Competing With Data & Analytics
When Paul Barth and Randy Bean launched NewVantage Partners in 2000, most organizations were not paying a lot of attention to data and analytics. That’s changed. As internal customers realize the value of their information, the demand for access to the flood of accumulated data has risen almost as quickly as the issues of security and governance surrounding it.
Though many organizations’ analytics projects are CIO-driven, the partners have seen more success when organizations “start in the middle” with a “real tangible business problem.” Solving something quickly, that gives quantifiable results, can help build a data management practice that ultimately influences the entire enterprise.
Barth and Bean spoke with David Kiron, executive editor at MIT Sloan Management Review, about the importance of governance policies and structures, the role of culture in the success of initiatives and the promise of value that many organizations are realizing from their analytics’ innovations.
How would you contrast the state of data analytics technologies in 2000 versus today?
Barth: When we were starting the company in 2000, people wanted insights, but not very much attention was paid to analytics or data. Even in large companies, there were a lot of fiefdoms and silos were not well shared. Except in very specific functions like direct marketing where you’re analyzing data and using direct mail and direct communications to market and sell your products, there wasn’t a broader understanding that data is both an asset and a potential liability if you don’t manage it carefully.
Over the last decade the notion that data doesn’t belong to an application, but to an enterprise, has taken hold. Companies understand that it’s a durable, proprietary and competitive asset. Protecting and securing data has become a priority.
The downside is that the systems and the data environment today are significantly more complex than they were ten years ago. All the investments in IT and applications and systems means the data is replicated more, and it’s inconsistent across systems and business processes. So the problem is bigger, at the same time that the desire to get access and make this useful is more urgent.
How does having multiple copies or redundancy lead to greater risk?
Barth: The problem with data is it’s very easy to copy and to have in many places, and our computer systems today don’t do a very good job of leaving an audit trail of where it’s gone. At one client, we identified over 500 desktop databases — Excel and Access and those types of things — that had confidential or sensitive information stored on them. When it’s copied that many times and it’s in an unstructured environment, it’s very hard to put in a business policy to control it. And if you don’t have a good knowledge of where your data is and where it’s going, it’s very expensive to try to reduce that risk.
What are some of the different categories of business value through data and analytics that you’ve identified?
Barth: One of the primary benefits you can get from harnessing your data and using analytics is process efficiency. For example, if you can authenticate your web customers, and can give them information that allows them to navigate and do transactions on their own, you eliminate your need to have your own employees doing work for your customers.
When you are developing products and services, you can be much more innovative with your data. You can see patterns; you can configure and customize what you bring to market. And you can assemble that into automated, customized routines that present something to the marketplace that your competitors can’t.
One of our financial services clients packages the data that it has around marketplace and different purchasing patterns, and sells business consulting on top of it. They can help businesses with things like optimizing store placement and location in different markets. They wouldn’t think of themselves as a data company or an insight company, but in fact they’re harvesting this information to innovate and bring new products to market.
In your experience, do companies tend to think about analytics in terms of giving them a competitive advantage, or improving their internal processes, or both?
Barth: It depends on the market. In 2008, everyone was talking about process efficiency and cutting costs. If you’re in more of a growth mode, you’re definitely looking for competitive advantage, the kind of innovation that we’re talking about.
Bean: For example, the Internet has raised the expectations of customers around the amount of customization that they should experience when they call, or when they log in. They expect that the website will recognize them, that they won’t have to waste time rekeying of information or dealing with offers they’ve declined dozens of time before. There is definitely competitive advantage in meeting or exceeding that expectation of personalized customer service.
Who is typically driving that kind of project in the organization, the CIO?
Bean: Yes, but the biggest predictor of success of these initiatives from our perspective has been when there’s a strong business sponsor involved. Purely technology-driven initiatives often fail because the business organization doesn’t see how this is going to contribute to improving the top line and bottom line of the business. Business sponsors that subscribe to the notion of information as a key driver to their business are the ones that get the broad base of organizational support.
Barth: In general, we have seen less success when you start at the very top, at the CEO level, and appoint someone a data czar and try to go down through the organization. In a large organization, this is such a new practice that it’s just too much to digest. It’s better to start in the middle.
