In the world of data quality, the most familiar examples where a large company's data quality can visibly be dodgy is in your own (their customer) name and address. This is especially true if you have ever had the joy of moving house and getting your (or at least my) bank to figure this out; after all how could they be expected to deal with a unique, unprecedented event like a customer changing their address? However, at least customer names and addresses typically follow a certain structure and have postcodes, and this makes them susceptible to analysis by automated tools.

Imagine how much trickier product parts information might be, with products being made up of many components, each of which may contribute to multiple products and being classified in all kinds of different ways, perhaps with complex and variable depth classification hierarchies thrown in. Yet it is clearly critical to get it right. It may be frustrating to run out of light bulbs at home, but imagine if the part in question is a key component for an offshore oil rig, or a commercial airplane. If a key part fails and there is no replacement for it available, the rental cost for a deep drilling rig of USD 400k per day or more will encourage a more thorough approach to parts management.

Just knowing which spare parts are actually available, and where they are, is a major challenge to large corporations, with all the data quality issues that one might imagine. It is worth flying a part half way around the world if it avoids downtime on a drilling rig in the North Sea, but this can only happen if you actually know that the part you need is lurking in a warehouse somewhere in Malaysia. Hence maintaining high data quality can have significant rewards, yet data quality tools that are whizzy at customer name and address analysis may not be suitable for this more complex data.

This is a world inhabited by offshore (often Indian) companies that manually scour through parts databases matching part numbers up to product catalogs, attaching attributes and spotting duplicates. It is also home to a few specialist data quality software companies such as Silver Creek and Zycus. An interesting competitor to these approaches is sparesFinder, a UK company founded in 1998. They have steadily built up an impressive customer list for their "Masterpiece" data quality software, with names such as BAT, Nestle, Philip Morris, Newmont and Lafarge amongst their clients, by offering a flexible solution ranging from software only to the full outsourcing of a data cleaning project.

Masterpiece has been from the outset a "software as a service" hosted offering, and comes with considerable parts-specific functionality, including 2,500 templates for equipment such as pumps, switches, gauges and tubular pipes. This may not sound glamorous but it gives them a head start when tackling data quality problems in the engineering world in particular. There is a pre-built connector to SAP, and the product can export data in assorted languages. This proved useful to Corporate Express, who used the tool to rationalise their office products catalog data from 12 countries to create a common European catalog, in the process reducing the catalog size by 35%.

As so often with European software companies, their depth in product is not matched by marketing spend, and the company is relatively little known despite having customers in 90 countries. With a client list that many companies would envy it seems to be a company with some potential if they can better exploit their project successes and get their message to a wider audience. Spare parts for (say) mining equipment may not be glamorous but, as the saying goes, "where there's muck there's brass".