Start here: if you're sure a specific receipt is missing, reply to support with the sender, subject line, date, and rough amount. We'll trace that exact email through your Gmail and tell you what happened — already in your sheet under a different merchant name, deliberately skipped as a non-receipt, or a genuine gap in our matching that we'll go fix. One concrete example tells us far more than a total count, and real gaps do exist — that's how we close them.
That said, here's what we usually find when a spreadsheet looks emptier than expected.
One purchase = one row, from the confirmation
ExpenseBot logs a row from the order confirmation, payment confirmation, or invoice — not from every email around a purchase. A single Amazon order can generate six or more emails; only one of them is the actual receipt. So if you count every receipt-shaped email in your inbox, you'll always see more than your spreadsheet shows. That gap is expected, not a miss.
The look-alikes we skip on purpose
To avoid duplicate rows, we don't log:
- "Delivered: 1 item" / "Your order is on its way" — shipping and delivery updates (the order confirmation was already captured).
- "Did your recent order meet your expectations?" — surveys.
- "Your refund has been received" — refund notices (handled by the Refund Offset feature, not a new row).
- "Add a payment method" / "Information about your recent purchase" — account housekeeping.
- "Order your favourites again" — marketing based on past orders.
A quick way to check coverage
Open your spreadsheet and look at the merchant column — do you see the range of places you actually shop? Sorting by merchant (or counting distinct vendors) is the fastest health check. It's common for someone who expected only a handful of entries to find hundreds of rows across a couple hundred merchants once they actually look — the real total is usually many times higher than a first glance suggests.
Year-long historical scans
If you ran a 1-year historical scan, give it an hour or more depending on inbox size. We email you when it's done — wait for that before judging completeness. The in-app banner also shows elapsed time so you can see it's still running.
And again: if after all that you still spot a confirmation we missed, send it over. We'd much rather see the specific example and fix a real gap than have you assume nothing was captured.