I spend a lot of time analyzing the product data quality, missing data elements, system accessibility because the data elements are impossible to pull out of the system or not collected to support our clients’ enterprise requirements for purchasing, engineering and maintenance decisions. I have to admit, I am always astonished by what I see (or don’t see) and the time and cost to pull data from a system. The reality is the data entered in these systems and the systems themselves are considered a support function (indirect or non-product activity) and not the core revenue generating stream for the business however the data is the life support of BI, accurate and available data is critical for smart and efficient business decisions. The missing gap in most business intelligence programs is a foundational flaw, referred to as data integrity and data quality or the lack thereof.

A business has two options, augment their BI decisions with a data quality scoring model, IDHS, a simple example “I am confident that our inventory budget should be 1 million dollars this year, based on the IDHS (+/- 30%) the actual budget could range from 700,000 to 1.3 million.”  The easiest reality is to budget the 1.3 million, with the plan to come in under budget, .3 million provides a safe cushion. This also alleviates the over budget spending and the tedious tasks of re-budgeting or canceling other important initiatives mid quarter / year.

The other option is to incorporate a structured and standardized Master Data Management process with Data Governance to collect, manage, cleanse (legacy / new data), enrich and disseminate information to the various systems. The goal is to create one master record set to ensure that decisions are based on accurate and complete data sets to implement meaningful BI reporting and analytics.

The results of data quality improvements are because of the requirements and processes of MDM. My definition is “An MDM program includes the Data Governance to define data requirements (structure, format and content), and the data processes to manage data activities such as collecting (extraction of BOM data or the data request web form), evaluating, matching (auto and mismatch), structuring, verifying and enriching to minimum data requirements, tracking history of change and data use, quality-assurance, reporting and distributing data (MAXIMO, ORACLE, SAP or another client’s systems) throughout an enterprise to ensure consistency and control. The MDM program will also include an on-going data maintenance process to manage data updates for this information.”

The following elements of data quality should be part of the governance program for your master data. This is critical to support a global enterprise. The discussions and metrics should include:

Accuracy: We intellectually understand the meaning of accuracy. An email address is either right or wrong, however in the product information world it can be a little more complex, this is where data governance is instrumental. The same spare part can be purchase from the manufacturer (one part number) or maybe a supplier (another part number)? A part number can be many different versions; for instance, a master org record is setup with a part number to purchase safety gloves, except one part number can’t buy you safety gloves; you must include the size as a description element in order to purchase. The result of an inaccurate glove record is you may receive all small gloves, but you really wanted large or you may not receive any gloves. Different manufacturers and suppliers have different ordering and purchasing rules.

Standardization: Is absolutely critical to BI reporting. Standardization is the map to how data is entered, referenced and stored to support ease of data access. The data elements should include classification naming, attributes, part numbers including formats, unit of measures, manufacturer and supplier names, addresses, web urls, relationships to parent companies and so forth.

Structured to support multiple uses: If you have one master organization and are only concerned with purchasing systems then structure may not be a concern, but to a global enterprise with multi-systems, the structure of use is extremely important as the data is disseminated to maintenance or inventory systems. In a purchasing system a ’Bearing, Ball’, part number ‘12345’ should only be set up once but in an “end use” structured environment, that ’Bearing, Ball’ is referenced to many pieces of equipment located  and used on other equipment and in other plants, it is also listed in engineering drawings, etc. If the multiple use structure is set up correct you can report “where used” for inventory sharing, internal purchasing programs supporting reduction in inventory.

Completeness: Having all data elements entered into the system required for the safe and efficient use of each item. If your data set has some missing prices and a report is provided the value of the inventory, obviously the report is inaccurate. The governance requirements include minimum required data elements. In the world of product data, the process may require a special speedy set up for a critical item that is urgent, however the MDM processes includes going back to acquire the missing information.

