Day 30: Power BI Mock Interview — Full Simulation

Minute 1–5 (Data Quality): Load via Power Query Transform. Check: data types (OrderDate as Date, Quantity as Whole Number, UnitPrice as Decimal), nulls in each column, duplicate OrderIDs. Add Revenue column: Quantity * UnitPrice. Minute 5–10 (Date Table): Create Date Table = CALENDAR(DATE(2024,1,1),DATE(2026,12,31)). Add Year, Month Number, Month Name, Quarter columns. Create relationship: Date[Date] → Sales[OrderDate]. Mark as Date Table. Minute 10–15 (Measures): Total Revenue = SUM, Total Orders = COUNTROWS, Avg Order Value = DIVIDE, Revenue LY = CALCULATE with SAMEPERIODLASTYEAR, YoY Growth % = DIVIDE difference. Minute 15–35 (Report): Page 1 KPIs — 4 cards. Line chart monthly trend. Column chart by Region. Bar chart top 5 products. Page 2 drill-through by customer. Minute 35–40 (Format): Consistent colours. Slicers for Year and Region. Conditional formatting on Region chart. Minute 40–45 (Validate): Check totals match raw SUM of Revenue column.

The three metrics that matter most for most businesses: (1) Revenue vs Target (with achievement %) — are we hitting our goal? (2) YoY Growth % — are we growing vs last year? (3) Profitability — are we making money on what we sell? Design: Three large KPI cards at the top occupying 80% of the page. Below each: a small sparkline (or 12-month mini trend). One slicer for Business Unit (if the CEO oversees multiple). A single line of text below each metric explaining what changed vs last month ("Revenue up ₹2.3Cr driven by North region"). Building: Create one measure per metric, build KPI visuals, add dynamic subtitles using SELECTEDVALUE and text concatenation. The entire page should be readable in 10 seconds.

New Customer Revenue = VAR MinOrderDate = MIN(Sales[OrderDate]) VAR MaxOrderDate = MAX(Sales[OrderDate]) VAR NewCustomers = FILTER( VALUES(Sales[CustomerID]), CALCULATE( MIN(Sales[OrderDate]), ALL('Date') ) >= MinOrderDate ) RETURN CALCULATE( SUM(Sales[Revenue]), NewCustomers ) Logic: for each CustomerID, find their very first purchase date (MIN across ALL dates). If that first purchase date falls within the current filter context period (between MinOrderDate and MaxOrderDate), they are a new customer. Sum revenue only for those customers. This is a classic cohort analysis pattern.

Respond professionally and redirect to outcomes: "I understand you want comprehensive visibility — let me ask a few questions to make sure the report serves you well. What decisions will you make from this report? Which of those decisions are most time-sensitive? When you say all regions — are you comparing regions to each other, or looking at each independently?" Then explain the design implication: "If I put all regions, all products, and all months on one page, each chart will have too much data to read meaningfully. What I can do is: show regional comparison on page 1, drill through to monthly detail for any region on page 2, and use a slicer to filter by product category throughout. This gives you access to everything with a cleaner experience. Would that work?" The goal is serving the business need, not literally doing what was asked.

(1) Data modelling discipline: always mention star schema, Date table, and correct relationship direction before discussing visuals. This immediately separates you from candidates who only know the surface. (2) DAX understanding: show you understand filter context, not just function syntax. Explain WHY a formula works, not just WHAT it produces. (3) Business thinking: connect every technical decision to a business outcome. "I used SAMEPERIODLASTYEAR because the manager needs to compare August this year vs August last year to understand seasonality." (4) Professional habits: mention validation, naming conventions, and documentation. These signal you build maintainable, trustworthy reports. (5) Honest self-assessment: know what you know deeply and what you know conceptually. Interviewers respect "I know this exists but have not implemented it yet" more than a bluffed answer that unravels under one follow-up question.