The devil is in the data warehouse details

Stephen Brobst is on his own now.

The chief technology officer for Teradata was in Toronto Thursday, meeting with customers and offering them advice on how to make better use of the reams of information sitting in various systems across their enterprises. He was also providing them with an update on the company following its recent decision to spin off from parent company NCR Corp. later this year.

“The response has been really great,” he says, adding that many customers see it as having the potential to bring more focus to the business intelligence specialist. “Some of the kiosks and new stuff going on at NCR right now – it’s just a very different business.”

Brobst took some time to discuss with Computing Canada the evolution of data warehousing and the future of analytics.

CC: How do you think customers’ expectations of what they will be able to get out of Teradata’s active data warehousing is going to evolve over time?

Stephen Brobst: We’ve been working on this active data warehousing concept for some time and a lot of our focus has been, quite frankly, educating the customer. So we’re the ones raising the bar in terms of expectations, because we know that Teradata has the ability to deliver this active business intelligence capability from a single source of truth within the organization, rather than requiring multiple redundant repositories of information. That’s a signicant total cost of ownership advantage to Teradata customers.

CC: Who is owning the management of active data warehousing in the enterprise? How much is senior management involved?

SB: The reality is, there are multiple stakeholders in a true enterprise data warehouse. You’re going to cross functional boundaries – you know, marketing, finance, risk and so on. And you’re going to cross channel boundaries from the call centre to – let’s say it’s a financial services company – to bank branches and the Internet and so on. So typically what happens is you look for business improvement opportunities that focus on specific business value propositions and how you’re going to get specific information to the decision-makers. And some of those are focused on enabling a certain kind of strategic decision-making capability – how am I going to build the efficient frontier from a risk allocation perspective. That would be one on the strategic intelligence side. On the operations intelligence side I might be analyzing patterns of transactions for money-laundering, and the stakeholders are going to be quite different. In a successful environment, you have an executive steering group committee, and the person who heads that up is appointed on a rotating basis, based on who has the big project that’s undergoing deployment at this moment.

CC: What then happens to the CIO or IT manager? Do they take on a more tactical role based on the guidance of the steering committee?

SB: Normally the CIO is on the steering committee. He or she has a seat at the table, but they are far outnumbered by the business stakeholders in a successful organization. The way I think of it is the business owns the data warehouse and active data warehouse capability, and the IT folks are the custodians. So there’s ownership versus custodianship. I run it on your behalf but you, the business, are the owner. That’s going to be the most successful model rather than have IT ownership. You really don’t want IT ownership, quite frankly, because the business case has to come from the business side.

CC: A lot of those IT people are focused on creating service-oriented architectures right now. How is that going to affect their data warehouse strategy?

SB: When they talk about SOA using reusable components, that means reusing data, and that favours and enterprise data warehouse approach where your fundamental architecture for reuse at the enterprise level. SOA has largely been used for transactional services – building services to do a deposit or a withdrawal. What data warehousing allows you do in that environment is create decisioning services – analyze risk associated with a transaction, or to evaluate this deposit for the possibility of fraud. So decisioning services get executed on top of the data warehouse, and through service brokers and so on those decisioning services can get integrated with transactional services where transactional services are running on top of the OTP databases, but you’ve got interoperability through Web services and the standards around that.

CC: How would you evaluate the capability of organizations to use data warehouses not just for historical analysis but predictive business intelligence?

SB: I would say most of our customers – I can’t speak for the other players – are doing some form of predictive analytics, but not necessarily in all subject areas. So the areas that are most often advanced in terms of predictive analytics are around anything to do with customer behaviours. Any consumer-oriented company – and certainly a lot of the B2B players as well – are going to have models to understand how to predict when customers will deflect, price elasticity models, those are fairly block-and-tackle, at least in our customer base. I should also say that with some of the other players I’ve seen situations where customers have been forced to peel out separate data marts because they were not able to execute their workload directly against the data warehouse. So they’ll peel out separate data marts into priority in file systems – in my view that’s a bad strategy. We really want to get to integration of the data, including the results that you derive from predictive analytics. So if there’s a rocket scientist among the market research people that has (used business intelligence) to predict the probability of a customer defecting over the next six months – I want those results integrated with the block and tackle use of the data by the knowledge workers, who are not necessarily the rocket scientists. I think it’s important to leverage the results of the analytics across the organization as a whole.

CC: How does the successful use of data warehousing change the capabilities of the business power user?

SB: You change capabilities of the business power user by giving them detailed access to the data. The key is detail. Detailed data without constraints so that with reasonable performance I can access the detail and ask new questions. Not to ask the same old questions that have been optimized and slapped on some report but to ask brand new questions. That’s a real breakthrough in an organization when you can offer that level of analytic capability. But the other thing is that in the end, the strategy is only as good as the execution. How do I get the intelligence from the ivory tower that was used to formulate the strategy to the front lines of the organization required for execution? There are a number of different words for it – operational BI, active data warehousing – but it’s all about how do I use information to execute the business strategy? Right now if you look at the marketplace in general, about 25 per cent of data warehouses have been activated. Gartner predicts that by the end of this year, it will be more than 50 per cent. So more than 50 per cent means it’s going mainstream. Right now it’s the leading edge players who have done it – the Royal Bank of Canadas, the Wal-Marts — the 3M, those very big, successful, marquee type of companies. The trend is definitely there, so the leaders have done it and gained experience, but it will become mainstream and the leaders will have to continue to innovate.

Comment: info@itbusiness.ca

Share on LinkedIn Share with Google+