Accessibility: The ability to pull information from a system is the foundation of reporting. This is a continual struggle when I am working with a new client. I often ask the questions, “Is the expertise available to be able to query and pull data as needed from existing systems?”, “Is the data stored parametrically or as concatenated text fields?”, “is the table structure extremely complicated?” Accessing the businesses information and providing the ability to slice / dice the information critical to BI.

In this fast moving, big data intense world of collecting and storing information for businesses, the reporting and analytics to enable meaningful decision making is critical, so I ask the question “What does data have to do with business intelligence? EVERYTHING”

 A key component for Master Data Management (MDM) is data cleansing which has multiple disciplines such as address cleansing or PIM (product information management). My expertise is in PIM, therefore my meetings have been focusing on data in the ERP and Inventory system.

My latest meeting was with an informed Material Manager, he understood the concepts of master data management, after the introduction meeting, he stated that “We had a data cleansing project and it did not work, I ended up going back and correcting the data.” Through the discussion, I came to believe that the data cleansing company, extracted the data and attempted to auto classify a half million records. As a purchaser of these types of services, I asked what was the process for mapping and quality checks?

The business issue is the buying team’s inability to utilize spend analytics and the solution is that the data needs to be referenced to the UNSPSC^® (The United Nations Standard Products and Services Code^®). The scope of the project is mapping the purchasing data to the UNSPSC^®. In my experience, I have identified four general levels of PIM data cleansing, 1) auto mapping 2) auto mapping with a manual review 3) verification and 4) enrichment. The cold hard facts are “buyer beware”.

The detail of the levels are:

1. Auto mapping: if you have a large collection of data, automation is a requirement however there are some issues. First, auto mapping incorrect, incomplete and inconsistent data will result in a system that will still have incorrect, incomplete and inconsistent data. The quality of the auto mapping is dependent on the structure of the data. If the data is structured to a noun or class, the auto mapping process will have high quality rate. If the data is set up as “free text”, the results will be dismal. This method will not address duplication or data quality in your system.
2. Auto mapping with a manual review: this process will take the results of the auto-mapping process and add a step of a manual review of the data. The question of the review, will all records be audited in the review, or is the process to review just the records that when the auto mapping just failed? How will consistency of the audit be managed? Again there are still the inherent issues as described in the auto mapping process.
3. Verification: In order to improve data quality, the data cleansing process requires verification with the manufacturer (service or product). The verification process assures that the purchasing record is set up to the correct manufacturer (referenced to the supplier via the contract), part number for restock ordering, UOM (Purchasing Unit of Measure), description with correctly classified i.e. BEARING, TAPER and the UNSPSC^®. Our process is to request the manufacturer to provide the UNSPSC^®. If the manufacturer cannot provide the UNSPSC^®, the item is correctly classified; the auto map to the UNSPSC^® will be successful. The verification process positions the data to identify duplication, manufacturer obsolescence and inaccurate data requiring additional information from the business to reconcile.
   
4. Enrichment: The fourth level of data cleansing quality, in addition to verifying, the data is enriched, this can be obtaining a price, warranty with the terms, additional description attributes, ECCN (Export Control Classification Number), recommended repair spare part information, eCl@ss, NSN (National Stock Number) or any other data element your business requires.

The conclusion is asking the right questions of how my data cleansing project will be implemented and managed are essential to making it a successful data cleansing project.

If we’re talking about a large company then inevitably with this new system comes the monolith project with whole organizations of people and processes, projects and documentation. The compulsion is to be sure that everyone, everywhere who has any relationship to it has their input and their needs accounted for. Along the way, the cost of implementation and other peripheral indirect costs have likely negated a great deal of at least any short term savings. Not to mention the potential increase in continuous maintenance costs and loss in performance. These are a few things I’ve learned from experience and I welcome yours.

Always have a specific objective when planning for development or evaluating software to purchase that overrides all others. Start with something like a mission statement, “We need this new system for….”

