A company orders ten packages of coffee beans from its supplier with delivery date X, total price Z and logs this into the ERP system. Shortly afterward, the order confirmation (OC) flutters in. Now the OC is checked and compared with the order in the ERP system – oh, what a shock – the OC says “ten packs of ground coffee, delivery date Y, total price V”, which causes problems, not only because the freshly ground coffee simply tastes better, but logically also because the data in the OC does not match that in the order. Admittedly, this example is banal. However, the difference is not much different even with x-times more complex order confirmations, as long as such discrepancies in data between the OC and the order have to be laboriously reconciled manually in the ERP system. What the company needs is automated checking and matching of order confirmations with purchase orders.

Do the items listed actually correspond to what was previously ordered? With this concern, the order confirmation provides clarity

Getting rid of monotonous activities

First of all, it is not a question of completely replacing humans, but rather of freeing them from monotonous tasks, such as manually comparing the order confirmation with the order in the ERP system, and only allowing them to interact in the event of errors. To ensure that this human intervention remains as minimal as possible and is limited to real discrepancies, the first step is to extract all relevant data from the order confirmations accurately.

The more accurate the extracted data, the higher the probability of dark processing

Intelligent, AI-based OCR as an essential step

The company sends the order confirmations to Parashift by mail, and they are received as PDFs. Now, the intelligent, AI-based OCR (Optical Character Recognition) automatically extracts all relevant item data, including the order number, supplier, delivery date, total price, internal and external item numbers, and other special features, if any, from this PDF. The data extracted by the intelligent, AI-based OCR is qualitatively enhanced, classified and sent back to the company as structured data.

Parashift takes care of the factually correct position data, the company takes care of the technically correct data

If position data still cannot be recognized automatically, the human in the loop comes into play and ensures that the factual position data from the PDF is correct and only then is it sent back to the company. The company thus benefits from a higher level of shadow processing and at the same time can now be sure that any discrepancies between the order confirmations and the orders in the ERP are technical discrepancies and not factual ones.

Whatever the specifics may be – with Parashift the extraction of this data is feasible

OC cannot be assigned due to discrepancies

Coming back to the example with the coffee beans, if the OC could not be assigned due to the discrepancies, the automated check would show the following errors:

  • Coffee ground instead of beans -> here the article does not match
  • Delivery date Y instead of X -> here the delivery dates are different
  • Price total V instead of Z -> here the agreed prices do not match

In daily business, these errors between OC and order are then conveniently highlighted so that the employee recognizes them immediately and, for example, only has to decide between the following procedures:

  • Recheck OC and purchase order in ERP
  • Process manually
  • Cancel
  • Complete processing

It is almost always the case that companies manage articles in their ERP system, in our example the coffee beans, under their internal article number, but the supplier logically manages the same article under a completely different number (from the company’s point of view with an external number). In order for the matching between OC and order to work properly, it is important to show both the internal and the external article number on the order confirmation so that these can then be extracted.

Internal and external article numbers are extracted so that the reconciliation works properly

Clear added value with automated OC verification

The advantages and added values that can be realized for companies with automated order confirmation checking are therefore clear:

  • Manual intervention only necessary in case of deviations in the technical data or complete absence of order numbers
  • Deviations are directly visible
  • Corrections can be made quickly or the necessary steps initiated
  • Employees no longer have to deal with the monotonous matching of order confirmations and purchase orders.
  • Employees have more time again for value-adding and essential tasks

What works for order confirmations can of course also be used, for example, for the automatic checking and matching of delivery bills against orders, invoices against orders, and so on and so forth