Analytics: Putting Information to WorkBy Brian P. Watson | Posted 02-10-2010
Analytics: Putting Information to Work
Tom Davenport admits that analytics is a "nerdy" topic, but it's also one of the hottest in the IT community today.
The IT and management professor at Babson College didn't expect too much of a reaction to his 2007 book, Competing on Analytics: The New Science of Winning (Harvard Business Press, 2007), but it caused quite a stir.
And now we can't hear enough about analytics and business intelligence. Two recent surveys--IBM's Global CIO Study and the Society for Information Management's annual survey of CIOs--found that BI sits at the top of the CIO's priority list this year.
So the timing couldn't be better for Davenport and his co-author, Jeanne Harris, an executive research fellow with Accenture's Institute for High Performance, to write a follow-up.
Where Competing on Analytics focused more on information as a competitive advantage, their new book, Analytics at Work: Smarter Decisions, Better Results (Harvard Business Press, Feb. 12, 2010, with Robert Morison), is more of a manual for turning the theory into practice.
Davenport and Harris spoke with CIO Insight Editor in Chief Brian Watson a week ahead of the book's release. Their message: CIOs can be heroes if they can give business leaders the information and tools they need to make better decisions. But first, they need to truly understand what information can do for their company--and then get past all the traditional hurdles to disseminating it in effective, productive ways. What follows is an edited version of the discussion.
What drove you to write a follow-up to Competing on Analytics?
Davenport: Basically, we felt like there was a much more positive reaction to Competing on Analytics than we expected. It's a fairly nerdy topic, but we struck a nerve and thought there was more to say. In particular, companies said they like the idea of analytics, but we don't necessarily build our strategies around or become analytical competitors, so we just want to make more analytical decisions. The first book had such an emphasis on competitive advantage; we thought we'd write another book that was more broadly directed and framework-oriented to provide people with some specific tools on how to make more analytical decisions.
Harris: Not only how to make more analytical decisions but how to set their own analytical capabilities, and give them thoroughly pragmatic advice about specific things that have helped companies improve those capabilities over time.
BI and analytics are a huge priority this year, based on various surveys. Why the imperative to do more now? What have CIOs and businesses been doing wrong?
Davenport: There's a big sea change going on right now. Prior to this, companies have been focused on implementing transaction systems, e-commerce systems, point-of-sales systems, ERP systems and so on. They've now arrived at a place where they've accumulated a lot of data; the next step is to figure out what to do with it.
Business intelligence and decision support are hardly new ideas, but they've really come of age. Organizations are interested not only in putting tools in place, but more around organizational capabilities--the analysis culture they need to pull that off.
Harris: Analytical decisions are such a hot topic right now, not only because we have the data and the technology processing power to use that information effectively, but because it's really an unmet need that goes back for decades.
If you look back at surveys from when the "CIO" term first came about, companies didn't just want someone who could slam in systems and build them infrastructure--they wanted someone who could help them get better information to make better managing decisions. Tom and I did a study about ERP systems going back to 2001, asking them, "What benefits did you most want from implementing ERP?" The answer was, "Better information for management decision making." Unfortunately, the benefit most slowly to be realized was better information for management decision making.
So there's been a desire for a very long time for IT to play a more proactive role not only in improving business transactions and processes more efficiently, but to help managers apply the information they need to make better decisions and put them into practice.
Davenport: IT organizations do an awful lot that's oriented to better decision making--not just business intelligence, but data warehousing, knowledge management, ERP systems. A lot of that is justified by its supposed contributions to better decision making, but the link between those activities is tenuous, to say the least.
We're starting to see a few IT organizations doing this better. A good example is Procter & Gamble, which renamed its IT organization "Information and Decision Solutions." But that opportunity for really affecting decision making has not been taken advantage of.
IT's Analytical Problems
Why is that? Have CIOs been too bogged down by the technologies themselves?
Davenport: I always say that CIOs have been pretty preoccupied with the plumbing as opposed to the water. But you have to admit it's been a few decades where the plumbing was changing at a very rapid rate--new valves and faucets coming out every year. It was a natural tendency to focus on getting the plumbing in place.
One of the reasons P&G is freer to focus on decision-related activities is because they've outsourced all their infrastructure to Hewlett-Packard. And with software as a service and utility computing, there's not much left really but thinking about how the stuff affects the business and how people make decisions.
Harris: I look at it from the business perspective. What are the benefits of outsourcing your infrastructure? It finally--at long last--frees you up to look at the important stuff, which is using the information from those systems to add value to your business.
It's easy to complain that IT has been too focused on the plumbing. It's certainly true, at times. But as Tom said, it's really hard to think about the temperature of the water people want to drink when there are holes in the plumbing spraying water all over the place and flooding the basement. It's only recently that we've reached a level of maturity in IT that [means] we can take that next step--to go back and focus on the fundamental mission of the CIO, which is to focus on how information helps the business.
