If you’re looking to implement master data management (MDM) around your operational ERP system, Bill Swanton has...
good news and bad news.
The good news, according to the Gartner analyst, is that most ERP systems have their own predefined data model. That means “many data model and architectural decisions have already been made for you, simplifying your MDM needs,” Swanton wrote in a recent report on the topic.
The bad news? Unlike other MDM projects, in which each data domain usually has around 20 to 30 different attributes that must be reconciled, MDM for operational ERP systems usually has to contend with 10 times that many or more per domain, according to Swanton.
That’s because product and materials data can have an almost infinite number of defining attributes. Whereas customer records include the predictable name, address, and purchase history attributes, products and materials can vary by color, design, size, shape and numerous other attributes.
“It’s very hard for one person to manage all those attributes,” Swanton said. In fact, each data domain covered in an MDM project for operational ERP systems could require a dozen people to manage the hundreds of attributes.
Master data governance, data stewardship pivotal to ERP
That’s why, perhaps even more than in traditional MDM environments, data stewardship and data governance are key to success, Swanton says. He advises companies like manufacturers and large retailers that are looking to tackle MDM for operational ERP systems to focus on two important areas before implementing any MDM technology.
First, companies should establish master data governance, Swanton wrote in the report. Master data governance is the process for making decisions about the business rules for master data and determining who will be responsible for the quality of the data.
In most large organizations, he said, department heads and executives who are most affected by bad data will need to take the lead. And unlike MDM for customer data, where often just one person can manage the handful of attributes, execs will have to bring in multiple workers to manage ERP MDM data.
Second, Swanton urges companies to devise a master data blueprint that includes technical descriptions and spells out business rules and responsibilities by organization for each attribute. “In most companies,” he wrote, “this has been a ‘hodgepodge’ of partial specifications, operating procedures and tribal knowledge.”
Only then will companies be ready to implement actual MDM technology, perhaps the easiest step of the three. According to Swanton, addressing data quality, data stewardship and data governance accounts for about 90% of the costs in most MDM projects. The technology itself represents only around 10%, he said.
ERP MDM can provide huge benefits for manufacturers, retailers
While laying the groundwork for MDM in operational ERP environments can be arduous, the benefits can be significant. Namely, manufacturers and retailers can expect to save significant sums by reducing errors that prolong the time it takes to collect payments from customers and suppliers.
For example, MDM could correct supplier data that results in invoices being sent to the wrong address, Swanton said.
MDM in operational ERP environments can also help large organizations streamline their supply chain.
A large manufacturer that has grown via multiple acquisitions, for example, could have different departments or divisions buying the same part from the same supplier but at different prices. A successful MDM project could help the manufacturer identify the situation and get a better deal on the part by buying in bulk.
“The elimination of redundant data from each legacy system [or as new items are added] is necessary to capitalize on the company's global clout,” Swanton wrote.
But he also cautions organizations not to try to do too much too fast. As with other MDM projects, he said, companies implementing MDM around ERP systems should focus first on the data domains whose poor quality and timeliness is creating the most financial pain.
For manufacturers and retailers, Swanton said, that will probably be either products or materials data.