Notes · 21 July 2026
What the store actually earned, without another app subscription.
Shopify can tell you what you sold down to the cent. It cannot tell you what you earned, because the rest of the picture lives elsewhere or arrives late: ad spend in Google and Meta, overheads in the accounting software, cost prices that drift, refunds that land weeks after the order. Most stores answer this with a profit app subscription, or with a spreadsheet that stops being trusted about a month in.
Our own store answers it with a system built once and owned outright: Claude, n8n and Xero doing the stitching the profit apps charge a monthly fee for. This note is how it works, what it costs to run, and where its honest limits are.
The stack
Three parts. n8n, an automation tool at roughly €24 a month, pulls revenue and cost of goods from Shopify, overheads from Xero, and ad spend from Google and Meta on a schedule. Claude reads all of it and renders a live dashboard: true net profit by product and by channel, refunds and payment fees netted off. And Xero keeps the numbers honest, because the dashboard reads the same bank-fed actuals the accounts are built on, not a copy keyed in by hand. Nothing is typed into a spreadsheet, so there is no spreadsheet rotting in the background.
The maths
The popular profit tracking apps price by order volume, from about €30 a month at the small end into the hundreds at scale (checked July 2026). Call it €50 to €130 a month for a typical small store. The owned build's running cost is the n8n subscription, on top of a Claude plan the business may already have for other work. Net of that, roughly €300 to €1,300 a year stays in the business, and the logic is owned rather than rented.
The part no app does for you
The plumbing took a few evenings, and most of that time went on deciding what profit should mean: which costs allocate to which channel, how refunds are treated, what counts as overhead. Those decisions are the real work, and they have to be made when configuring any profit app too. No app makes them for you.
The honest limits
If a business does not already pay for Claude, the subscription eats a fair chunk of the saving; this wins when Claude is already doing other jobs. A profit app is set up in an afternoon, and this is not. APIs change, so once or twice a year something needs a small fix an app would have absorbed. And the in-month view is a guide rather than the final number: profit settles about a week after month end, once overheads, refunds and the payment fee reconciliation have landed in Xero. Mid-month tells you whether you are on track; the month-end run is the answer. The profit apps carry the same lag; they just do not mention it.
We now install a scope-locked version of this build into other stores as the Profit Dashboard: €400, standard integrations, one overall profit view, with a profit-definition worksheet settling those decisions before the install call. If you would rather build your own, the free Context Kit is the place to start, and the thinking above is most of the work.
Daniel · The Clarion