Improving the invoice processing process requires insight. Key Performance Indicators (KPIs) are a good starting point because they provide insight into potential opportunities for improvement. Provided they are meaningful KPIs. There are a number of steps you need to take to arrive at reports that provide the right insight.
To measure is to know; this well-known statement also applies to the Accounts Payable process. The goals of measurement are just different for everyone. For example, an Accounts Payable manager wants to know the processing costs per invoice, the number of invoices processed per FTE, matching percentages or the turnaround time of an invoice. For a finance manager, insight into the percentage of invoices paid on time is important, and for a buyer, the percentage of goods purchased outside of contractual agreements is of interest. But how will you measure this?
Defining the right KPIs and then getting these values out of the software system in the right way proves very difficult in practice for many companies. Lack of knowledge of internal practices or about the exact operation of the invoice processing system sometimes leads a process improver to look at a specific value, which ultimately says nothing about what he wants to know.
Variables
An example is when a manager aims to measure the productivity of his Accounts Payable employees by looking at the number of invoices processed per person per day and therefore wants to know who clicked the "Save" button how many times. However, this is too shortsighted thinking, because you then assume that the number of edits for that "Save action" per person is the same.
This way of measuring completely ignores the complexity of an invoice and the amount of handling an Accounts Payable employee must do to process it. For example, an Accounts Payable employee may spend much longer on an invoice with a thousand order lines than an invoice that has only thirty, but again, that depends on, for example, the portion that is processed automatically.
So in this case, it is important to get a good view of the amount of manual labor required to process an invoice in the system before using a particular value as a KPI.
Sensible KPIs
I also see in practice managers who want to know everything. This is usually because they do not know where to start. However, wanting to know everything quickly leads to too many KPIs and that is not an effective way of measuring: KPIs are of course no substitute for looking over one's shoulder.
Therefore, it is useful to have targets that are SMART in wording, linking them to sensible KPIs that are interrelated so that they interact with other processes and departments in the organization. This allows you to paint a coherent picture of a department or process in which you can quickly observe changes that require your attention.
Start with the right questions
How to do that? Starting by asking the right questions of yourself and colleagues: From whom does the question come and what is the basis of the question? Is the question about efficiency (process improvement) or effectiveness (purchasing results)?
Next, what are you going to use the data for: do you want to do something with it yourself or do you need to report on it? There is a significant difference whether you want to use the data to compare people's performance for reward or whether you want to use the data for process improvement.
You can use the processing time per invoice, for example, to see the effect of certain automation measures. But if you use that same processing time to target the employees whose invoices they process have an above-average processing time, it will probably backfire. By doing so, you will only encourage them to avoid, whenever possible, the exceptions that take time to resolve.
So what do you want to measure and with what objective? How do you then make the translation to your process and automation systems? What elements in your process say something about this and what assumptions should you then make by considering or not considering variables?
Human interpretation
Conversely, when using KPIs, whatever they are about, it is important to remember that they are indicators. In other words, you are not being handed a conclusion but information that requires human interpretation.
For example, you may have devised an indicator that measures the difference between an invoice date and first touch with the receiving software to develop an inkling of whether suppliers are antedating. Or you compare ordered, delivered and invoiced quantities of (bulk) goods to get an idea of suppliers who are "delivering/billing by margin. Or perhaps you've devised an indicator that alerts you when a link or application has not provided input for more than an hour or to measure the performance of a system.
In all these cases, you are not relieved of the duty to think about the KPI. Perhaps the apparently backdated invoice has been sitting in a colleague's drawer for a week before it was presented to the scanning line. It could also be that invariably slightly more was delivered than ordered, because a buyer in one's own organization is responsible for an ironclad inventory and therefore urges the supplier not to under-deliver. Or perhaps the applications were temporarily not linking because they were being updated. In short, realize that what you measure and at what point you do so determines what conclusion you can draw.
If you make a comparison with a benchmark number or an average or bandwidth measured over a larger whole, you can place the KPI in roughly three action categories: green: 'everything is going well', orange: 'could be better, but no acute action needed' and finally red: 'this seems to be going wrong, I must investigate immediately'. This prevents you from micromanaging on individual outliers, which doesn't really help anyone.
The signal function combined with sensible KPIs not only provides insight, but also regularity and calm in your daily work. Unless everything in your dashboard is lava glowing, but then you have other things to do than puzzle with KPIs.
