GXS Insights

The ½% Solution – Part 2

by Bryan Scott Larkin, Industry Marketing Director, GXS

As originally published in the Journal of Trading Partner Practices.

One of the worst kept secrets is that, for many companies, the retail supply chain is damaged if not broken. Excess costs or lost revenue opportunities can be found in incorrect shipments, out-of-stocks, compliance penalties, unnecessary consumer returns, too many expedited shipments, and trucks turned away at the receiving dock. Both retailers and suppliers employ teams to rework these problems as if they were inevitable, but they rarely look to address the source of these to eliminate them entirely.

We discussed a few very important ways that can help suppliers attain a ½% increase in the bottom line. Some of these practices also can have a positive effect on retailers’ bottom lines either through the supplier’s efforts or by retailers engaging in the practices themselves. We started at the end of the process and began working our way forward. We covered the settlement process and aspects of managing logistics in a way that reduces stock-outs, minimizes errors, and reduces compliance penalties. In this, the second half of the article, we’ll move upstream a bit further and examine other aspects of the Retail Compliance Council’s Perfect Order Index and focus on eradicating pesky errors that are at the root of many, if not most, compliance problems.

The Purchase Order Process

We discussed the latter end of the purchase order process in the last article. Here we will review the rest of the purchase order process for areas that can cause the problems mentioned above. The purchase order process relays information regarding what a retailer is interested in purchasing. It usually includes product and price, ship-to location and other information. For many retailer/supplier relationships, the purchase order may be just the start of a sometimes problematic process.

Early results from Advance Ship Notice (ASN) research being conducted by the Vendor Compliance Federation (VCF) and my employer, GXS, show that retailers average more than three purchase order changes per purchase order. This corroborates results obtained by VCF in an earlier study. Discussions with suppliers point to the sales department’s desire to handle change orders manually to help minimize the chance that an order will be canceled or decreased in size. The hope is that personal discussions can change a buyer’s mind. Even though the purchase order change often can be generated electronically, these purchase order changes are manually shared even when the original purchase orders are passed electronically to suppliers. The manual intervention to handle purchase order changes can lead to idiosyncrasies in an otherwise automated order-to-settlement process. For instance, when retailers and suppliers use applications to align ship notices with purchase orders, the systems are often fed by EDI or other B2B and supply chain systems. If a purchase order change occurs manually, it might not show up in these systems or in reporting systems because it didn’t come via an automated and monitored process. This can lead to missed or improper shipments that appear to be correct because they align with the original automated purchase order. However, they do not align with the final purchase order requirements as arrived at because of all the subsequent manual changes.

Example Perfect Order Index

Example Perfect Order Index
On Time	x	Complete	x	Damage Free	x	Accurate Invoicing	=	POI
95%	x	95%	x	95%	x	95%	=	81.4%

  Retail Compliance Council: Benchmarking the Perfect Order

Another problem with manually handled change orders is the data must be revised by hand. This leads to the potential of key stroke entry errors which generally occur more than 1% of the time (more on this later). Even if the information eventually gets into monitoring and delivery systems on the supplier side, errors in data entry will introduce fulfillment errors on a regular basis. I’m not sure retailers and suppliers fully realize the number of purchase order changes that occur and the problems they can cause when handled manually. Retailers and suppliers should embrace a best practice of automating the purchase order change document. Without this automation, the order-to-settlement process will be guaranteed to produce errors that lead to compliance penalties. With compliance penalties amounting to between 1% and 2% of gross sales revenue, according to The Credit Research Foundation’s Customer Deductions: Impact on Receivables 2006 Edition, reducing this component of the problem may have a significant impact on a supplier’s bottom line.