Determine the Real Needs. Try to separate the “must haves” from the “nice to haves”. Bells and whistles are great but there needs to be a true benefit. Seek a balance between development time, software performance, hardware performance and user experience. I always try to put special emphasis on the user group which stands to benefit the most. Having many users who can do their job faster and more efficiently can add up to real savings versus the few users who have a special need which bogs down the project and performance.

Change is inevitable. If some requests for additional features come along, evaluate them against the mission objective. There is nothing wrong with listening and investigating ideas for project add-ons as long as the benefits outweigh the costs in time and money, but there needs to be a limit or you’ll never complete the project. Good ideas can always be implemented later if it makes sense then you’ll have the benefit of the research already done, but be quick with the research. Evaluate the impact for doing it now or waiting. Here are some good questions to start with: 1) How much more money?  2) Would this be faster/cheaper for programming to do it now versus waiting and doing a more complicated enhancement?  3) Is the impact to the users great enough to warrant it?

Know the roles. Good ideas can come from anyone. Every project must have a project champion who makes the final decisions (and live with them) and also eliminate roadblocks. You need a user advocate who has done the job and knows what it takes. Have programmers who possess both talent and vision, not just code crunchers, and listen to them.

Have good documentation, and “Good” is subject to interpretation. This is another area where the KISS principle is very often not utilized. If you have to hire ten people to sit in meetings just to maintain your documentation you’ve probably overcomplicated it and certainly increased your project cost. I try to start with these principles:

1. Document the people on the project and their responsibilities. Let there be no question as to who does what.
2. Everyone who has a job to do needs to understand what they need to do and have the documentation to reference.
3. Keep the language simple. Focus on getting the point across. If it takes a rocket scientist to understand it you’ve failed.
4. Of course, document the issues, decisions made, by whom etc. but be sensible. Document enough to cover for the “he said/she said” but content is most important. No bonus points for flash.
5. Know who is supposed to have what done and when. Another obvious one here but I see too often where target dates are determined top down with little or no thought to cost or the tasks. Don’t let the tail wag the dog. Pushing hard to get the job done is fine but be realistic. Listen to the people who know before making bold predictions.

Here are a few data management tips:

1. Data needs a simple way to be imported into the system. Data comes from a number of sources so a dynamic mapping and import procedure to an internal processing area is useful for data analysis.
2. Yes, there needs to be an area to work on data before it is promoted to a Master Data Status. Software developers need to understand that data is never in a pristine state ready to be entered as a Master Data Record. Never!
3. Data processing requires a managed work flow through the system. Imagine the issue to have thousands of records for analyzing and many employees trying to manage who has what records outside the system. Just not functional work scenario.
4. Never copy data from one software module or grid to another, always reference. Cost per record to manage the data is increased every time a person needs to manually update an aspect of a record more than once.
5. Performance of the software is imperative. To really capitalize on software and technology reporting and analysis need to be done on thousands of records at a time. Time is money.
6. Provenance tracking is extremely imperative especially when “Cataloging @ Source” is the foundation to the quality of the record. Data should be identified with history: where the data originated, contact information, data and time, a revision level, file name, all associated records on the file, etc. MDM system developers, can you start to see the importance of this information?
7. Data needs to be cleansed and profiled; it is important that the software processing tools understand all aspects of the data. For instance search rules should not be so rigid that it takes an analyst manual actions to find a duplicate record because of an extra space or a slash. A worse case scenario is to take the data out of the system to work the data in excel, I am not going to even comment any more on that scenario except that it is totally unacceptable to remove data from a system to try to normalize it. Remember there is a lot of data brought into the business and the cost to manage the data is not core to the primary business, it is an indirect cost. The solution is not outsourcing to a “low cost, low skilled” worker in another country when much of the preprocessing can be done at the expense of CPU time.
8. Data changes, if you have a number of different modules in your software package what is the strategy to support aggregation of the changes to the different business units using the data? Does your software only update in one module and the other modules are in an out of sync situation? Again remember software should be designed to simplify the processes to support the business needs.
9. We live in a global economy language translation and localization of data is more important now than ever. What are the methods translate and maintain localized data?
10. Reporting and exporting of information is critical. It is a requirement to export a segment data set to send to a business customer or run a report of the activities of the work. A MDM system must be able audit data activities through the complete process of import through promotion to a master record.