Five years ago one of our clients in the servicing industry said, “We pride ourselves in trying to achieve one-call resolution. So a customer calls, they get everything done in a single phone call. But there’s one piece of data that we’re missing and we can’t find, and it usually results in us having to do a call-back, which is extremely expensive.” He could quantify it.
But as he started to look, he found the root cause was a data issue that crossed multiple parts of the organization. We were able to build a governance function and a data analytics working group that could deal with that issue. Once they realized they could deal with that issue, they then started to deal with more and more enterprise issues. So starting with a real tangible business problem that’s not too hard to solve is a good germ for growing a data management practice that ultimately influences the enterprise.
Have the dynamics between those who might own the data and those who want access to it changed over time?
Barth: In the past, the data steward — the one you might think of as owning or having responsibility for a data asset — might have been reluctant to share for a variety of reasons. The data consumers want the data, and the insights. The challenge has been when there is only an ad hoc or kind of a person-by-person negotiated process. Then when changes occur, the consumers might be surprised and it may throw off their reports or their analytics. On the other hand, if consumers are using something inappropriately the stewards might be concerned that they have a security issue or an issue around the appropriate use of data that they lose control of. Just establishing a governance process alleviates a lot of the tension in that relationship, because there’s a process for raising issues and resolving them.
Bean: These programs often take root by establishing data governance policies and boards that bring together the consumers of data and the stewards of data into regular weekly or monthly interactions.
One of the things that governance bodies do is to maintain information lineage. This is a type of metadata that basically maps where data was originated, where it’s stored, what the system of record is, where it flows, where there’s some added value to it, through every use of it. That way if a change occurs, you know exactly who will be impacted. You can analyze that impact to minimize it or to get the most benefit.
But that means the consumers of data have a responsibility to use the right, approved data, and not just something that’s easy or convenient or on their desktop. They have a responsibility that if they have new requirements, that they document those and share those with the people who are maintaining the sources. So there’s a mutual responsibility, and by having both sides taking an active role in the maintenance and usage of that data, a lot of that pressure and friction that used to exist is alleviated or addressed.
What is the role of culture?
Barth: Culture can be a big barrier to leveraging information and analytics. Sometimes executives or managers don’t really want a bright spotlight of data revealing exactly how their processes work, or exactly how things are functioning. There is risk associated with being too transparent about how your operations are working.
Oftentimes the benefits of improving the quality, the usability, etc., of your data do not accrue to the person who’s responsible for recordkeeping and stewarding the data.
Compensation is one of the biggest barriers to developing a healthy data strategy and data ecosystem, because if everyone is locally compensated, the value of a shared data asset has to be sold to many different players, and if those constituents just don’t agree, you’ll end up fragmenting and corrupting the data environment.
Bean: In most organizations, the data has developed in specific silos, lines of business, and those folks have been rewarded and their career success has been based on meeting the very specific line-of-business goals. But typically, the greatest value of data and information is the opportunity to utilize it and see it as a shared asset across the enterprise. Instituting programs across the enterprise requires cooperation of business owners from across the organization, and aligning them in a way that is not consistent with the directives that they’ve had historically and how they’ve been compensated, and how they’ve been successful historically.
Data management projects and data strategy is as much about change management in most companies as it is about the technology or business issues it addresses.
How do you tell an organization to begin? How would you think about developing more data-capture capabilities, for example, versus data governance structures and policies?
Barth: First identify the business drivers that are on the plate in the near future. That is, what strategies are we trying to get out the door, and what are their business impacts? And then, how does information and analytics support that, or enable it? As you drill down into more and more detail, you identify what we call critical data elements essential to implementing a business change.
For example, if you wanted to have increased self-service on the Web, you need to be able to identify your constituents both when they come into a physical branch or into a call center as well as on the Web; and you want to identify them as the same individual, but maybe in different roles. You may find two dozen elements that are critical to know to support this new personalization process. Working backward to figure out the lineage of that data, you may build a group to help you build a prototype of a process that can become part of your business process over the long term, and one that you can repeat over and over as you identify additional critical data elements.
