Keep value and
volume in view.
Track both revenue and pack units when reviewing product performance. Investigate price and pack mix before changing the assortment.
PUBLIC DATA, A LITTLE CURIOSITY, BETTER QUESTIONS.
Unwrap the insights.
Beneath every number is a story.
A chocolate-inspired exploration of sales, promotions
and the decisions behind them.
FERRERO ROCHER SALES · 2025
Independent portfolio studyPYTHON / SQL / TABLEAU
By Mila Ho
Independent portfolio studyPYTHON / SQL / TABLEAU
By Mila Ho
THE COLLECTION / 01-03
Three layers of
understanding.
Start with what happened, look a little closer,
then ask what to do next.
01 / UNDERSTAND PERFORMANCEAUSTRALIA · FULL YEAR 2025
A bestseller has
more than one measure.
The 32-piece pack leads recorded revenue. The 16-piece pack leads units sold. Two valid answers to two different business questions.
Recorded revenue110.37M
Unspecified dataset currency units
Product units sold288,641
Packs / units as supplied
Weighted margin rate32.51%
Sum of supplied margin ÷ revenue
Sales records2,777
Supplied rows, not distinct orders
PRODUCT MIX / RECORDED REVENUE
Australia baseline · product packs have different prices and sizes.02 / QUESTION THE PROMOTIONAUSTRALIA BASELINE
A bigger discount.
A better outcome?
Look beyond promotional revenue totals. Compare volume per record, revenue per record and the mix of products being sold.
NO PROMOTION / 0% DISCOUNT104.2units per record
vs.47,312 revenue per record
433 supplied recordsBUY1GET1 / 50% DISCOUNT104.5units per record
25,778 revenue per record
460 supplied recordsREAD WITH CARE
Open the workbook and choose “Promotion Analysis →” ↗Similar observed volume per record does not prove that a promotion worked - or failed. Discount directly enters the revenue formula; product mix and selection differences also matter. There is no control group or promotion-spend data here.
EXPLORE THE EVIDENCEOpen Tableau in a new tab ↗
Your turn to explore.
Filter by country and navigate between Sales Performance and Promotion Analysis inside the workbook. The editorial findings on this page describe Australia.
▥
The interactive workbook
Two dashboards, seven markets, your next question.
Loads an external Tableau Public view.03 / SHAPE THE NEXT DECISIONFROM OBSERVATION TO ACTION
Good insights end
with better questions.
Keep value and
Test before
scaling discounts.
Compare the same products and channels, then design a controlled promotion test. Measure incremental volume and contribution after costs.
Validate the
business context.
Confirm currency, retailer geography and channel definitions. The public dataset contains labels that need verification before commercial use.
What this dataset "can" and "cannot" tell usClick to expandClick to collapse
This is an independent analysis of a public Kaggle dataset, not verified Ferrero company data. It contains 20,000 records across seven countries in 2025. Currency is unspecified; monetary comparisons across countries are not economically validated. Margin is supplied, with no independent cost basis. Retailer labels do not identify physical stores. The promotion comparisons are observational, and one year cannot establish recurring seasonality. No findings here demonstrate causality or promotional ROI.
BEHIND THE WRAPPER
From raw rows
to a clear story.
A reproducible workflow: prepare the data, define the measures, validate the model and communicate the result.
Prepare & validate
Convert Excel dates, standardise labels, document missing promotions and reconcile revenue and margin formulas.
Read the notebook ↗Model & investigate
A sales fact table, six dimensions and seven reporting queries. Joined row counts and country totals reconcile to the cleaned data.
Explore the SQL ↗Visualise & explain
Interactive performance and promotion dashboards, with weighted measures and a consistent country filter.
Explore the dashboards ↗A LITTLE CURIOSITY GOES A LONG WAY.