Home » Case study » How quality in freight forwarding can be delivered faster and more efficiently thanks to RPA
LINK International Transport uses Robotic Process Automation tools to speed up the accounting processes associated with invoicing. The use of software robots meant that accountants no longer have to work like robots.
The Customer: LINK International Transport
The number of robotic processes: 6
Solutions applied: UiPath
A measurable benefit: 75% (and over time, thanks to machine learning, more) of correctly read and automatically booked invoices.
LINK International Transport specializes in road transport and logistics services. The company, which has been present on the market for 30 years, provides full and partial truck transport services in Europe. LINK’s operations centre is located in Wiązowna near Warsaw, and transport bases are located in Gliwice, Łódź, Poznań and Wrocław. Employing over a thousand employees and managing a fleet of over 1200 transport units, the company executes nearly 61 thousand orders annually.
Freight forwarding, logistics and transport are one of the key drivers of the Polish economy. We are a real European leader in this area. Despite promising results and forecasts, no one in the industry can rest on their laurels. Companies continuously strive to reduce costs and optimize processes, increase productivity, efficacy and operational efficiency. The acceleration of even seemingly irrelevant business processes translates into profits in other areas. When you can act faster, you can act cheaper and gain a competitive advantage as a result.
Every day, tens or sometimes hundreds of invoices from several thousand carriers with whom the company cooperates flow into LINK International Transport. In total, the company receives about 50 thousand invoices annually. The worst thing is that it is not evenly distributed over 365 days. At the beginning and the end of the month, there were peak periods in which employees of the four-person accounting department literally could not free themselves from invoices. Practically every day, three of them had to devote part of their work time to handle invoices, namely to enter data taken from paper documents into two IT systems.
“With so many documents, efficient work management is not enough. Automation is needed. The employees of the accounting department were tired and frustrated as a result of the constant rewriting of data, especially as other, more demanding duties awaited them. That’s why they took the initiative to automate the process in some way,” says [LINK representative].
A short market analysis indicated that the best solution would be to work with Mindbox, a partner of UiPath, a leading provider of Robotic Process Automation (RPA) technology.
Mindbox developed a solution based on Abbyy’s Optical Character Recognition (OCR) technology and UiPath robot that enabled automation of cost invoice from carriers. Paper documents first go to the secretary’s office, which applies a bar-code sticker and enters the scan of the document into the system where OCR technology reads the data recorded on it. Although it might seem like a simple task, it is not.
“Most contractors have their own invoice standard. Only in individual cases do models of documents repeat themselves. The system must verify whether the invoice is actually intended for LINK International Transport – mistakes do occur. The currency in which the payment is to be made is also checked on the invoice. At the initial stage, it is also checked whether a given contractor exists in the database at all. This is a security measure in case somebody tries to make a scam”. –enumerates [Mindbox representative].
An additional complication is a fact that invoices are issued by foreign contractors of the reverse charge VAT0 type – this also verified by the system. At the same time, the system verifies whether we are dealing with cash accounting in a given case, and also detects documents correcting previous invoices. The constructed robot can even cope with exceptions in which the basic currency of the document is euro, but VAT is paid in PLN to a completely different account. In this case, you need to read and verify two bank account numbers.
“The invoices read by OCR go in the form of structured XML files into the communication interface with the forwarding system used by LINK International Transport, where all transport orders are stored. There, the XML file, which has been prepared by the OCR system, is supplemented with appropriate fields. One of the parameters that OCR reads is the order number. This allows to link the invoice to a specific order and contractor,” says [Mindbox representative].
Further enriched XML document is handled by the robot, which is responsible for its booking. The bot logs on to the SAP system and processes a package of documents collected that day according to the rules. He searches for the invoice, fills in the appropriate fields, checks if the bank account number on the invoice matches the one in the system and then sends the document for approval after processing. The robot also checks if there is no difference between the data in the forwarding system and the invoice – in this case, it returns the complaint to the accounting department.
The important thing is that the robot collects information from each stage of the automated process, also during the OCR reading and document enrichment in the shipping system, and then creates a report analysed by business employees.
“One of the challenges in automating this process was to implement a mechanism that will close a fiscal month on a different day than the last calendar day of the month. For example, April this year was closed on 5 May. This was solved with a configuration file, in which the month’s closing day is manually determined. Based on this, the robot does the accounting,” says [Mindbox representative].
Of course, OCR is not able to guarantee a 100% efficiency of reading data on invoices. Human involvement at this stage is necessary – the operator checks and verifies the correctness of the recorded data. Thanks to this, it is acknowledged that the data that goes to the robot performing accounting is correct.
Although automation did not allow to eliminate human interference altogether, the time spent by accountants on tasks related to invoices was significantly reduced. The previous process required 100% human involvement at each stage. Now it only applies to exceptional situations, which could not be described by the logic of the robot’s work.
“The implementation of an OCR (Optical Character Recognition) solution is a breakthrough. Until recently, every scanned invoice went into accounting. Employees opened a document image and entered data into two applications. Now their role is only to check if the characters read to match the original and click the Enter key. If they detect a mistake, they only graphically mark the error area. The solution is very user-friendly. Anyway, if OCR has doubts concerning a field, it also informs the operator. Then the operator checks the field and can easily change the values”. – says [LINK representative].
Currently, the correctness of data reading is 75%. This means that a person has to correct automatically read data on every fourth invoice. Over time, however, the reading efficiency will increase. The system is enriched with machine learning algorithms – the more invoices flow through the system, the more effective the model is. The error rate will decrease with time, although it can be assumed that it will never reach zero.
The system is so intelligent that when a new supplier arrives, up to 3 to 5 documents are enough to learn how to read each subsequent document correctly. This means that at the learning stage, the operator sometimes has to indicate some field on the document manually.
The automation of the invoicing process is, of course, only the beginning of LINK International Transport’s adventure with RPA. The success of the first robot encouraged the company to automate further processes. Preparations are already under way to robotise two more processes in accounting.
Interestingly, the project was implemented at the peak of the COVID-19 epidemic. “Despite the limitations of the epidemic, the project was executed very efficiently. It was carried out within three months. All the time, the project team, consisting of representatives of LINK and Mindbox, remained in remote contact, without having to visit the company headquarters. Even in such exceptional circumstances, the project was successfully completed on a high level.