Bad Data

While transaction-based keystroke errors can introduce bad data as detailed above, the real root of many retail supply chain problems is the initial product data itself. While many attempts have been made to fix this problem over the years (UCS2 and GDS are examples), the problem has persisted. There are several reasons for this, but we’ll focus on just a couple.
One reason is that people appear to be unaware of the sheer volume of errors introduced by manual data entry processes. The best outcome from manual data entry today is 99.5% keystroke accuracy. This is achievable only through double-blind data entry techniques where two individuals manually enter the same data into a system and the data is electronically compared before being committed for production use. Any discrepancies between the two are resolved by a third person. This process is used by some professional data entry firms that promise 99.5% accuracy. A colleague of mine has also spoken with a high-tech manufacturer that uses this process exclusively and has seen 99.5% accuracy results as well. Without double-blind entry, professional data entry firms offer 2%-5% error rates. Assuming some level of validation rules, perhaps a 1% keystroke error rate can be reached. However, even with a 1% rate, if you are entering data into a 10-character field, then across 10 instances of that field, 1 instance will be bad (10 instances of a 10-character field equals 100 keystrokes).That translates to a 10% error rate for your usable data.

Another reason for the persistence of bad data is that companies are often so functionally separated that the “process” of doing business is not fully understood or appreciated by employees in a way that allows for correction of cross-functional procedural problems–something noted in the last issue with regards to root cause analysis of compliance penalties. Functional, rather than enterprise-wide processes may optimize a department but not the enterprise and can result in duplicate data entry. Both retailers and suppliers enter the same product data manually into multiple back office systems. With a best hoped for 1% error rate noted above, what starts out as the same data ends up as different data in various systems and these errors make it difficult to manage day-to-day activities and can also lead to problems with internal reporting and potential Sarbanes-Oxley compliance issues.

Another reason for the persistence of bad data is that companies are often so functionally separated that the “process” of doing business is not fully understood or appreciated by employees in a way that allows for correction of cross-functional procedural problems–something noted in the last issue with regards to root cause analysis of compliance penalties. Functional, rather than enterprise-wide processes may optimize a department but not the enterprise and can result in duplicate data entry. Both retailers and suppliers enter the same product data manually into multiple back office systems. With a best hoped for 1% error rate noted above, what starts out as the same data ends up as different data in various systems and these errors make it difficult to manage day-to-day activities and can also lead to problems with internal reporting and potential Sarbanes-Oxley compliance issues.

Solving the Bad Data Problem

So, what’s the best practice for both suppliers and retailers for resolving data integrity problems?

That question has been asked frequently in the past few years. Originally, Global Data Synchronization (GDS) was seen as the solution. However, experience has now shown that a substantial amount of the data passing through the Global Data Synchronization Network (GDSN) is inaccurate. One study performed by the GS1 US Data Accuracy Task Force suggests that, of the products synchronized through GDSN to-date, 80 percent had data accuracy problems. In effect, GDS is really about making sure everyone is using the same bad data.

The data problems are so bad that most retailers are still just trying to get the data right for the products they already sell. Most have not even attempted to utilize GDS for new item introduction – an area that was touted as one of the great benefits of data synchronization.

That leaves both retailers and their suppliers trying to figure out how to resolve the issue of data integrity as a separate challenge rather than as an integral part of their data synchronization efforts. In fact, some retailers have chosen to continue to review all products–measuring and weighing them–even when the data comes through the GDSN. Some have hired outside consultants to do this and analyze the results. Others have implemented complex Product Information Management (PIM) systems, partially as a means to deploy data validation and inspection rules. Some have undertaken more than one of these steps. Usually they contact the suppliers to fix the data.

There are three key elements to solving the data integrity problem. First and foremost it takes vision and commitment from corporate leadership. Executives need to take notice and encourage steps 2 and 3 below. Second, corporate processes need to change. Data governance programs need to be put in place to minimize the chance of data errors. Such programs assign ownership of attributes to individuals and they enforce accountability for timely and accurate data administration. They also include the elimination of duplicate data entry–where the same data is entered into multiple internal systems. Enterprise application integration needs to be embraced to assure product attributes are entered only one time with the resulting data being shared electronically across all appropriate internal systems. All other systems should have manual update capabilities removed for such attributes, thus guaranteeing information can’t be changed except for by the owner. Finally, there needs to be a method to assure data accuracy. In fact, while listed last here, it is easier to start with a data validation initiative than it is to modify all the internal processes with a data governance program.

