Written by Tamr
A Job Really Well Done
Take a moment, when you can, to read Rachael King’s WSJ piece on global manufacturer Flextronics International Ltd and the software it has developed to manage its “complex global network of 14,000 suppliers in real time.”
Flextronics — which builds and ships products for at least a dozen industries via facilities in thirty countries and four continents — created the “Flex Plus” software “to help identify potential problems with suppliers earlier,” King writes, “and redirect work to keep inventory moving.”
We don’t know the Flextronics folks or the specifics of the software they’ve developed. But from afar, we’re pretty amazed at what they’ve done. Building this kind of solution meant integrating “data from various systems such as inventory monitoring, manufacturing, quality, outbound transportation and delivery and figur[ing] out how to deliver large quantities of data in real-time” (to everything from the ridiculously cool touchscreens pictured above to local computers, tablets or smartphones).
This is really hard to do, even in a highly centralized, top-down environment. But Flextronics rejected the “control tower approach” in favor of a more distributed goal. “Flextronics wants the whole company to see the same information globally,” King writes, “but let local managers make decisions.” King goes on to explain that:
One of the biggest challenges in creating the software was building uniform part numbers and naming conventions across the company’s global systems. “We allow the factories a certain degree of autonomy and that creates challenges in master data management,” said [Flextronics SVP/CIO Gus] Shahin, in an interview [with the WSJ].
At Tamr, we’re seeing this same dynamic playing out across not only supply chain/procurement, but also customer data integration challenges as well: companies are looking to “democratize” data visibility across the org and provide analysts and decision makers a clean, unified view. But the sheer variety of data and the diversity of its sources within enterprises like Flextronics are too much to handle for strict top-down, rules-based Master Data Management approaches, which depend intensely on data expert knowledge for manually creating rules and for entity resolution. This dependence on top-down rules and assumptions creates a system brittle to data changes or additional sources, requiring constant manual maintenance that limits integration speed, scalability and agility.
Tamr was architected to deal with the very problems that Master Data Management products face within the modern enterprise and in the age of Big Data. Tamr serves as a complementary component to Master Data Management and empowers companies — like Flextronics did — to embrace the volume and variety of data within the enterprise, not shy away from it. Whereas traditional Master Data Management products were built top-down to deeply manage a few data sources, Tamr was built to manage hundreds or thousands of sources through a unique blend of probabilistic machine learning and expert sourcing. With Tamr, a Master Data Management customer can dramatically improve management speed, scalability and agility — while maintaining data quality throughout the entire process.