AI solves assignment problems faster than humans - Part 1
Fully exploiting efficiency potential
You have been driving the automation of your financial processes for years. And then you realize: there are still annoying losses due to inefficiency. For example, invoices that should process automatically but still get stuck in manual workflow loops for 2 to 5 days. Almost every company recognizes this phenomenon. One of the most common causes is the absence of a clearly identifiable contact person to whom the invoice should be routed in the workflow.
As a result, inefficiencies can increase several times over, depending on how many loops the invoices go through on average. The traditional solution is still manual work, where someone guesses the likely contact person, enters it and sends it on. That takes time. Often the hypothetical assignment is also wrong, so the invoice goes through yet another loop.
How helpful are assignment tables in accounting?
Assignment tables are designed to help with the automatic assignment of invoices that are not related to an order, for example. This is a good idea, but has its own problems: First, the assignment tables must be maintained manually. Second, the routing defined in the tables is not always accurate. As a result, assignment tables sometimes, but by no means always, help clear up ambiguous invoices. Still, even with assignment tables, invoices end up with the wrong recipients far too often.
The first step to improvement: recipient analysis with avvaneo Discover
Recipient analysis is a feature in avvaneo Discover that provides accurate "self-knowledge". It shines a spotlight on this specific source of reduced efficiency. The scope is derived from thousands of individual cases that, on casual inspection, do not appear to be associated in any systematic way. But usually, the recipient analysis reveals quickly where the potential for improvement is hidden.
For example, you will find answers to analysis questions such as:
- How many workflows were there in the selected period where the invoices were forwarded to the wrong recipients (in percent and absolute numbers)?
- How many days did it take on average to find the right recipient and deliver the invoice correctly (average and total for selected period)?
- Which suppliers' invoices most often end up with the wrong recipient?
- How many days does the average correction process take per supplier?
The answers are based solely on your current data, i.e., they analyze what is actually happening. The dashboard offers filter options to further narrow down the serious problems.
In this example with fictitious data, you can see: addressing errors of almost 10% which would be too much. With every single misrouting, the total effort in processing incoming invoices increases, time is lost, and the company loses money which need not be lost.
Considering all the traditional methods to reduce the error rate, there is now a way that works automatically by itself and even automatically gets continuously more effective: The use of self-learning algorithms, i.e. artificial intelligence. Our AI solutions for the simple and sustained reduction of assignment errors is already in use. It leads to rapid improvements. More on this in the second part of our blogpost mini-series coming up next week.