Intelligent recycling

In common with many large organisations Steinbeis Temming has experienced major problems in managing its invoice processing effectively. Peter Turnbull explains how SER has worked with the recycling giant to overcome the inefficiencies in its financial operations.

Entry processing is widely recognised as a key area where bottlenecks frequently occur within a company's workflow processes. A large accounts department for example will typically handle a mass of different documents which must be examined scanned and sorted according to supplier and document type. This is hugely time-consuming repetitious and tedious leading to low staff motivation and a high error rate with resulting delays in processing and increased costs.

This was precisely the problem faced by German recycling company Steinbeis Temming Papier GmbH & Co as it sought to improve both the internal efficiency and customer service of its accounts payable department. Located in Glückstadt near Hamburg Steinbeis Temming is Europe's leading manufacturer of recycled paper with 800 employees and producing 258 000 tons annually - the equivalent of 30 tons per hour. Recognised as a global leader in its sector the company has invested heavily in state-of-the-art technology producing coloured papers for a broad range of uses - from office stationery to advertising - and is a valued partner in establishing ecologically sound paper recycling facilities worldwide.

Yet like many such leading edge companies Steinbeis Temming operated a manual process for dealing with invoices a labour intensive and time-consuming activity that had become cumbersome and out of step with its highly efficient manufacturing processes. It received around 300 documents daily - mainly though not exclusively in paper form from approximately 3 000 suppliers - distributed manually to the accounting team with invoice data entered into the company's SAP/R3 financial accounting system by hand.

Other problems highlighted the unresponsive nature of the Steinbeis Temming system and the need for drastic improvement. To answer questions regarding the payment status of an invoice for example it was necessary to search through the data recorded in SAP/R3 which only displays data anonymously with no visibility for invoice numbers or dates. Processing and searching was thus made difficult as the name of a supplier for a particular invoice could only be determined by opening the document.

In response the accounts department had for some time been looking for a suitable method of automating the process. "Initially we examined more conventional rules-based approaches from a number of suppliers confirms Hermann Vorrink, organisation and data processing director, Steinbeis Temming, but these did not offer a practical solution for us as the programming of complex rules took too long. Indeed we saw that the regular adaptation of rules to our constantly changing business processes would become a cost trap as the savings achieved through automation would in part be dissipated in amending the rule set."

For Steinbeis Temming the answer to achieving significant process improvements across its financial operations lay with SERware a knowledge-enabled eBusiness solution from SER Systems. At its heart lies SERbrainware the company's award winning neural network-based knowledge management technology. SERbrainware is an intelligent learning engine that reads analyses and acts on text with human-like intelligence - in essence it automates the process of dealing with the daily stream of new information by classifying storing and sorting information in ways which were previously only possible if individuals read it. Critically it learns by example: It analyses both structured and unstructured text from any media type and is trained by people who show it how they would read and make decisions on the basis of available information.

Key benefits

In putting forward a convincing business case for such a knowledge-enabled solution SER has recognised that discussion of such benefits as productivity and efficiency gains improved customer service and better information management though important and attractive to line managers are in themselves insufficiently quantifiable to be the sole financial justification for purchase. As part of a proposal therefore SER typically prepares a full ROI analysis - including a comparison of current staffing levels with those needed with the SER solution and the payback timeframe compared with more conventional business solutions - showing an early and dramatic impact on the bottom line across say a postroom or accounts payable function.

Steinbeis Temming already had an established relationship with SER having implemented a document management system in 1997 including a certified SER IS/3 interface to the company's existing SAP archive link. Documents are no longer stored in binders and kept on long space-hungry shelves but stored on optical disks - scanned in at the central IT workstation and then available for booking. Historical documents including COLD data from the company's MAS sales software are archived in the SER document management system with documents from the current and previous fiscal year held in the SAP system.

The use of electronic DMS had clearly brought benefits in speeding up the filing and retrieval of existing documents but alone could not address the manual sorting of incoming invoices. Booking such relevant information as the invoice number date date sent net value and sales tax still had to be entered by hand onto the SAP/R3 booking screen a process which continued to take up to six hours per day.

Following the recent introduction of SERware however invoices are automatically captured classified and analysed as soon as they are received. Specific data fields are then extracted from the invoices and instantaneously fed into the SAP workflow for further processing within Steinbeis Temming's SAP system. Invoice images are archived in the existing SER document repository and are immediately available from within the SAP system together with other related documents. The result is that the SER solution leverages the capabilities of SERbrainware to classify and identify the correct data fields for input into the SAP system with unprecedented speed and accuracy.

Fast implementation

In May 2000 a pilot programme was introduced. As with other similar pilot implementations at the outset SER worked very closely with the client team in taking it through the process. As the programme developed however it stepped back into a more consultative 'implementation partner' role as Steinbeis Temming staff took on more responsibility for the process.

