Online Retail II

A UK online gift wholesaler, December 2009 to December 2011. Every number on this site is read from the dbt marts in this repository — nothing is typed in by hand, and the page rebuilds from dbt build.

Revenue over time

The extract stops on 9 December 2011, so the final month is not a month. Drawn in, it would show a fall of -69.1% and a return rate of 28.3% , both artefacts of the same thing: November's returns land in December while December's own orders mostly do not exist yet. So it is left out of the chart above.

The mart carries an is_partial_month flag for exactly this reason. A dashboard that plots the last bar of a truncated file has told the reader the business collapsed.

Whose money is it

Roughly a fifth of the lines carry no customer id. The usual first line of an analysis of this dataset is WHERE customer_id IS NOT NULL, and it quietly removes

So the denominators here are deliberately different. Revenue is measured over every line. Retention is measured only over customers who can be followed across time — because with no id there is no way to tell two visits from two people.

Best sellers

Postage, bank charges and stock adjustments are not in this table. They are real money — 62 codes and −£95,225 of it — but nobody ordered them, and a best-seller list that keeps them ranks postage first. They are removed once, in fct_order_lines, and the removal is shown on its own page.