Are You Ready to Reengineer Your Decision Making?

When Thomas H. Davenport was writing about business analytics five years ago, the term du jour was “business intelligence.” Before that, it was “online analytical processing,” and before that “executive information systems,” and before that simply “decision support.”

Call it what you will, this thing he was looking at was essentially two activities: “There’s reporting, which is getting a handle on what has happened in your business,” Davenport says. “And there’s analytics, which, to me, is explanatory and predictive. It’s why this happened and what might happen going forward.”

Five years ago, Davenport says, analytics practitioners spent about 95% of their time reporting on the past and only about 5% on analysis. “Most companies were not really focused on the issue and, if they had analytics, it was a very siloed thing: a little bit in market research, a little bit in quality, maybe a little bit in actuarial for insurance companies.”

That’s changing. Some companies have gotten aggressive about analytics and become case studies for the smart use of analytics in their strategy: Progressive, Capital One, Harrah’s Entertainment, Google, UPS. But for most companies, the potential for analytics to become a critical tie to decision making remains an untapped opportunity.

Davenport, who holds the President’s Chair in Information Technology and Management at Babson College, says the link between analytics and decision making needs to be relearned: “What I’ve seen a lot with the proliferation of data and data warehouses and business intelligence systems is that the tie to actual decision making has been lost. We generate a lot of data, we supply a lot of tools and we say to people, ‘Okay, go at it, have fun, play, make better decisions.’ But we never actually ensure that they do.” In a conversation with MIT Sloan Management Review’s editor-in-chief, Michael S. Hopkins, Davenport explains what it would take to change that.

What’s changing in the world of analytics that CEOs ought to know?

There are a couple of things. One is that we used to have this distinction in organizations between transactional information systems and decision-oriented systems, and I’m starting to see that distinction break down. You have big enterprise systems vendors like SAP [AG] and Oracle [Corp.] saying, “We’re going to put all of our technology into memory so you can get an answer immediately out of a transaction system.” The integration of basic transaction systems, ERP [enterprise resource planning] systems, CRM [customer relationship management] systems, point of sale systems, web ecommerce environments – all of which generate vast amounts of data – means being able to get good answers about what’s going on in your business without having to go to nearly as much trouble as you did in the past.

This is less a technology issue, I think, than a management issue. We’ll see an increased focus on decisions and how they’re made. We’ve reengineered a lot of things – our processes, our organizational structures – and I’ve been trying to persuade people that this is the time when we could really start to reengineer our decision making.

Describe what you mean by reengineering decision making.

I did a study of 57 companies that had improved their decision making in one way or another, and I asked them what they used to make those decisions better. The number one intervention tool they cited was analytics – about 85% said analytics. But right after that was change in culture or leadership, followed by better data, followed by change in business processes, then the education levels of the people doing the decision making.

The average number of intervention tools was 5.3 per decision. That suggests that, if you want to make decisions better, you have to use a multiplicity of tools.

Now, decision making has been one of those areas that is largely the province of executives. It’s invisible; nobody knows when it starts and when it ends. Sometimes it’s not even clear what the outcome is. And the process is not anything you’d be proud of in most cases.

Why not, exactly?

Let me give you an example. I advocate that a good first step for executives is to identify what their company’s top five strategic decisions and top five operational decisions are. I usually do informal surveys when I’m speaking about this topic and I’ll say, “Raise your hand if you have a sense of that.” And maybe one in every 500 executives will say, “Yeah, we know what our key decisions are.” It’s quite astounding.

I argue that they have to do some sort of inventory of the decisions that they care about and what kind of information is used to make them and how often do they take place and who’s supposed to make them – the whole decision roles issue.

In our study I found a whole new set of decision analysts who are not just doing the back office analytical work for decisions, but who are helping frame the decision, helping communicate with stakeholders, and so on.

Do you think this use of “decision analysts” is more the exception than the rule? The province of executives and managers is to hold those decisions close.

It is the exception. I was talking about this recently with a very smart fellow who heads a decision engineering group at a big computer company, and he said, “In all my years here, no executive has ever come to me or my group and said, ‘Help us make better decisions.’” He said occasionally they will come and say we need to make faster decisions. And while he does that, he can often help make them better decisions. But, he said, more often people will say, “Help us justify this decision that we want to make.”

