Data Analytics |

Should This Amazon Seller Adopt FBA? A Data-Driven Answer

Mon, 20 Jul 2026 00:00:00 +0000

Analog Heroes is a fictional US-based Amazon seller of bullet journal supplies. The business is healthy and growing, but every order ships merchant-fulfilled, and shipping now consumes a third of revenue: 31.7% in 2024, 34.0% in 2025. The question the owners want answered: do any SKUs justify the fees of Fulfillment by Amazon, and if so, which ones?

The dashboard below answers it with a ranked shortlist. Sales velocity and projected annual savings are computed per SKU against a simplified FBA fee model (size tiers, fulfillment fees, seasonal storage rates). Eight SKUs clear both thresholds at baseline. The Prime Uplift slider on the FBA Opportunity page adds the speculative part: assume the Prime badge lifts conversion by up to 30%, and watch the shortlist grow from 8 candidates to 15.

Explore it live (best viewed on a desktop screen):

Every number in the report is synthetic by design. A seeded Python generator produces two years of order lines in the layout of Amazon’s real seller reports, calibrated to realistic patterns: a hero product line carrying 77% of revenue, a Q4 gifting spike at 3 to 4 times baseline, and a second demand cycle in January when new year resolutions sell journals and dated planner inserts. Because the data is generated, the repository can be fully public, and the pipeline demonstrates privacy-by-design anyway: the raw format includes the buyer PII columns that real Amazon reports contain (with fake values), and the SQL staging layer strips them so nothing personal ever reaches the analytical model.

The pipeline runs on DuckDB with all transformations in plain SQL: staging, a star schema with conformed dimensions, and an FBA economics mart that feeds the dashboard. A pytest suite of 39 tests guards the build, from seeded reproducibility down to revenue reconciling to the cent between staging and the fact table. The full build process, from design document to phased implementation plan to review gates, is documented in the repository.

Where Does a Snack Shop Actually Make Its Money? A Business Report in Google Sheets

Thu, 24 Apr 2025 00:00:00 +0000

Snack Attack Shack is a fictional snack retailer from the Masterschool orientation challenge. The official submission was a group project, but my place in the Data Analytics track rode on my orientation results, so I also built a complete version of my own: the report I would want to read if the shop were mine, covering where the money actually comes from, which customers are drifting away, and what’s quietly rotting in the warehouse.

The answers were concentrated ones. Twenty-five customers, 28% of the base, generate 69.6% of revenue; the top 25 products carry 68.2%. Regular customers drive 96.8% of everything, which makes the dormant segment a reactivation opportunity rather than a write-off. On the operations side: several products sat flagged for urgent reorder while discontinued items still held excess stock, and the carrier comparison split cleanly into a fastest option (Speedy Express, 8.6 days average) and a cheapest one (United Package, €8.76 average), which is a choice, not a problem.

Everything runs in Google Sheets, from cleaning through loyalty segmentation (new, non-returning, dormant, regular) to AI-assisted demand and profitability forecasting. Two logs in the repository document the work end to end: every transformation from raw data to final numbers, and exactly where AI assisted, because analysis you can’t retrace is just opinion with charts.

Read the , or explore the in Google Sheets.

Why Are a Quarter of Databel's Customers Leaving? An Excel Answer

Thu, 06 Mar 2025 00:00:00 +0000

Databel is a fictional US telecom from DataCamp’s case study “Analyzing Customer Churn in Excel,” which I took on during my self-study retraining, before Masterschool. The case ships with a suggested solution; I built the dashboard I would actually want on my desk if 26.86% of my customers were walking out the door.

The numbers tell a specific story. Of 6,687 customers, 1,796 churned. Nearly half of them (44.82%) left for a competitor, with customer-service attitude a distant second at 15.98%. Churn climbs past 40% for customers aged 79 to 88, concentrates in short-tenure accounts, and spikes geographically: California churns at 63.24%, wildly out of line with every other state. Each of those findings is one click deep in the dashboard, with dynamic filtering and an interactive state map doing the work a static report can’t.

Under the hood it runs on modern Excel: dynamic array formulas (LET, SWITCH, FILTER, PIVOTBY) instead of hidden helper columns, so the whole pipeline from raw data to dashboard stays inspectable in one workbook. The full transformation from raw export to final dashboard is documented step by step in the repository’s changelog.

Watch the walkthrough: