Mastering PIM

Barbara Mowry President and CEO, Silver Creek Systems, Inc.

Product information management is all about contect, says Silver Creek President and CEO Barbara Mowry.

By nature, many CEOs don’t speak for the products their companies produce and prefer to steer a purely financial course. Barbara Mowry, President and CEO of Silver Creek Systems, outwardly fits this role, with entrepreneurial and previous officer credentials at two Fortune 500 companies and current board seats at a NASDAQ corporation and the Universities of Minnesota and Colorado. Lately, she’s joined the Board of Governors at the 10th District Federal Reserve Board in Denver. An unlikely interview, thought DM Review Editorial Director Jim Ericson, until he stumbled across Mowry’s earlier Chairman/CEO role at Requisite Technologies, one of a few companies that wrestled with elements of product information management (PIM) long before most of us heard of it.

DMR: I covered the supply chain, e-catalog and standards movements back in the dot-com boom. Did PIM and Silver Creek evolve out of that?

Barbara Mowry: Not exactly, but indirectly. In the bubble period you’d see a system in Japanese but with the data in English. One of our founders asked what it would take to get the data into Japanese. You know, just a silly disconnect. But you can’t do that unless you understand context, which computers are awful at. Once you understand the context of information, holy smokes, you can translate it into a different language and translate all kinds of consumable information. The founders of this company saw the meaning of context, and now the approach is different and built on very deep intellectual property and patents, but it’s part of the problem we never solved cost-effectively or scaled at Requisite.

DMR: We talk about holistic master data management [MDM], but my background tells me that product master data presents some unique challenges.

BM: We completely agree with you. When I was an officer in Fortune 500 companies, we spent a lot of resources on customer data, making sure we didn’t duplicate mailings by confusing B. Mowry with Barbara Mowry and so on. We’d merge data and purge out duplicates and all that stuff. Then it evolved into uses for online systems, call centers and other things that led us to this notion of CDI [customer data integration]. I’m not saying CDI is easy, but it is pretty mature. The good news about the customer side is you have well-defined standards. The post office tells you exactly what you need to have in an address. Pattern-matching technologies work very well. When it comes to product data integration, there are almost no standards or governance process. Product data is really difficult because the word order is different, grammar is different, it’s abbreviated and so on. It requires a different technology approach because you have to understand product data in context. When people try to solve this, you see huge teams of people, lots of business rules, scripts, cross-reference tables, all kinds of kludged-together things. We know you can’t scale that, the quality is not there; and then the CEO goes and buys another company, and now you have a real mess on your hands.

DMR: Are industry standards for product information getting any better?

BM: When I look back on progress since the 1990s, I can’t see any difference. Standards by definition are the lowest common denominator, and I have yet to meet a company that wants to see its data commoditized.

DMR: Beyond standards, you are talking about a lot of requirements.

BM: We’re not here to say we solve every single part of the information problem, and different companies will contribute to a solution. We think of the world in terms of information supply chains. Somebody in the company has to sit down and ask what key components of data are needed to accomplish a strategy. Then you put together a strategy and ask what technology components are needed. Then you look for the holes. I believe solutions require data quality, data governance, data integration, data enrichment and data publishing. You need quality data, but there is no single form you can use for everything. The data I need in my e-commerce system is not the same data I need in inventory or item masters. Integration is a component because if you have quality information but can’t move it to where it’s needed, it won’t solve the problem. You need data enrichment because sometimes you don’t have the data [you need], you have a manufacturer name and part number and no clue to what it’s describing. You need data publishing companies. We have a customer, a distributor that needs to publish their information into 70 different catalogs with 70 different versions of the data - not all the data in all the catalogs, and all the data expressed differently in each catalog. You have a single source of the truth, but now you need 70 versions of it, and you certainly don’t want to store all 70 versions.

DMR: Distributors have traditionally done a lot of heavy manual lifting to assemble usable and comparable product information for customers.

BM: To use your example, a distributor’s strategy is to be the single place companies come for the best information about what they’re trying to buy. How do you create that? Distributors have the mother of all matching problems, because it’s a many-to-many problem. They get all this stuff coming in from manufacturers in all different forms, spreadsheets, portals, all different ways. Abbreviations are all different, and then they have customers trying to buy things that need information in their own different ways.

DMR: How do you address that?

BM: In the case of a distributor, we might help with their quoting process. When a bill of materials comes in, we standardize it to a common form and match it up against their giant part master file with semantic technology we call the DataLens System. Now you have both ends of the equation standardized to the same form; we can match and dramatically increase the number of quotes they can respond to. In one example, we increase their quote/rate matching by over 20 points, which is worth tens of millions of dollars. We do this on the fly without storing any information. There are many use cases. In e-commerce and retail, you have a high velocity of information change and want your catalog information on a Web site with [efficient] search. In health care, organizations group purchase to negotiate better prices across higher volumes. That’s about standardizing and matching disparate sources of data under a single contract. ROI is all over the place if you get back to thinking about managing a company’s information supply chain across key drivers of a particular business.

DMR: How do you support MDM and PIM platforms?

BM: A PIM system allows you to gather data in one place with workflow around it, and a large cost of a PIM system is in services. There are two main problems. One is how to get information into the system to create this single source of truth in a cost-effective manner. The other is how to keep the information accurate by maintaining it. We complement platforms and partner with every PIM vendor. We help get the data in there cost-effectively in a way that allows you to maintain it forever and provide governance so you don’t get bad data to begin with. The PIM system takes care of storage, approvals, workflows and all the stuff that goes into that. We also help on the outbound publishing side where you may need publishing in many different forms. We are complementary to platform systems and provide a kind of wrapper to allow for the data governance process.