hours of prep pay back quickly. Keep practicing SQL aggregation and pivot tables so the next assignment feels routine.
