Time to rationalise your parts data?

"After seven long years trying to rationalise and classify all our parts data globally, we’re just starting to reap the rewards of our efforts,” says Craig Ashby, IBM’s world-wide PC division and eServer business process support and classification expert. Ashby, who started his IBM career at the firm’s Greenock plant in 1967, was invited in 1994 to head a project at IBM’s PC division in the US (Poughkeepsie, New York State), which wanted to rationalise part numbers on a global basis. “The PC division was really put forward as the guinea pig I suppose. We were the only IBM division using the parts database classification software for the next two years. After this time, everyone else wanted it too,” quips Ashby. The software was called ‘PIE’ and at that point had hundreds of IBM users. “Various departments were using it: procurement, production control and design engineers, to look up information on parts and sub-assemblies,” he says. But the system had its limitations. “It was a mainframe system out of the US. We needed a client-server version really. We spent two years customising the software [with the help of Aspect, now part of i2 Technologies]. After the takeover by i2 in 1998, the software was renamed to ‘Explore’. “Our main drivers at this time were really two things: we wanted world-wide forecasting of parts demand and, to achieve that, standardised descriptions. Parts were being described differently in different areas of the globe: the same bracket for example, could be used by 20 different IBM plants around the world, all described differently by the engineer or buyer.” According to Ashby, he then had to go through a painful process, which lasted for 18 months, where he cross-mapped all the different descriptions into a single parts classification system (Explore). “It was me, on my own, full time,” he says. So has it been worth the pain? Ashby: “Without a doubt. We’re now able to articulate demand at a world-wide level. And we can present this to procurement people at a commodity level using common nomenclature... It underpins our whole global e-sourcing and procurement strategy.” (see MCS July/August, 2002, page 24). So how does the classification software work exactly? Ashby explains: “We agreed on a hierarchy of part descriptions or categories. There are 700 in total and about 1.1 million part numbers in the system. It’s very detailed, with each category having 20 or 30 attributes to further describe the part. “For example, a hard drive might be one of the 700 categories. Against this, there are probably around 20 attributes such as: function; cache memory; id; head disk type; internal or external; number of pins; rpm motor speed; interface type; and so on.” He estimates there are currently thousands of Explore users within IBM, with around 100 at any one time. “The biggest problem we had was engaging the engineering community at IBM. They thought it was a wonderful idea in theory, but when you said to these guys, ‘Look, you actually need to scrub those 5,000 parts’, it became more difficult to get their buy-in.” Lone crusade In fact, after hearing too many engineers around the company say, “I’m too busy to help,” Ashby had to do most of the engineering parts rationalisation also by himself. Beyond procurement, another key benefit has come in the form of better inventory management. Ashby explains: “In the company’s PC division, the manufacturing operations are spread globally. Once a week, each manufacturing site in the division inputs its stock levels and part numbers (or revised ones) to Explore. We have a home-grown application which then aggregates all this stock information on a world-wide basis. We simply couldn’t achieve this kind of visibility before without the classification software.” And it’s not just improved stock visibility. Ashby provides another example: “Explore gives our designers the ability to use part selection. For example, if we wanted to design a new card for a [printed circuit] board, the designers could specify the types of components they require from Explore, searching on specific attributes. It avoids duplication of design effort. We don’t have to re-invent the wheel every time. It’s saving us crucial time in getting products to market.” Explore now holds every single component IBM uses in its products. What’s even more impressive though is that ‘attached’ to most (not all) categories in the classification system are approved lists of suppliers of components. “We have set up commodity ‘councils’ who are responsible for deciding which suppliers sit against each category and attribute. We’ve pretty much covered all the electronic component categories [cables, power parts, etc], and for the mechanical parts we’ve covered about a quarter. It’s an ongoing task really,” says Ashby. Cut new parts release You can really start to get a feel for how important Explore is to IBM now, when Ashby estimates that the firm releases around 50,000 to 60,000 parts every year, all of them controlled through Explore. And there’s even more. For example, because of the high volume of new product brands being designed these days (and hence a higher volume of part number releases), IBM senior executives wanted to be able to quantify globally how many part numbers were being released per brand to find out which design teams were contributing to the overall increase in part number releases. Again, Explore provided the information directly. Ashby comments: “We now distribute and allocate part numbers to development teams at IBM in a very controlled fashion. When designers are designing new parts, we have to authorise this first. We look at the percentage of new parts in a design and check whether any of these ‘new’ parts already exist in Explore.” Although Ashby does not reveal actual savings from its classification software, it’s quite clear that the company has driven cost from its bottom line in a big way by rationalising its parts information. As well as benefits to purchasing teams, designers are now ‘forced’ to consider existing designs and parts rather than designing components from scratch every time. The efficiencies are palpable. Ashby has some advice for manufacturing firms: “Classifying your parts data is beneficial for any company that deals with components across single or multiple sites, but it’s been a long road to success for us. Seven years in fact.” He continues: “There are barriers to overcome, such as the resistance of your design engineers, availability of parts data and scrubbing existing components. Firms will probably underestimate the cultural issues here. Also you have to have dedicated staff to maintain the information and control the release of part numbers.”