Organisations approaching scanning in accounts payable think first about hardware, and the largest improvement available is organisational. Three rules, free to adopt, improve extraction quality and elapsed time more reliably than a better scanner. They are simple to state and broken surprisingly often, usually because nobody wrote them down.
Once, and canonically
A document that is photocopied, annotated and then scanned exists in several versions, and the one in the system may not be the one somebody wrote on. Scanning first and treating that image as canonical means everybody works from the same document, and annotations become structured notes rather than marks on one piece of paper.
Early, at arrival
The received date is the date the invoice reached the organisation, and payment terms usually run from it. Scanning at arrival makes that date real rather than a guess, and makes the invoice visible to payables immediately rather than after internal post. The paper can then travel at whatever pace suits.
Consistently
One resolution, one orientation, straight, both sides where relevant. Consistent input beats a superior engine fed a mixture, every time. Where individuals scan on their own devices, a short written standard plus an occasional check is enough, and a capture app that rejects a poor image at source removes most of the remaining variability.
Questions people ask about accounts payable scanning
What resolution should we use?
High enough that small print and stamps remain legible when zoomed, tested on your worst supplier document rather than chosen from a default.
Should we scan both sides?
Yes. Terms and remittance details are often on the reverse and their absence surfaces at the worst time.
Who should scan?
Whoever opens the post, as part of opening it. Passing paper to finance to scan adds a day for no benefit.