Honestly: there is no universal number, and any page promising exactly “X minutes per invoice” is selling you someone else’s average. But your own number can be measured with ten invoices, and the mechanics of where savings come from are the same everywhere. Manual invoice handling splits four ways: keying data into the system, matching the invoice to an order or contract, waiting on approvals, and resolving errors and supplier queries. Automation hits the first two first (entry and matching) because they are pure document work. With 73% of accounting firms having implemented automation, the question is no longer “does it work” but “how much of your process is that kind of work”.
How do you measure your own savings with ten invoices?
Take ten consecutive purchase invoices and time three things: minutes spent keying data, minutes matching and checking, minutes waiting on and chasing approvals. Multiply by your monthly invoice count. That is your baseline – without it, every “savings” figure is marketing. The pattern you will likely see: entry and matching dominate when invoices arrive as PDFs; when suppliers send e-invoices, the entry step disappears at the channel level – structured data needs no reading. That is why the first automation win is often switching on e-invoice receiving, not buying software.
Where the savings come from, in order
Entry goes first: the document is machine-read, the entry is born as a proposal. Matching second: the system reconciles the invoice against the order and the bank movement itself and flags only exceptions; turning review into sampling. Third, approvals speed up because the invoice no longer sits in someone’s inbox but routes by rule to the right person. What remains after automation: exceptions, interpretation and supplier conversations when something is genuinely wrong. Do not plan that part to zero: it is the valuable part of the job.
When does automation pay for itself?
Price your measured monthly hours against the automation cost, and add the price of errors to the equation: correcting a manual entry costs more than the entry itself, because the mistake surfaces later. If your measured routine is under a couple of hours a month, skip automation: the joy will not repay the setup. If it is measured in working days, the answer is already written in the equation.