USAMA FAYYAD is President and Chief Executive Officer of digiMine, Inc., a Seattle-based company that provides hosted data warehousing and data mining solutions, which he co-founded in March 2000. Prior to digiMine, Dr. Fayyad founded and led Microsoft Research’s Data Mining & Exploration Group from 1995 to 2000. His work there included developing data mining prediction components for Microsoft Site Server (Commerce Server 3.0 and 4.0). He also developed scalable algorithms for mining large databases and led their development into server products such as Microsoft SQL Server 2000 and OLAP Services. He helped establish a new industry standard in data mining based on Microsoft’s OLE DB API. From 1989 to 1995, Dr. Fayyad founded the Machine Learning Systems Group and developed data mining systems for the analysis of large scientific databases at the Jet Propulsion Laboratory (JPL), California Institute of Technology. During that time he received the most distinguished excellence award from Caltech/JPL and a US government medal from NASA. He remained affiliated with JPL as Distinguished Visiting Scientist after joining Microsoft. Dr. Fayyad has a PhD in Engineering from the University of Michigan, Ann Arbor (1991). He also holds two BSEs in Electrical and Computer Engineering (1984), an MSE in Computer Science and Engineering (1986), and an MS in Mathematics (1989). He is Editor-in-Chief of several renowned technical journals and newsletters, and served as program co-Chair of KDD-94 and KDD-95 (the international conference on Knowledge Discovery and Data Mining) and as General Chair of KDD-96 and KDD-99. He is a Director of the Association of Computing Machinery (ACM) SIGKDD and serves as Editor-in-Chief of its newsletter SIGKDD Explorations.
TWST: Would you give us a brief summary of digiMine?
Dr. Fayyad: digiMine is a technology service provider that delivers solutions to help our clients understand the behavior of their customers through the various customer interactions they have. A typical client of ours might be interacting with their customers online, via the call center, via catalogs or via the direct mail channel. What digiMine does is help clients capture every one of those customer interactions, build data warehouses to model that information and store it, and then deliver applications to help our clients to figure out what to do with that data, both from our perspective of understanding what that data means as well as acting on it to impact the business, revenues, and optimize the interaction with customers. So it’s focusing on understanding customer behavior, understanding who is about to churn on you, and understanding who is a good customer and who is a bad customer. It goes all the way to applications that deliver real-time capabilities to do things like cross sell, up sell, increase revenue from the customers or predict what the customers might do next. We are a technology service provider in a sense that we have built our own software and our own intellectual properties; in fact, we have about 30 patents pending on a lot of our technology. But we deliver it in a service model, so we build and maintain the data warehousing infrastructure, as well as deliver the applications so that our customers get to just use them and benefit from them rather than deal with all the risks associated with building complex software systems.
TWST: What does the typical client look like today? How is that client universe changing or evolving?
Dr. Fayyad: Our clients come typically from one of four verticals ‘ financial services companies; big retailers such as Nordstrom, Barnes and Noble, J.Crew, etc.; large manufacturers that do not deal with consumers directly but are trying to understand them because they may touch the consumer via their Website or viadirect marketing communications; and large media companies and publishers that have a lot of customer interactions through subscriptions and services such as Dow Jones & Co., The Wall Street Journal, CBS MarketWatch and Microsoft. The basic idea is to help companies understand the complex data that they collect from every interaction they have with a customer, whether it is a customer who is subscribing to a service like telecommunication service, such as AT&T Wireless, or a customer who is reading news online or shopping on the Web.
TWST: Do they tend to face the same problems and questions in dealing with those customers?
Dr. Fayyad: Yes. We are really in one technology vertical, which is understanding customer interaction data and delivering the results of collecting and managing all that data in a format that makes sense to a business user such as a marketing manager, an executive, a merchandiser, etc. So the key challenge that digiMine solves that nobody else can solve today is that we can overcome the problem of data warehousing and deliver results that make business sense. Data warehouses have been filled with failure stories of companies trying to build warehouses and never being able to maintain them. But more importantly, it is not just about maintaining the data storage, but about adding value through our applications that help the business user understand customer behavior via interactive analytics and taking action through predictive applications such as cross selling, up selling and reducing churn rates.
TWST: Do you actually have products and services? Does this encompass simply the online contact or does it extend beyond that?
