An accounts payable automation case study is a marketing document, which does not make it useless. It makes it a document to read with a specific list of omissions in mind. Case studies are published because the outcome was good, they describe businesses selected for that outcome, and they consistently leave out the same five things. Knowing what is missing lets you extract the genuinely useful part, which is usually the description of the before-state rather than the numbers. This page is the five omissions, the questions that fill them in, and how to tell whether the study describes a business anything like yours.
The five things a case study omits
The implementation cost and internal time. The exception rate before and after. How long it took to reach the quoted state. What the baseline actually was and how it was measured. And the businesses where the same project did not go well.
The questions that fill them in
Ask the vendor for a reference at your size, in your sector, who went live at least a year ago. Ask what proportion of invoices still need a person. Ask what the implementation actually took in internal days. The answers are usually given honestly and are almost never in the published study.
Use it for the before-state, not the numbers
The most useful part is often the description of how the business worked beforehand, because you can compare that with your own. On the worked example on this site, 400 invoices at three touches of six minutes with a 12% exception rate is 134.4 hours a month: $4,300.80 at a $32 loaded rate, $10.75 an invoice and $51,609.60 a year. Compare their before-state with yours, and treat the after-state as a ceiling.
Questions people ask about accounts payable automation case study
Are case study numbers made up?
Usually not, and they are selected. A true number from an unrepresentative business is more misleading than an invented one, because it survives scrutiny.
What size business should a reference be?
Yours, within reason. A case study of a business ten times your size tells you about a project you are not running.
Is it worth asking for a failed implementation?
Asking what a difficult one looked like is fair and often answered candidly. Vendors who have thought about it give a better answer than the ones who claim there were none.