Data Analytics where Topmanagers should keep an eye on
This four-part series deals with the analytics of very high incoming invoice amounts. In an audit of the purchasing process, you want to be quickly familiar with the risk-based transactions, because: high amounts of invoices = major potential risks. I will show you what is scientifically justified by high amounts of invoices and how to recognize them. With this background knowledge, you can focus on the most material and significant business transactions. This means for you: More efficiency in the next SAP Audit.
Part 1 of the series: “Very high incoming invoice amount”
1. Where Topmanagers should keep an eye on
2. Getting serious: What are high incoming invoices?
3. Analytics of high incoming invoices
4. Two graphics for evaluating high incoming invoices
What should I have my focus on?
Whether in the context of an SAP audit or the operational review of incoming invoices, top managers and auditors should ask themselves: What are the “big fishes” that should be looked at in any case? How do you get the unusually high incoming invoices? Of course, you can simply sort by the amounts of all incoming invoices and then get a list of the largest invoices – but: are they “unusual”? How can one define “unusual” at all? Can there be invoices with more “normal” amounts, which are “unusual” for the specific vendor? Let us think through this question in a structured way! I would like to explain how I would do it in order to analyze the SAP audit meaningfully. After this series you know:
- how an unusually high input calculation is scientifically defined,
- how unusually high invoices are identified,
- and what should be considered in SAP data analytics of unusually high invoices.
How do you recognize unusually high incoming invoices?
First, you should not try to find unusually high input invoices in the complete totality of all input invoices. What you should do is: divide the basic population with respect to a particular context and then search for exceptionally high incoming invoices in the respective context. The obvious point here is to divide the total population by the individual vendors and then search for exceptionally high invoices within the vendors. The reason for this is simple: one can rather assume that the probability distribution of the invoice sums within a vendor is more stable than invoices of all vendors, when they are examined together and the distributions of the invoice sums of all vendors are mixed with one another.
The basic totality of the incoming invoices is therefore divided according to the associated vendors. This process is called stratification within the statistics. Within each (vendor) layer, it is now evaluated which input statements are “unusually” high. But what is a “reliable high”?
To answer this question, it is sufficient to attempt simple basic concepts of descriptive statistics. We consider an “unusually high” when an invoice with respect to its sum can be regarded statistically as an outlier.
In the next article of this series, you will learn how to identify these “outliers”, then follow these transactions in more detail.