It’s very politically difficult for people to say, “I need help making decisions.” They get paid the big bucks, and they think they should be good at decision making.

But so many great tools are relatively under-explored, under-exploited. Not just analytics and use of more data, but the whole wisdom of crowds thing, which SMR has been writing about, and the behavioral economics stuff, and the new findings in neuroscience. This is a really fertile period for learning about decision making, but it’s not applied extensively in most organizations.

What are the experiences of companies that are doing it well? And where are the trouble spots?

I was talking to Chevron [Corp.] about this, and they have a quite extensive decision analysis methodology that they train virtually every executive in. In effect, you can’t become a senior executive at Chevron without being trained in their decision analysis methodology.

The big part of their methodology is that, for every decision that costs them over $100 million, they have a forced look back, a review. In a nonrecriminatory fashion, they figure out what went well and what didn’t and what they might do better next time. A culture of honest self-examination is all too rare within organizations, and I think that’s an impediment for many companies.

Doing it well across channels is difficult, particularly when a company gets good at one channel. Harrah’s [Entertainment Inc.], which I’ve done a lot of work at from an analytics standpoint, said, “We were great at direct mail and database marketing, and then the Web came along and we thought it was great, but, man, breaking out of our existing habits was really tough.” Or 1-800-Flowers.com [Inc.] – really good at online, but they told me, “You know, integrating the call center data thing has been a tough thing for us.” It’s not that any of these areas is inherently more difficult than another, it’s just how you put them all together.

Sometimes I talk to the web analytics people and they’re a totally different crowd from the direct marketing people, who are a totally different crowd from the call center analysis people. For one thing, the web analytics people tend to wear more black and have more earrings and facial hair. But the main thing is that they need to use technologies that integrate them more. Pull all that back into the broader picture of what customers are doing and saying.

The analytics of collaboration are another trouble spot. The tools for internal collaboration are getting more sophisticated all the time, whether it’s the older knowledge management stuff or the newer Enterprise 2.0 stuff. What companies don’t do much of yet is to view collaboration as a mission-critical activity that needs to be measured and optimized. In the same way that we know how long our customers spend on our website and how many unique visitors we get, we could measure who’s collaborating with whom about what. I call this the science of collaboration.

MIT Sloan’s Andrew McAfee talks about the need to put science back in management science – the huge opportunity to be more rigorous in analysis and just have a more scientific mind-set. You’d agree?

Yes. I think now there’s a cultural shift in organizations to saying, “where’s the evidence for that particular assertion? Have you done a little test?” Michael Schrage at MIT is talking a lot about that now, too, about how experimentation is much more possible now, not just online, but almost everywhere. [See the MIT SMR interview with Schrage, “Value-Creation, Experiments, And Why IT Does Matter.”

I think there is an awareness that while reporting is good, it’s not enough. We need to do more, we need to understand why those data turned out the way they did, what it might do in the future and how we might optimize a particular one of those variables. All of which is quite possible.

A couple of names come up frequently – Harrah’s, Progressive, Capital One, UPS, Google. What other companies claim they’re competing on analytics against a company that does not make that claim?

You’re starting to see it in retail, where Walmart [Wal-Mart Stores Inc.] was quite good at supply chain analytics. Sam’s Club is starting to give personalized offers. J. C. Penney [Corp. Inc.] has been pretty active. Target [Corp.] has been active. Macy’s [Inc.] was a has-been in analytics, and they’ve started to try to do a little bit more over the last couple of years, but have not done well historically. In grocery, Tesco [PLC] is quite strong. In the U.S., Kroger [The Kroger Co.] used a lot of the same approaches that Tesco did and has been successful with it. Shaw’s has been weak, Stop & Shop relatively weak, Safeway weak.

Gaming is an easy one, because there are three big competitors. There’s Harrah’s, there’s Wynn and there’s Sands, which is Sheldon Adelson. Harrah’s is the analytical competitor. Adelson competes on conventions. And Wynn [Resorts Ltd.] competes on luxury and style. Wynn and Adelson dabble in analytics – they have loyalty programs – but they don’t really compete on it at all.