Dr. Fayyad: Yes, we offer our products and applications as a service. The scope extends beyond just Web data obviously. We usually start in the online world with the Web presence because those are the systems that are newest and generate the most interaction data. But we typically augment that with data about customers, with data about products and then we pull in data sources from other customer interaction points such as call centers and direct mail. In the case of automobile manufacturers, it’s data from the dealers on what cars were sold, data from point of sale for retailers and so forth. So the idea is to essentially complete the picture that the our client has of their customers by bringing in these multiple data sources and making the experience, whether it is online or offline that our client has with their customers, much more effective from the point of view of increasing revenues or understanding behaviors or what’s working and what’s not. The value added here is that the analytic reports that they get back from digiMine on their interaction data is always packaged in a way to make sense to the business user ‘ to a retail merchandising manager interested in cross selling and up selling, to an auto manufacturer concerned with manufacturing priorities, to an online publisher trying to predict subscriber behavior, or to a wireless telco provider who needs to deliver better services and reduce churn or decide who the customers with the highest lifetime value are.
TWST: IT spending in general has experienced a downfall together with the economic downturn. As companies look at what digiMine provides, what are the areas you can reach that would offer them something compelling?
Dr. Fayyad: The phenomenon of IT spending being reduced has worked phenomenally well for digiMine because we offer our technology as a hosted service ‘ there is no software to deploy. There is no long implementation cycle. What we have found is that a lot of our clients who used to spend millions of dollars on software implementations, bringing up hardware systems and so forth in-house, have essentially lost their appetite for buying big software products and have been extremely interested in our services so that they don’t have to deal with the risks and the spiraling costs that are typically associated with the large software deployments, especially in the area of data warehousing and business intelligence where digiMine delivers its products. In that sense the downturn in the economy has been a big help for us. digiMine has more than tripled its revenue in the past year, in 2002 over 2001, and we continue to see similar growth as we go forward. The other side of it is what’s attractive to the customers, which is the fact that we took the whole area of data mining technology and business intelligence and delivered it in a package that is oriented toward the business user, not the techie. Basically, as a business user you don’t even know that very complex data mining technology is running in the background to enable you to perform an application such as intelligent cross selling, up selling or predicting churn. It just happens in the background and the magic is in the box ‘ it’s not in your face. You don’t have to hire a whole bunch of statisticians and analysts dedicated to the software to help run it in-house. We get all that done for you in our data centers and facilities.
TWST: Who is the typical customer that you deal with and how is that contact point changing or evolving?
Dr. Fayyad: From our customer side, our typical point of entry is typically with marketing, the VP of marketing or a marketing manager with very big responsibilities. Anybody on the business side who is faced with the challenge of saying, ‘I have a large user base,’ or ‘I have a user base and I need to grow it,’ or ‘I have a large user base that I need to retain and I have to figure out my customers and understand them better,’ is typically our entry point. Although, in some cases, in some of the large organizations that have been through a lot of the failures of building data warehouses, business intelligence and data mining systems, we find that the IT side of the house, the CIO or the CTO, is contacting us and saying, ‘We have tried it one way and it didn’t work ‘ it’s too expensive, there is no ROI in it. I heard about your model and I would like to give it a shot.’ I would say the mix there is about 80% from marketing and 20% from IT itself.
TWST: In the competitive landscape, who are you competing against? Is it in-house attempts to perform these functions?
Dr. Fayyad: Yes, our largest competitor is trying to build it in- house by buying all sorts of software packages and trying to build internally grown data warehouses and data marts and having IT deliver the solutions to business user. On the other front, we come up against another type of competitor, somebody like IBM or NCR, a very large data warehousing or data mining vendor that is trying to sell software and/or hardware systems within that company to get them to do it. So it’s typically either totally home grown or home grown with the help of some of the larger solution providers.
TWST: How are you approaching the marketing and sales requirements for digiMine? What are your strategies there?
Dr. Fayyad: To date we have been pushing very hard going after Fortune 500 companies through our own direct sales force. We are beginning now to form alliances with companies that are service providers, selling to the retail vertical or the manufacturing vertical. Think of solution providers that build and maintain data systems for larger manufacturers, they typically have the knowledge and expertise of understanding the problems that that manufacturer is going through, but they don’t have much depth in the areas of data warehousing and data mining. By plugging in with digiMine, these solution providers will instantly have a means of delivering a sophisticated data-driven applications to their clients because we help them deliver a very sophisticated data warehousing and business intelligence and data mining solution. But at the same time they help us by plugging in the outputs of our systems into the business systems of our clients. The other partners include advertising agencies, who basically are trying to help their customers capture data from things like advertising campaigns and understanding what’s effective, what’s working and what’s not working. So these partners may typically bring us leads or bring us into accounts and we work closely together with them.
TWST: There must be a lot of new entries into this area these days. Does that provide you with consolidation opportunities? Is this a market where you can easily differentiate yourself?
Dr. Fayyad: The differentiation is fairly easy along the fronts of delivering solutions that make sense to business users because a lot of this technology today is esoteric; it’s really designed for PhDs in statistics and data mining. I can’t think of another company that is delivering hosted business intelligence and data warehousing systems at the depth of digiMine. There are companies that deliver pieces that we may be interested in. We did an acquisition of a Menlo Park company back in mid-2001. We continuously are looking at other companies that might fit in; companies that add maybe more depth on capturing of online data or some of the offline applications that we don’t do today. In this market, we have been through a cycle of evaluating a few companies and we found the match wasn’t there. So even though the market is right for the opportunities, it is tough to find the right opportunity for consolidation. There are not that many interesting companies out there.