Such a program will assure product data integrity, otherwise known as product data quality (PDQ). Some of the afore-mentioned PIM programs provide this functionality today.

When retailers and suppliers develop validation rules separately, the supply chain does not benefit as much as it otherwise might. The reason is the duplication of efforts in developing and maintaining the validation rules. This is particularly onerous for suppliers as they have to maintain rules for all their retailers. Even though there is some standardization through the GDSN, optional and dependant data along with retailer idiosyncrasies can require significant levels of effort to maintain the validations. Historically, suppliers have maintained a table for product numbers that correlates their product numbers to the way each retailer references the products. These tables are used to translate incoming purchase orders so they may be fulfilled accurately. Maintaining accurate, retailer-specific product data – and validations for such – is akin to having one of these tables for every product attribute.

In effect, suppliers must maintain a matrix that correlates their attributes to the way each retailer wants to see the data.

While PIM might be a good way to store such data, having each supplier maintain the validation rules to ensure the data integrity is challenging and costly. A shared service that provides these validations for all suppliers on behalf of retailers is the least costly model for the supply chain – and for suppliers. Sharing the costs of implementing and maintaining validation rules for every retailer can significantly reduce the financial impact to any one company while also ensuring that validation rules never fall behind. In fact, a study by A.T. Kearney showed that it costs each company between $60 and $80 to fix bad data after it has already impacted orders and fulfillment. However, the cost to fix that data up front in a shared system is significantly lower. Considering the $60-$80 doesn’t include the cost of the problems caused by the bad data, these are compelling numbers.

The numbers are so convincing that one international data pool recently announced the inclusion of a shared PDQ solution as part of its basic functionality for the future. Such recognition of the value of PDQ in a shared environment should not go unnoticed. Software as a Service (SaaS) has received significant positive coverage from analysts such as McKinsey and Gartner, and PDQ is a prime example of where SaaS meets the retail supply chain.

Still, the most important thing for retailers and suppliers to realize is that they must get their data right before any of their current processes, or future projects, will bring them the results they really want.

Taking Action

The Perfect Order relies on accurate data. So does a successful GDS program as well as RFID. While compliance penalties can be significantly reduced from their 1% to 2% levels, perhaps the greater impacts will come from smoother operations in order management and fulfillment. Optimized shipments, greater sell-through and stronger partner relationships will be the results of this ½% solution. But what are the practical steps and best practices retailers and suppliers should take?

1. Automate the Purchase Order Change Transaction
   Take errors out early and the whole process stands a better chance of success.
2. Automate Logistics Transactions
   Automate to eliminate errors and keep everyone informed.
3. Implement Logistics Transaction Quality
   Don’t just automate – make sure your shipments are accurate and alerts are sent when they aren’t.
4. Address Product Data Quality
   Redeploy some post invoice compliance staff to the front end to reduce penalties rather than negotiating them.
5. Implement Scorecards
   Rate your customers, carriers and suppliers to identify those you can trust and those that need work.

If suppliers follow these guidelines, the reduction in errors and greater efficiency will be felt on the bottom line. Top line growth should occur as well through improved relationships with retail customers and consumers. Retailers will see benefits as well–from happier consumers, from reduced resources committed to resolve issues with product data, and from better planning based on visibility to incoming shipments. Will suppliers and retailers see ½% improvement to the bottom line? In many cases I believe the answer is yes – and perhaps more. The results have been proven with early adopters that have already addressed one or more of the above steps. Now it’s your turn. Don’t wait to succeed!

Bryan Scott Larkin is the director of strategy and marketing for the retail and consumer packaged goods industries at GXS. He also is a business technology advisor to the National RFID Center and a member of the Board of Governors of EIDX.