Why wouldn’t competing on luxury and style be more effective if it were aided by the use of analytics. Wouldn’t analytics be complementary rather than contradictory? Wouldn’t questions about whether more profits are generated by marble floors or Oriental carpets be something you’d want data about?

I think that whole area of style and distinctive brand – luxury – is still largely a matter of human intuition. I don’t think it has to be. But hardly anybody in the fashion industry uses data. In an SMR article I co-wrote, “What People Want (and How to Predict It),” we talked about how there aren’t very many people in the movie industry who use analytics or other predictive decision-making approaches. Will Smith has done really well with it, but not too many others. I think maybe in five years that’ll be one of the changes that will pervade the more intuitive aspects of life, but it doesn’t now.

How much pressure do you think organizations are going to be under to aggressively incorporate analytics and other predictive approaches into decision-making? Do you think executives can still carve out pretty big competitive-advantage gains by diving into analytics?

I think the speed depends on the industry. In some industries where there’s a lot of data, like financial services, retail or online, it’s already a business necessity, and getting competitive advantage will require some creative new applications. In others, you can still get early competitive advantage—in any industry that’s style related, especially, anything that is artistic. Management consultants have a huge opportunity; they haven’t been very analytical at all. Manufacturing–still a lot of opportunity. In any business-to-business environment, where there’s not a whole lot of data, opportunities exist for companies to set themselves apart. But for consumer-oriented businesses, I think, the cow’s out of the barn for the early applications. They just have to work a little harder at innovating with analytics.

Do you think everybody knows what analytics is now? Or have we still not reached that point?

The problem is that the definition is changing. “Business intelligence” was the umbrella term for looking at the past and looking at explanatory and predictive models. Now that umbrella term is “business analytics.” There’s a bit of titular inflation going on and “business analytics” sounds more sophisticated, I guess.

The whole category is just using data and analysis to understand and manage your business more effectively, as opposed to simply capturing your customer’s address or keeping track of your employees’ vacation balances, that transactional kind of stuff.

Imagine the executive who says, “Jeez, we don’t use analytics at all. But I know we’ve got the necessary data.” What kinds of skills and capabilities does she or he need in order to take advantage of the opportunities?”

I actually have a handy little acronym for just that purpose. It’s the DELTA model. My new book, Analytics at Work: Smarter Decisions, Better Results, is structured around it.

D is for data. E is for enterprise orientation, which is viewing this not as a series of little silos but something we try to be good at as an enterprise, by getting people talking, sharing data, sharing solutions. L is for leadership. T is for targets, where you figure out where you want to apply it in your business. For Harrah’s, it was loyalty. For Progressive, it was pricing risk. For Google, it was page rank and then, later, advertising. What’s the target for where analytics is going to really set you apart?

And then A is for analysts. You really need a lot of smart people. All this automation and data mining, where supposedly a computer’s going to find all the trends … well, it doesn’t happen. Scratch a highly analytical company and you find highly analytical people in large numbers. That’s a big problem, actually, because there aren’t enough of those people around who can tell good stories with data and explain what’s happening in common sense terms.

I’m trying to think of whether leaders typically have this kind of a skill set or don’t. Are they typically analytical? Do they typically set an example in terms of their own decision making?

Not exactly. There’s a big, big gap between the most analytical and the least analytical. American business has a fair number of CEOs with engineering backgrounds, and they tend to be relatively analytical. At the same time, an awful lot have sales backgrounds, and they’re not analytical at all. Clearly, you could do a lot of analytics with sales, but people don’t generally go into sales because they like numbers. Executives with legal backgrounds also don’t tend to be very quantitative in their decision approaches.

That said, an inclination toward using analytics can be born in surprising ways. I was with the superintendent of the Charlotte-Mecklenburg Schools recently, and he got into analytics by learning it from Bill James the famous baseball statistician. He said he used to read James’ The Baseball Abstract and he thought, “You know, we need that kind of data in education.”

Ah, sabermetrics – changing the culture of American business.

Exactly. The head of Humana [Inc.] told me that he got into analytics after he read Moneyball. The inspiration comes from surprising places.