I am a firm believer that software should not dictate a business process but should be designed to streamline and add efficiency to lower the cost the activity. If you are designing MDM systems, your team should include experts in data management, data quality and business process expertise with applicable experience. Businesses should not be paying for customizations to your software to be support basic 101 management of data.

What is your Life Cycle Data Management Strategy?

1) Drawing Libraries – The items in the library need to be cleansed and profiled to a classification schema. The schema requires standard naming conventions and technical descriptions. The schema can be designed within your company, priority purchased from another vendor or you can opt for using an open classification dictionary for public use such as the ECCMA eOTD.

2) Common Component Listing – provides a listing of preferred components that support the inventory management strategies for your organization. All equipment designers and builder are required to use the common components identified. Note: common components are set up in the drawing libraries.

3) Spare Part Acquisition – Place the components on purchasing contacts at the beginning of design, this will facilitate the ease of spare parts planning and purchasing. An item on contract provides purchasing the data needed to run analytical algorithms in order to better negotiate pricing organization wide. If the item is set up accurately to a standardized classification dictionary with technical descriptions only one time the whole organization can realize the benefits of the Life Cycle Data Management Strategy.

4) Inventory – supports optimal inventory management by promoting the ability to plan stocking levels and strategies with nearby facilities. Think about the implementation of spare parts sharing or an internal purchase first program. The most important requirement is the standardization or normalization of the data; the part needs to be classified only one-way and should be shown in every system the same way.

5) Maintenance –The use of standardized components coupled with a data management strategy allows the organization to streamline the number of different components used to serve the same function on different equipment. Also reducing the number of parts in inventory and maintenance management tasks.

Life Cycle Data Management Plans starts with component standardization and cleansing the data in your equipment drawing libraries and all downward systems including maintenance. This strategy avoids duplicate inventory items and at the same time promotes an internal purchase philosophy that puts a priority on inventory sharing before issuing supplier purchase orders. Standardizing inventory with information elements such as predefined stocking levels, identification of critical inventory, functionally equivalent item identification and purchasing analytics as well as enhanced vendor management are all necessary steps for a manufacturing business to remain competitive in today’s world of lean low overhead manufacturing.

The data quality issues that industries around the world are experiencing have occurred as a result of many years of manual inventory and purchasing record maintenance, through mergers and acquisitions of companies and business units as well as data migrations from various legacy systems into new fangled ERP black holes. There are a number of reasons why.

A common data trap frequently fallen into is assuming that just because you are implementing a new ERP system your organization will now have quality data. Remember the old computer motto – “Garbage In, Garbage Out”. Let me tell you based on first hand experience that there is nothing “sexy” about bad data when the production line is down or any other time.

Data Cleansing and Data Profiling is a very tedious and detailed oriented service. There are a number of key rules to follow whether the profiling and cleansing work is done internally or outsourced to someone who specializes in data cleansing. Here are some rules to consider before a project is started:

1) Conduct a detailed and comprehensive data mapping through all internal systems including engineering, purchasing, asset management, plant inventory management, etc. The goal is to standardize and document all data sources within the enterprise one time and ensure that each department is accounted for and determines what data elements are required to complete their business required tasks.

2) Build a central data cleansing database and make sure all locations using each item are referenced. This ensures that updated information will be passed back to the various legacy systems. You will need old information and updated information for this stage of the process.

3) The data cleansing database should include a balance of electronic scripting for data corrections and manual auditing. A solid process for answering questions needs to be set up. My preference is that the system should use a web utility that tracks data change history and other data related information such as contact information, issue resolution status, classification, questions and answers, etc.

4) The data needs to be referenced to a classification schema and a standard implemented for descriptions and properties. The schema can be designed within your company, priority purchased from another vendor or you can opt for using an open classification dictionary for public use such as the ECCMA eOTD.