A conventional system will typically require a long implementation phase as the various business rules are defined analysed and adapted in the context of what the rules-based software solution can understand. By contrast the time required for the neural network-based SER system is extremely short. "For us confirms Vorrink, the SERbrainware system could be transferred from pilot phase to real everyday business in just 14 working days improving productivity almost immediately."

In common with other organisations implementing similar SER solutions the primary drivers for change have been the desire to cut costs and improve efficiencies within the relevant department at the same time enhancing customer service through increased speed and flexibility. Yet as Vorrink emphasises the ability to have an effective automated system 'up and running' in such short time frames has proved an important additional benefit.

Similarly many companies have accepted that in areas of their operation that combine high volume repetitive tasks with low grade skills there will inevitably be a significant number of mistakes - it has been shown for example that on average an error rate of up to 20 per cent can be anticipated during invoice inputting. The SER solution which learns by example in the same way that the human brain assimilates information is capable of analysing document content and adapts itself to changes by continued learning without further programming costs. During the short pilot phase the system had already correctly and completely classified more than 85 per cent of all supplier invoices - even in cases where the relevant booking information was spread across several pages.

As the number and range of examples increases so SERbrainware's classification capability continues to improve. "The result is that the 300 invoices we receive on a daily basis can be classified in less than three minutes with a correct classification rate currently running at 92% says Vorrink. The number of documents that are not recognised and which therefore require subsequent manual classification will continue to decrease as the number and variety of examples 'learned' by the software grows.

As part of the new workflow process, incoming invoices are given to an employee for checking and then scanned in at an IT workstation. The SER software then classifies the invoice or other document, extracts the relevant information and forwards this directly to SAP/R3. The incoming invoices now appear in the SAP inbox and are available for booking, with three major advantages over the previous workflow process:

The time required for a single booking has been reduced by more than 70 per cent.

The SAP/R3 booking screen is automatically completed, as the system relays data such as invoice number, date, net value and sales tax directly to the booking system, achieving a process oriented workflow.

The list in the SAP/R3 in-box is now transparent, making it possible to establish the creditor without the need to open the document.

Cultural challenge

Training forms a pivotal role throughout implementation. Help and education regarding product features is obviously important, though the intuitive nature of the software demands less training than comparable rules-based systems. Far more important, however, is the need to address the cultural impact of what is undoubtedly a revolutionary system. In order therefore to achieve the level of company-wide 'buy-in' essential for successful adoption of such a mission-critical system, at SER we have developed a 'train the trainer' approach that has proved particularly successful where a major implementation or roll out is involved.

Targeting key staff within end-user organisations such as Steinbeis Temming, SER provides a series of training classes and workshops to provide both in-depth product support and to train them on how to achieve the right mindset in employees to get the best value from the new system. This is often achieved by focusing on the benefit to be derived by each individual, for unless employees recognise what it will do for them, they may not respond favourably to what is likely to be perceived as radical change. In the case of post room or accounts receivable staff, for example, both the company and employees may see how automating the process of dealing with incoming documentation will remove the drudgery of highly repetitive tasks, providing the opportunity for more value-added and stimulating activities. This is especially relevant where high volume, repetitive tasks require high level skills and intelligent decision-making, such as patent administration or medical insurance claims.

In an IT sector that has frequently over-promised and under-delivered, SER's emphasis on the importance of tackling cultural hurdles head-on has been endorsed by recent KPMG research that examined the primary reasons why the benefits of knowledge management products and projects can fail to materialise. The most often cited reasons, including failure to integrate within normal working practices and lack of time to learn often-complex new systems, are cultural rather than technical. This clearly places a significant onus on the client organisation to play a full part throughout in communicating to the whole workforce what it is seeking to achieve, rather than working in isolation within the IT department and presenting the total solution internally only on completion.

Knowledge logistics

With SER's document management and business process software in place, Steinbeis Temming have already taken a significant step along the road to a complete four-part 'knowledge logistics' solution, combining knowledge management and business automation to capture and exploit knowledge and intellectual capital. The initial stage is that of knowledge-enabled content capture - using corporate know-how to drive business processes from the front-end capture of information. This is then distributed via knowledge-enabled e-process support and stored using knowledge-based repository management. Finally, personalised knowledge delivery services enable individual knowledge workers to subscribe or register an interest in particular types of content on a personal enterprise or global level.

Like Steinbeis Temming, many companies already have in place some form of knowledge-based repository. Looking ahead, however, they will involve all areas of their operation in using knowledge-enabled business process support to streamline systems and processes. At SER we believe that, in the future, improving the quality and speed of decision-making through the capture and use of an organisation's intellectual capital and corporate know-how in this way will increasingly form part of the drive for competitive advantage.

Peter Turnbull is head of marketing at SER. He can be contacted at: peter.turnbull@SERuk.com