This week is the annual SAP Sapphire conference in Orlando, Florida.  I didn’t make it to the show this year, but I thought I would offer some insights on ERP and its increasingly interdependent relationship with B2B e-commerce.   The ERP vendors have spent much of the past few years focused on rewriting their applications to support a services oriented architecture approach.  SAP has its Netweaver initiative and Oracle has Fusion.  However, one area I think the ERP vendors have underestimated is the need to redesign their applications to support the extensive level of outsourcing that is becoming predominant amongst their customer base.

The Outsourced Supply Chain

In today’s manufacturing market, outsourcing is becoming more the norm than the exception.  Companies have become more and more specialized within their value chains.  OEMs that traditionally have been manufacturing-oriented are increasingly outsourcing many of their supply chain functions to third parties (contract manufacturers, freight forwarders and third party logistics providers).  Back office functions such as accounts payable, human resources and IT management are being sourced as well to specialized BPO firms.  The overall result of this outsourcing phenomenon is that manufacturers are more dependent than ever on business partners to perform daily operations.  It also means that enterprise IT systems are more dependent than ever on moving data to and from business partner IT applications.  In order to gain visibility to outsourced, external processes, manufacturers must be able to synchronize data in real time with their business partners.  For many manufacturers who outsource critical manufacturing, logistics, distribution and service functions, a growing percentage of the data housed in corporate ERP systems actually originates outside the enterprise.

50% of ERP Data Originates Outside the Enterprise 

I was talking to a customer the other day who told me that over 50% of the data in their ERP system comes from trading partners.  This comment puts in perspective the critical role that B2B integration technologies play in enabling ERP.  It also underscores the need to ensure that bad data isn’t flowing in from external interfaces and thereby corrupting information quality.  Much has been written about the challenges with maintaining what is called “Master Data” for products, customers, employees, assets and suppliers in the past few years.  But I am not referring to bad address information for consumers who move around every few years (i.e. customer master data).  I am referring to non-master, transactional data related to a specific order.  The order data originates from large customer accounts as well as contract manufacturers, third party logistics providers, banking institutions and business process outsourcers.

ERP Needs a Firewall to Protect it from Bad Data 

Who cares about bad data in an ERP?  Both line of business managers and IT personnel should be concerned about the consequences of this growing problem.  Bad data pollutes a manufacturer’s ERP system.  The result is a loss of productivity.  But to be more specific, data quality errors result in three negative impacts to a manufacturing organization:

1.       Longer time to process – Time sensitive processes may be delayed while accounting, warehouse and customer service personnel research and resolve data issues.  For example, if an invoice is posted to an accounts payable system without a general ledger number to apply the cost against, then an accounting clerk must phone the supplier or internal buyer to capture the appropriate data.  If a purchase order is pushed to an order management system with an invalid part number or SKU then the sales organization must contact the customer to discuss an appropriate substitution. 

2.       Higher cost to process – Personnel must spend time and effort manually correcting data in the ERP application or fixing problems resulting from the processing of the bad data.  Higher volumes of manual processing unnecessarily inflate costs and erode margins.  For example, in the retail industry studies have determined that over 60% of invoices have data errors.  And each error costs between $40 and $400 to correct.

3.       More mistakes during processing – The probability of an error increases exponentially as soon as manual processing begins.  By comparison, however, human intervention could be relatively inexpensive when compared with scenarios in which bad data goes undetected.  What are the costs of missing a contract commitment with one of your top 5 accounts?  Or of fulfilling an order incorrectly with the wrong parts shipped to the wrong location? Such data quality errors might seem like insignificant problems when viewed at a microscopic level for each individual order.  However, the costs of increased Days Sales Outstanding from invoice processing delays and customer penalties from failed order fulfillment commitments can quickly compound to have a macroeconomic impact on financial performance.

Why Firewall your ERP? 

Why aren’t ERP applications designed to capture these types of business process and data quality errors?   Actually, SAP and Oracle do provide extensive business logic and data integrity checks within their applications.  When an end user keys in data to a graphical interface the native ERP business logic will detect a wide variety of errors.  However, when the data flows through a B2B gateway and is subsequently uploaded into the ERP database there is very little data checking that occurs. What can manufacturers running ERP do to prevent bad data from corrupting their enterprise applications?  One option would be to hire a systems integrator to develop custom code to enforce data integrity standards for B2B imports and exports.  Additionally, a new set of user interfaces would be required to manage exceptions identified by the data checks.  Any customizations to ERP applications come with a significant overhead.   With each new release of the vendor’s software, the customer must perform extensive regression testing and often software updates.  I think a better option is to deploy an application at the edge of the enterprise that inspects incoming and outgoing documents from trading partners for data integrity issues.  Such an application would effectively be acting as an “ERP Firewall” for bad data.  The firewall would inspect the contents of EDI, XML and other files in a Demilitarized Zone for quality of content.  Bad data would be rejected to the sender or held in a queue for exception processing.  Good data would be passed straight through to the ERP for immediate processing.