5) Free text is not our friend in the data standardization world. If all possible use a system that has built in data rules and ensure anyone entering data into the system understands the standards and the importance of quality data in addition to the high cost to businesses using bad data.

6) Data Cleansing and Profiling the proper way is not “cheap”, but the cost of cleaning the bad data is always less than the expenditures incurred by cleansing your data multiple times or continuing to operate your organization based on erroneous information generated from one or multiple dirty databases.

Cleansed data permits the removal of duplicated inventory items, an internal purchase philosophy that puts a priority on inventory sharing before issuing supplier purchase orders, standardizing inventory with predefined stocking levels, identifying critical pieces of inventory, identifying functionally equivalent items, use of engineering component standardization libraries and facilitates purchasing analytics as well as enhanced vendor management.

To realize cost savings from new process efficiencies, these separate legal entities need to integrate the information flow and internal groups within each entity such as purchasing, manufacturing engineering, plant maintenance, facilities management, warehousing, commodity management, and asset recovery. Each area must share the mission-critical master data related to the spare parts. Truly integrating the information flow within the conceivably 50-plus business units that indirectly work together across the automotive supply chain to deliver just one item to an OEM sounds literally impossible and cost prohibitive. However, your opinion may change when you read the next couple sentences.

It is estimated that process failures and bad information cost business $1.5 trillion or more in the U.S. alone (Larry English, 2007). A study of large companies, a majority of which have revenues of more than $1 billion, found that 31 percent believe that their costs for incorrect data are $1 million or more per year (Dave Waddington, 2008). The most common element needed by (and from) all involved in the supply chain of the spare parts that keep our lines running is data quality and content as information is transmitted from one organization to another.

Figure 1. Typical Supply Chain Spare Part Data and Information Flow

There is a lot of activity and even more information available around Master Data Management (MDM). MDM and data quality initiatives have become an industry trend these days. To champion a successful MDM effort, formal strategies regarding data standardization in content and structure, as well as import, storage, display, and transmission from your enterprise resource planning (ERP) systems to industry partners are mandatory.

Every supplier, OEM, and manufacturer is using a unique set of data standards to attempt to achieve true “quality” for their data. But how powerful, efficient and beneficial to the automotive industry can the use of silo developed standards be? If all partners were using the same data standards, naming conventions, and requirements to describe spare parts, we can greatly streamline the process needed to exchange the information and at the same time reduce the number of physical and business process failures resulting from the low-quality descriptions contained in our legacy systems, and in most cases, new state of the art ERP systems.

The elements required to achieve a symbiotic information flow for the automotive industry are the same:

A common understanding of what data is needed for a particular class or type of item;
A common method to store the data;
A common method to display the data; and
A common method to transmit the data to those entities that do business together.
The answer is to simplify and standardize the methods used for the exchange of structured, accurate, and efficient data-sharing in an automated fashion, rather than manually sharing as it has traditionally been done. The Electronic Commerce Code Management Association (ECCMA) and DATAForge LLC have formed the Automotive Industry Content Standardization Council (AICSC). The purpose of the council is to facilitate the addition of automotive industry specific terminology to the electronic Open Technical Dictionary (eOTD), create identification guides for quality descriptions, or data requirement statements for individuals, organizations, locations, goods and services.

This also helps develop an automotive supply chain specific spend analysis classification. The dictionary being maintained by ECCMA and the AICSC is ISO standard and public domain; any organization can benefit from its use. ECCMA and the AICSC work with automotive-centric businesses to standardize the way data and information is stored, viewed, and exchanged.

Figure 2. Quality Description:

ECCMA has brought together thousands of experts from around the world and provides them a means of working together in the fair, open, and extremely fast environment of the Internet to build and maintain the global, open-standard dictionaries that are used to unambiguously label information. ISO 22745 spare parts data is capable of being used in any ISO 8000 computer application (neutral exchange), is easily translated, and must stand the test of time (long-term data retention) by using a public domain concept identifier.

