22.10.2021 | Reading time: 5 minutes
Do you know the situation of having to type data from an invoice or other paper document into a system, for example for the customs declaration? Unfortunately, this is often time-consuming and also error-prone, because the currency can be mistyped or a number error can creep in with the invoice values. Digitization in customs processing [?]
Do you know the situation of having to type data from an invoice or other paper document into a system, for example for the customs declaration? Unfortunately, this is often time-consuming and also error-prone, because the currency can be mistyped or a number error can creep in with the invoice values.
Digitization in customs processing is advancing all the time. While some are already looking at the most innovative technologies of artificial intelligence (AI) and the digital mind set of the future, paper documents and their data transmission are still the order of the day in many offices, which is why many business processes are caught between analog and digital. Especially in customs processing, many documents – delivery bills, packing lists, T1 documents, invoices, etc. – are still passed on and processed in analog form. So how can we simplify life here without spending the day transferring data?
The solution is called OCR (Optical Character Recognition) technology. If you’re saying to yourself, I don’t understand what you’re talking about, then read the following article and get an answer to what OCR actually is and what it does exactly.
What exactly does OCR mean?
OCR technology refers to automatic data extraction that makes it possible to recognize letters, numbers and other characters from scanned documents or image files and convert them into text documents. OCR technology, in combination with artificial neural networks and intelligent data processing, thus provides the basis for digitizing customs processing. In simple terms, this means that you receive a PDF document, e.g. an invoice. This document is fed into the “OCR engine”. This automatically extracts data that is important for a customs declaration.
How can OCR technology be used in customs processing?
Although OCR technology can recognize many things right from the start, it still needs to be trained in the first step. Especially for small to medium-sized companies, the ready-made solution approach is the better alternative. In this case, an already existing construction kit is used, which, however, still has to be adapted to the individual needs. Therefore, it is necessary to import many sample documents into the system in order to train the technology. Similar documents that occur in large numbers can be trained well. So if your company always has the same invoice layout and you get a certain mass together, this is helpful for a good quality of data extraction.
It is also important to look at the entire value-added process and involve service providers if you want to use OCR technology. In the best case, documents are sent directly from the system and not further processed as a scanned variant.
What the technology cannot solve is the possible poor data quality of the documents themselves. If incorrect invoice values and customs tariff numbers are listed, the best technology is unfortunately of only limited help. Therefore, the master data quality should be checked first.
Do we still need OCR technology? The future of customs clearance is paperless anyway!
Practice shows that there is still quite a gap between wish and reality when it comes to digitization.
In the next five to ten years, a tool such as OCR recognition can still accelerate and improve processes. Already in about one to two years, the technology will have reached a very high level of maturity, where it can be assumed that completely new documents can be electronically processed and made available in structured format from day one without any or only little human intervention by virtual machines (according to Borisav Parnakovic from Digicust).
Another reason for using ORC technology is the complexity of customs processing. With such a large exchange of information and documents between many different parties, the result is a large volume of PDF documents. Getting these in a structured XML or JSON format from the start is often a pipe dream.
Borisav therefore believes that AI-based OCR recognition should be seen much more as a driver of digitization, and that it will continue to play a role for some time to come due to its flexible deployment capabilities.
What to look for when choosing an OCR vendor?
There are a number of criteria to consider when selecting an OCR provider. In order to use OCR technology successfully, the provider’s in-depth industry experience is an important point. The combined knowledge of customs and IT in one company can circumvent any communication and expectation problems on both sides.
Learning ability and machine validation is another important aspect to achieve automation, which must work well with the chosen software. The faster the AI learns through feedback, the faster it is possible to send the data to the next systems in an automated way or through only a little manual validation.
An easy-to-use user interface (i.e., what you see on the screen interface) is also not to be underestimated for the application. The less effort the user has to put into manual validation, the better the customer experience. Of course, this requires not only a great software interface, but also an AI that learns all possible configurations through the slightest human influence.
A main criterion for you is certainly also the price. However, it is important to differentiate here. The problem is that it is often difficult to compare between providers. Therefore, we recommend having documents tested to see how good the “standard technology” already is without having to make adjustments. The complexity of the documents also plays a role. If you want to extract few information and header data from always the same document layouts, a template-based approach is the right one. However, for commodity items and a large number of different document/invoice layouts, this approach does not work well, so 30% to 70% in additional costs must be applied. Therefore, for more complex projects, an AI-based OCR is the only solution to achieve efficiency gains. Highly complex projects such as customs processing also require additional services. The more complex the project and application, the more expensive the product.
It is important to note that there are different billing methods. Some providers charge for document processing services per document, others per page.
If the technology provider has a suitable process for data delivery, the customs service provider will feel efficiency gains right from the start and can enjoy a daily increase. The important thing is that the customs service provider’s effort is kept to a minimum. This is particularly important if the digital transformation is to be realized successfully.
What do we take away from the topic of OCR technology?
Transcribing data from e-mails and the like is an error-prone and time-consuming process. OCR technology can provide a remedy. Important are process chains as a whole, good data quality and a considered selection of the service provider. Although OCR technology may eventually be superseded by other technologies such as blockchain, etc., it is a good interim solution. After all, as with the Digital Mindset, it’s not just technologies that need to evolve, users need to evolve here too. OCR can be a good way to directly improve efficiency.