The software-vendor community argues energetically that any large company can save a bundle by automating its accounts payable (AP) processes, particularly those surrounding the processing of invoices for payment. How big could that bundle be?
To answer that question, APQC studied 116 organizations, each with more than $1 billion in annual revenue. The process-efficiency data on these firms comes from APQC's Open Standards Benchmarking database.
The first step placed organizations into one of four buckets, each containing 29 firms (29 X 4 = 116). Each bucket reflects a step up in the level of success in matching invoice line items the first time (Figure 1). Note that this analysis only applies to invoice line items requiring a two- or three-way match, and that it excludes line items for freight and tax. The performance gap between the best and the worst is alarming but, alas, not surprising.
In bucket 4, the best of the best performers have a 99 percent match rate, thanks to an extremely high level of automated communications, typically the machine-to-machine kind known as electronic data interchange (EDI). When EDI is utilized, an ERP system sends a purchase order to the supplier's ERP system, which in turn spits back a mirror image as an invoice.
See a larger version of Figure 1. Performance Gap: Best vs. Worst.
"EDI is the ultimate touch-less event," says Henry Ijams, managing director of PayStream Advisors, an advisory firm that assists organizations that want to improve their procure-to-pay processes. "But because EDI is so well-defined [it was originally developed for manufacturing companies], it doesn't work for everything. So, you have to tightly manage your procure-to-pay process to make everything work properly. You have to order correctly. The supplier has to ship properly. And then you have to receive properly."
Ijams points out, "Even if you are using other forms of automation, for example, an electronic invoicing network, the people in AP will still have to intervene and resolve discrepancies from time to time. The AP people are not the ones who mess up in the first place, but they have to fix broken invoices, which are inevitable even in the best-run operations."
That explains the range of 85-to-99 percent for first-time matches in bucket 4. When it comes to the cost burden of having to manually resolve 15 percent of invoice line items, Ijams says, "it's important to keep in mind that touch-less transactions typically represent only 25 percent of the total cost. It's the dirty invoices that send your costs soaring."
Examining bucket 1 reveals the best first-time match rate of 36 percent. In other words, a woeful two-thirds of all line items must be handled manually. The impact of all that manual processing on the overall cost of an AP function is startling. Remember that the labor component of total finance costs hovers around 60 percent (Figure 2). This includes all possible costs associated with the purchase of goods or services; for example, labor, systems, fees to external service providers, etc.
See a larger version of Total Cost of AP per Number of Invoices Processed Annually.
The comparison shows that the median cost of the AP process (measured in terms of cost per invoice) for bucket 1 is more than twice the median cost for bucket 4. (The median reflects the value above and below which an equal number of responses fall.)
Imagine the typical large organization with revenue upwards of $1 billion. Business model aside, you'll likely find that company processing more than 500,000 invoices per year. Clearly, it pays to invest in touch-less AP processing. However, don't forget that it also pays to invest in sound end-to-end business process management, including regular measurement of process efficiency and analysis of the root causes of quality breakdowns.
Mary Driscoll is a senior research fellow with APQC, a nonprofit benchmarking and best-practices research organization. She is a regular contributor to Business Finance.