Years ago, Tom and I wrote a report called "The Director's Cut." The idea was that people built systems, but it became such a huge effort to implement them. After a while, even though they knew better, they just focused on keeping the darn thing in. It was such an exhausting process. One client referred to it [as being] like a python swallowing a pig--afterwards, you just want to climb up on a rock and digest for a while before you do anything else.
I think that's true of IT as well. We have to master these basic processes and infrastructure and have a digestive period before we're ready to do anything with the data. It's like a director going back to movies they weren't completely satisfied with. It's about going back and doing it right this time.
You have a methodology, DELTA. What are the key takeaways on how CIOs and businesses can use it to succeed?
Davenport: When Jeanne and I did the first study of how companies can succeed at analytics, back in 1998 or 1999, we found that data and leadership were the two prerequisites. If you didn't have those, you couldn't get very far. So those are two of the DELTA principles.
That's Analytics 101. Analytics 202 is, we need some analysts who can really manipulate that data effectively, and we need a target to focus on, because we can't be analytical about every aspect of our business. And we need to think about an enterprise approach to unite all these little silos that have gone up.
Harris: In Competing on Analytics, we talked a lot about CEOs and how important they are to being an analytical competitor. But we recognize not everyone has a Ph.D. CEO who's going to be that passionate. Many people at many different levels in the organization can play leadership roles; it's just naturally going to affect the scope of the analytical projects you'll be able to do. In the new book we try to talk about those different kinds of leaders.
Davenport: At different levels.
But are all those different people in sync, strategically, or do they have varied priorities?
Davenport: There's still a fair amount of frustration with IT that it takes too long to make the data available. While we write about taking an enterprise approach and connecting the different silos, there's still not that much that would stop an enterprising marketing manager or supply chain manager from developing their own approach to business intelligence. So it really becomes an issue of governance, in many cases, to bring the whole organization up, and not just the particular functions.
Harris: The other frustration is IT is much more of this notion of data "push" as opposed to analytical decision-making "pull." I hear from business executives that IT wants to give them whatever data is a by-product of whatever system they've just developed, as opposed to sitting down and partnering with us to understand what information we really need and helping us go and get it. It's an age-old issue; it's been around since the dawn of computing.
How to Measure Analytical Success
What advice would you offer to CIOs for developing and executing a comprehensive information-driven strategy?
Davenport: There's a positive message and negative message. Positively, you can think about what particular decisions you want to influence. It's rare to have clarity on what the five most strategic and five most tactical decisions the organization needs to make are. You can start to apply the data and technology to where it's really valuable to the organization.
We talk a fair amount of identifying some sources of data that are a little proprietary. It's not the data hygiene-related issues, but also something that's really distinctive.
The negative message: Stop doing stuff that doesn't really affect how decisions are made. If you can't draw a solid line between what the information and technology are, and what decisions will be made from them to benefit the business, you probably shouldn't be building that system in the first place.
Harris: One of the reasons we put in the "targeting" section of the DELTA model was because we found that people were targeting the wrong business problems. They were targeting what they need to do versus what would really make a difference to the business.
And how can they measure the success of the changes they put in place?
Davenport: The good thing about analytics is that the metrics it often leads to are not IT metrics, but business metrics. Take pricing. We talk a little bit about it in the book, and then I wrote a Harvard Business Review article on some work that The Stanley Works had done on pricing. They were very much focused on how they make better pricing decisions. The way you measure their success is in six points of additional gross margin. The way you measure effectiveness of marketing analytics is the average lift you get from a promotion.
Those are very important business metrics, and they're not the usual things that IT people wrestle with, like spending as a percentage of revenue. They're real impact metrics.
Harris: We've been trying to emphasize the concept of analytics not just to make better decisions, but to improve business performance. Too often there's a disconnect between making a decision and taking action, or making a decision and getting any improvement in the business. One of the things we've found in talking to companies around the world is the importance of embedding analytics in the decision-making process, all the way to execution and outcome.
Why do CIOs still struggle to execute on their strategies?
Harris: We've talked to several companies where the analysts were very proud of the analysis they did and the conclusions that could be reached from the analysis. But they had no interest in whether anyone actually read the memo or executed on their conclusions. There are some dysfunctional cultures out there where people are insufficiently engaged in their work and aren't focused on seeing it through to the end.
In a way we see that in IT organizations. Years ago, people created self-help functions. They were more interested in throwing data over the wall and not having to worry about if it met the business needs or not. If we're ever going to move past that, IT needs to have a better understanding of what drives business performance.
If you look at a lot of the recommendations about what skills and capabilities analytical talent needs, you can argue that IT people need those same kinds of expertise. They need to be engaged and able to form a trusted relationship with the people using their data to understand the business objective they're trying to accomplish. They need to be engaged in what the real business problem is, as opposed to, "How quickly can I send this person a report and get them off my back?"
Davenport: Analysts were historically back-office people; you give them the data, and they come back in a month or two with a model. We found that people who are doing this successfully now are much more engaged in the whole process--helping to frame the decision in the first place, helping to deal with stakeholder analysis, and so on. They're very closely engaged with decision makers throughout.