Jacqueline Roberts is vice president of DATAForge LLC. For more information about ECCMA, visit the ECCMA Web site.

View web publication:

 http://aiag.informz.net/admin31/content/template.asp?ps=4683&sid=4683&brandid=4002&ptid=406&uid=0&mi=390242

To realize cost saving from new process efficiencies, these separate legal entities need to “integrate” the information flow to manufacturers and within each manufacturer to internal groups such as purchasing, manufacturing engineering, plant maintenance, facilities management, warehousing, commodity management, and asset sharing / recovery need to share the mission critical master data related to the spare parts. A truly integrated information flow could conceivably touch a number of business units that indirectly work together across the supply chain to deliver just one item to a manufacturer. The most common element needed by (and from) all involved in the supply chain of the spare parts that keeps the equipment running is data standardization, data quality and an electronic method of transmittal. A study of large companies, a majority of which have revenues of more than $1 billion, found that 31% believe that their costs for incorrect data are $1 million or more per year.1

Data standardization and data cleansing cost should be covered with cost saving initiatives. In addition to the initial data cleanup; strong data governance processes should be implemented for on-going data setups.

1Dave Waddington, “Growing Adoption of Master Data Management by Business?” citing an Information Difference survey of 112 companies, 65% of which had revenues of more than $1 billion, IT-Director.com, IT Analysis Communications Ltd., June 23, 2008.

I once spoke with a purchasing agent employed by a Fortune 500 company and this is how outsourcing was explained to me…”even if takes someone in a foreign low wage country 3 attempts to get the work correct, we are still are saving 25% over their competitors in the US.” Of course, I had a number of responses, including: Was the cost to manage and audit the work 3 times included in the cost saving analysis? Of course not, the cost savings estimate is only documented at the RFP phase.

Each day our company evaluates our internal and customer processes to build automation and intelligent software applications that increase throughput, improve accuracy without manual intervention and provide our customers with a continuous stream of process improvements. I believe long term our cost are competitive, the challenge is educating new customers to understand the unique and beneficial processes that allow them to capitalize long term implementing  our data quality solutions.

My hope is that I will never see another response to an RFP “Need more competitive pricing or to include “off shore” solution – This is required for more competitive proposal and for further consideration”

How long will it take US salaries to race to the bottom so work can be outsourced back to the states? I hope that this is not the answer, let’s discuss what US vendors need to do offer the long term value add processes that off shore options do not?

Buyer/supplier information synergy to me means that at any given time, if a piece of data is needed to complete a product description or a piece of data is needed to place a product or service order that is not available in your ERP applications, your suppliers are willing and able to provide the piece of data automatically without human intervention.

That probably sounds ridiculous.

This situation probably sounds more familiar:

1.) New equipment is installed
2.) The recommended spares list is loaded directly into ERP
3.) The new equipment malfunctions, goes down
4.) Production stops
5.) The maintenance department attempts to replace failed component only to find there is no inventory
6.) Maintenance frantically calling machine builder/suppliers/plant engineering to identify the failed component

This may be an over simplification or might not be something your organization can relate to, but, there are several types of technology and business process that enable prevention of this scenario and other problems that plague the large manufacturers supply chain.
—a common unified schema or dictionary used by all commercial organizations to label product information (eOTD)
—a common method of transmitting this information directly into ERP prior to the need arising to replace a component (xml)
—develop a supporting business process to ensure the needed information is requested (people)

As manufacturers/suppliers/BPO providers we all need to work together to move to a common method of requesting, transmitting and receiving the information we need to keep our operations running.

How would a buyer know that these are the same parts? Two different manufacturer names and two different part numbers; this scenario will cause duplication in a purchasing system. The result is the additional work of creating and maintaining two contracts but also cause downstream effects such as excess inventory with more than 1 stocking location, lack of a volume purchase or a global view.

 Question: Is the answer to always to confirm the actual manufacturer and set up supplier references?