TWST: What has been the funding history and the financial history with digiMine? Are there still needs there to address?
Dr. Fayyad: We have been extremely successful in raising funding. We have raised $45 million to date, the most recent two rounds with $20 million with Mayfield in August 2000 and $20 million in May 2002 led by Mohr, Davidow Ventures with participation from American Express and a few other companies. Right now the company is well capitalized so we see no immediate need for capital in the near future. We believe we can turn profitable and go well beyond that on the cash we have today, but we can be open to it. The company has been generating revenue for over two and half years now. Our next milestone right now is an IPO hopefully, assuming we continue to execute on our current plan.
TWST: What is the agenda when you look out over the next six to 12 months? What would make that time frame a success?
Dr. Fayyad: Within a year digiMine would have at least crossed the milestone of having been successfully delivering its products and services for a while and I think assuming that the markets are turning around and that our traction with customers both in increasing our revenues from our existing customers (not just through renewals, but through up selling), as well as acquiring new customers would have more than proven the model. I believe that digiMine would be by then positioned to claim that we are indeed revolutionizing how people are thinking about data warehouses, business intelligence and data mining technology in general and we believe that that’s going to generate a lot of interest in the market. We also have two major product introductions into new verticals that we are extremely excited about.
TWST: Introduce us to your management team and the senior managers that make this team work. Are there any areas there that you are looking to augment or change?
Dr. Fayyad: We are looking for a strong player to lead our direct and indirect sales, and develop our channel and partner strategy. So there is a position for a strong VP of Channels and Sales. We do have a very strong CFO onboard who is qualified to take us public. Our Chief Strategy Officer is Nick Besbeas, who comes from the background of dealing with consumer data and direct marketing and another co-Founder, Bassel Y. Ojjeh, our COO, whose background is in data warehouse operations and data base technology development. So we think the pieces are in place. We will definitely grow the team by adding a strong VP of Channels and Sales. We are looking for somebody who is going to have the operations management experience to round out the team and allow me to focus more externally versus both externally and internally. So those are the pieces we think are missing today.
TWST: What do you see as innovations or surprises or changes in technology that might dictate what clients are requesting and what you can provide?
Dr. Fayyad: As a technology service provider, we are in a unique position where we build our own software and IT, yet we have to ship it everyday to our customers because we run it on their behalf. Now that gives us a unique opportunity to essentially be plugged in almost all the time with our customers. Knowing their needs, we know where the solution is coming short and where the solution is satisfactory and doesn’t need more attention. So our big emphasis today is on extending the current offering today by adding in capabilities that raise the level of interaction between the end user and the product. For instance, digiMine’s customer segment manager defines arbitrary segments of consumers or customers based on their behavior and then tracks them at the segment level, elevating the level at which you think about data. You don’t need to know about the details, you just say ‘Anybody who is at risk of leaving my business in the next four months should belong to segment A, anybody who spends at least $1,000 every six months and interacts with me at least twice a year should be segment B,’ and so forth. And then the business user only thinks about segments. They don’t need to think about the low-level definitions and how they are derived and applied day in, day out in the background. The other area is rolling out a new vertical. The philosophy behind digiMine is to be rolling out in new verticals every six to 12 months. What this involves is building the business applications that make sense for the vertical, hiring the key people who understand it and rolling out the solution once it’s ready for sales and marketing. So those are the big milestones for 2003 ‘ enhancing that existing product platform and rolling out at least one more vertical. But we are not ready to announce that vertical yet.
TWST: Is there anything you’d like to add?
Dr. Fayyad: The comment I would leave you with is around why we started digiMine and why we think it isa huge opportunity. My background comes from data mining, both on the scientific side and on the technology platform side through my experience both with NASA/JPL and at Microsoft. And one of the observations that drove me to start digiMine has been the fact that for most data mining applications and data mining algorithms there is no data to mine. People collect a lot of data, but they don’t know how to collect it and maintain it in the right form for it to be useful to the business and ‘mineable’ by these data mining algorithms. That is why digiMine has combined the data warehousing, the business intelligence and the data mining components in one complete solution ‘ if we leave the pieces to the customers, it will not work. In fact most data warehouses in the industry are what I typically call data tombs because you write data into them and you never read it again. By solving that problem, we stand to change how the world thinks about data mining, data warehousing and business intelligence ‘ through lower costs that leads to a dramatic ROI, less risk and a strong focus on the business user.
TWST: Thank you. (DWA)