Write down the 6-10 numbers you check every week (for example orders, sales, average order value, returns, ad spend, cost per sale, new email sign-ups, best-selling products) and where each one comes from.
From your shop, ad accounts and email tool, export last week and the week before as CSV files (spreadsheet files). Most have an Export or Download button on the reports page.
In Claude, create a Project called 'Weekly numbers'. Paste the 'Report rules' prompt into the Project instructions so every week's report looks the same.
Upload the files into a new chat in the Project and type 'This week's report'. Claude builds the table, the changes and the three things to look at.
For the first month, compare each figure in the report with your shop's own dashboard. If something doesn't match, tell Claude which file or column it should use.
Every Friday afternoon, download the files and run the report, so it's waiting on Monday. It takes about 10 minutes.
Once the numbers match for a few weeks, check whether your shop or spreadsheet has a Claude connector, so Claude can read the figures directly instead of you uploading files.
You make the weekly numbers report for [business name], a [what you sell and where]. Each week I upload this week's and last week's exports. Calculate: [your list, e.g. orders, sales, average order value, returns and return rate, ad spend, cost per sale, new email sign-ups, top 5 products]. Show: 1. A table: number, this week, last week, change in % (up/down arrow). 2. Three short lines: what went well, what got worse, anything odd in the data. 3. Three things worth looking at this week, each linked to a number above. Use only figures from the files. If a number can't be calculated, say which file or column is missing. Currency: [MYR / USD / ...]. No forecasts unless I ask.💬 Explain a change
In this week's report, [number] went from [x] to [y]. Using only the uploaded files, break it down: which products, days or ad campaigns account for most of the change? Give me the top 3 reasons with the figures, and say clearly if the files can't explain it.
From https://usecasy.com/use-case/claude-weekly-kpi-report/ · Guide checked 26 October 2026 · Prices are rough and shown in USD unless stated (MYR where relevant).