Analysis for people who sell things online.

Every store dashboard shows revenue, because revenue is the one figure the platform holds completely. Margin needs cost data the platform does not have, and shipping cost is often in a third place entirely.

That is why the most common surprise in ecommerce analysis is a bestseller that loses money on every order: it is highly visible in the revenue chart and invisible in every margin chart, because there is no margin chart.

Growth from new customers and growth from returning customers look identical in a revenue chart and behave completely differently over a year. A store growing on new customers alone has to keep buying them.

The number to watch is what share of orders come from people who bought before, tracked monthly. It moves slowly enough that a single month tells you nothing and a six-month trend tells you a great deal.

A portfolio refund rate is nearly useless: it averages a fine product with a broken one. The useful version is refund rate by product, compared against the catalogue average, with the products that stand out named.

A product refunding at three times the average is usually a listing problem, a sizing problem or a quality problem, and which one it is can normally be read off the returns data itself.

An orders export first. Every figure in the report is computed from one table, and with an orders export that is the one holding revenue, order volume and repeat purchasing. A returns export and a product cost sheet can go up alongside it: each is read, converted and handed back as a download, and the report names any relationship it can see between them, but nothing is computed across two files.

It does not need to be tidy. Merged cells, several tables on one sheet and four date formats are the normal case.