Gross product sales: noncancellation rows with positive quantity, positive unit price, and five-digit product StockCode optionally followed by one uppercase letter. Missing CustomerID remains included. Line value rounded to GBP pennies HALF_UP; sum integer pennies. Orders count distinct InvoiceNo among eligible rows. Negative/cancellation rows, nonproduct codes and nonpositive prices remain separate exception counts; counts overlap. This is a reproducible dataset metric, not audited revenue or net sales.

Country groups: January top three by eligible gross product sales, plus Other countries, held fixed in February. Products: each month top three product codes by eligible gross product sales, descending, code as tie breaker.
January history has December and January; February adds February. No missing prior month is invented.
Cause commentary is unsupplied: human input required; numeric changes do not prove causes.
Data attribution: Chen, D. (2015). Online Retail. UCI Machine Learning Repository. https://doi.org/10.24432/C5BW33
License: https://creativecommons.org/licenses/by/4.0/
Changes: selected Dec 2010-Feb 2011 rows; computed declared aggregates; no source row was edited. Our own presentation template is used.