Bean: These initiatives tend to fail when they’re perceived as “another large data project,” because so many organizations have gone down that path before. It needs to be stated in very clear business terms and started on a more incremental basis.
We usually counsel organizations to start fairly focused and identify quick wins that justify the investment, both in terms of cost as well as time. As organizations are able to identify successes along the way, that builds momentum, it builds organizational support, and it helps the organization more broadly see the benefit.
The way many organizations, for example large financial institutions, have begun to address the data issues is doing it in the context of some of these other regulatory initiatives; so it will be described as a Basel II initiative or a Sarbanes-Oxley initiative.
These are often led by a chief risk officer or a chief financial or compliance officer. But the net result is that for the first time the organization is able to begin to pull together a lot of their data across the enterprise. As a byproduct of meeting the regulatory goal they’re able to create the foundation for an enterprise data management capability.
If that’s coming out of the risk part of the organization, who recognizes that opportunity for greater enterprise benefits?
Barth: CIOs are often the first to recognize that there’s an opportunity to build this in a way that it’s useful for more than just one function. They have this central function where they see a lot of information and technology that is commonly used. They see the infrastructure, and they’re usually very oriented towards reuse, leverage, and simplifying their environment.
It needs to be supported at the top. The CIO needs access to the other C-level executives and business partners to get them on board, and there needs to be a direction that says, let’s take advantage of this data asset and the leverage we can get out of it. We know of one case where they didn’t have that, and they are spending a lot of money but not getting anything other than a Basel II reporting application and database out of it. In another case, they’re building an enterprise asset and filling out their data strategy.
What are some of the difficulties that organizations encounter?
Barth: Because these are cross-enterprise initiatives, they’re often driven by people who are brought in to be a change agent as opposed to long-time line-of-business executives. So by definition, they’re often somewhat short-timers. They’re being brought in to shake up the organization, change it from the way it’s been in the past, to where it’s perceived as needing to be in the future.
In the first few years, you absolutely need a change engine. You need someone to disrupt those status quo, to put controls and guardrails around things that are destroying your data environment and your ability to do analytics; and you need somebody with a vision who’s saying, yeah, we’re going to go in this direction
Usually a change agent has to face an initial set of resistance, where nobody believes them. People don’t believe this will be any different than the last time; they believe we’re special; this is too hard to do. Either we don’t have the skills or you don’t have the skills and we won’t get this done. And we’ll see a lot of that either behind the back or in the face kind of confrontation as these are getting fleshed out.
In the early days, oftentimes people will just say, I’m just going to take my ball and go home. I don’t need to share, or I don’t need your data; I’ll get my own. And those things need to get resolved. These organizations need to learn how to develop a matrix-oriented set of decision-making around a shared asset that everybody needs to steward and needs to respect. So, the lack of a real personal, direct incentive for participating in the early days can be another area where there’s some drama.
Bean: We see a lot of what we call passive resistance. People will say yes, but then they don’t follow through, or they don’t actively participate. The justification is often that there are other things within the lines of business that they’re in that take precedence.
So, if you’re in a line of business and there’s acute pressures to continue to sustain profits or build revenues or recognize process efficiencies and cost savings, and there’s a fixed amount of time and energy and that’s in competition with an enterprise-level initiative, people will focus on what has the most immediate near-term impact on their own goals and their own success. The enterprise shared-services type of initiatives often are tied to longer-term measures that are hard to identify on a quarter-to-quarter basis.
What other obstacles have you seen?
Bean: On multiple occasions we’ve seen organizational data initiatives sponsored from a senior level that have been dropped a year or 18 months into it. Often it’s a change of executive leadership; sometimes it’s changes in the economic and business environment which generate a kind of kneejerk reaction where executive management will say, we must pull the funding or suspend the funding for all projects that are not immediately critical to driving down costs, and stock market performance.
That’s why we’re big believers in demonstrating some very quick successes that provide a financial justification to keep these things going. Shrink the change, try to make this manageable, understandable and tangible in terms of the results that you get from it, and then iterate on that and continuously improve on your abilities.