Jeanne Harris (pictured) is an executive research fellow and a senior executive at the Accenture Institute for High Performance in Chicago. She leads the Institute's global research agenda in information, technology, analytics and talent. She is also co-author, with Tom Davenport, of Competing on analytics: The new science of winning (Harvard Business School Press, 2007) and (with Bob Morison also) Analytics at work: Smarter decisions, better results.
On a visit to London, she briefed Computer Weekly about how she, at Accenture, sees the current data management scene. What follows is an edited version of that interview. In it she speaks about why analytics has attained such prominence in recent years, how education needs to evolve for data analytics, the role of the data scientist, how to build a data science capability, how corporate IT should shape up, the "second machine age" and how data analytics needs the complement of intuition.
Q. Tell me about the Accenture Institute for High Performance: the scale and scope of that and how you see it.
A. The Institute for High Performance is Accenture's strategy business think tank. That's the best way to think about it.
It's not the only R&D group within Accenture. We have a group called Accenture Technology Labs, for example, and you may have seen the Accenture Technology vision that comes out of that group.
Think of the Accenture Tech Labs as the people whose charter it is to look at what's coming from computer science schools, software companies, hardware companies and entrepreneurs' startups. It's their job to figure out how technology can enable business to do new things and how that might be applied to business.
The charter of the Accenture Institute for High Performance is to look at the political, business, social, economic and management trends, and consider what new pressures and expectations they are going to place on technology.
I'm responsible for the IT strategy for the Institute.
Q. So, broader than just analytics?
A. Very much so. My most recent article was on the implications of venture capital industry changes and how that's affecting tech startups and what that means for job creation. I had an article in Accenture's publication, Outlook, which is about the culture of tech employees, specifically in Silicon Valley, and what that means to other businesses elsewhere.
Q. But analytics is hot right now?
A. It's an interesting subject because it is, in one sense, very timeless. If you go all the way back to Peter Drucker, he would say, go back to the 1500s and Italian and German banking -- it was all about how to get the best possible information to make the best possible decisions. That's the fundamental role of a manager -- to plan, execute and monitor.
In fact, if you go back to the ancient writings about IT when IT first came to business, the whole idea wasn't just to process payroll, it was to give managers better information for decision-making. So that notion, that cause, has been there, truly, since the 1960s, and yet it's really only recently, relatively speaking, that we've been able to fulfill that objective.
People ask me why analytics has taken off now, and I think there are a couple of reasons for that. One is that, when I was a consultant in the 1980s and 1990s, people had the same desire to use data for decision-making, and to incorporate it into their processes, but it seemed like whatever they wanted to do, the data volume was too big, or it was too poor quality, or it was not integrated.
Whatever it was, it seemed like the technology to solve that problem was always in the future, and it seemed like it was always just out of reach. Well, I think all those data and technology obstacles have largely fallen away.
We also have a generation of leaders now who have always had PCs, if not laptops, so they're used to having data and the ability to manage and manipulate it, whereas the previous generation had to make decisions based on their gut.
If you took an MBA course in the 1970s, you were taught about strategy. How you differentiate yourself and how you wreck barriers to competition.
They're irrelevant today. So, what's left is to make the best possible decisions and to optimise your processes and execute them with rigour, and be innovative and agile. All those things are powered by data, so it's forcing people to look at data and analytics in a new light.
Q. Do you see that being reflected in MBA courses now?
A. Slowly. One of the biggest dialogues going on globally among business school professors is how to incorporate some of these concepts into the curriculum.
There's a proliferation around the world of masters degrees in analytics -- predictive analytics or business analytics, they all have different names and slightly different models.
Q. In terms of your research now, what are you focusing on?
We did a survey that looked at people who self-identified themselves as data scientists or analysts, trying to understand what, if anything, was the difference. We also tried to look at whether there was a difference between people who said they were data scientists in Silicon Valley and the work they did, compared with those in London or New York.
One hypothesis was that a data scientist in Silicon Valley looked materially different than one elsewhere -- but that's just not true. We found that data scientists are, not a different animal, but a sub-species of analyst. They don't always accept that notion -- they themselves see it as quite different. Data scientists see themselves as much more entrepreneurial. They see themselves as much more likely to be working on something strategic to the business.
And I'm participating in the debate about "data scientist unicorns", which is the idea that the combination of skills and knowledge you would want in a data scientist are too tall an order. If you were to read and believe all the articles that say what a data scientist has to be able to do, it is a rare beast.
My frustration is, because I'm not just a researcher, but I'm also a part of Accenture and we deal with the real world, unicorns do not exist. So we decided to do a study to understand how managers are responding to this.
We found that there is a team solution for data scientists. We identified seven or eight specific skills that need to go into that data scientist team, and then talked to people like the chief analytics officer at Monster.com and places like that to understand how they were constructing those teams. And really, that is the case if you talk to the companies that are most advanced from an analytics perspective.
Q. I would have thought that one of the key things for a decent data scientist would be asking the right questions of data in a business context. It kind of comes back to the unicorn thing, but isn't it very hard to find people like that?
A. It depends on your perspective. Does it take someone with a unique skill set, or is it a failure of leadership to give them proper direction and support in context so that they understand what it is they're trying to accomplish?
I love these people. I love spending time with data scientists. But an executive for a big UK company in the UK suggested to me that if I left the data scientists to their own devices, I could come back and they'd have a wonderful algorithm that would give me 80% of the answer, and if I let them go another year it would be 85%, and they'd keep working on the same thing until it was perfect.
But the reality is, from a business point of view, I need that 80% today, and I need to take that and figure out how to put those insights into practice. And then they will get better data, and that will allow them to give me a better algorithm, better insights and returns.
There's a virtual cycle, and I think it's a responsibility of management to give them that direction, to set them on problems that matter, to help them understand what success looks like, to keep them focused and to complement them with people who can help translate business into technology and back again.
Q. We've spoken about data science capabilities and transcending the problem of finding fantastic individuals and instead going for a collective approach. Do you think organisations on the whole should build their own data science capability, or do you think they should buy it?
A. Many organisations want to have some core, in-house capability, but it makes sense to not rely exclusively on that, and the reason for that is twofold.
One reason is because business needs are dynamic. They're going to change, so you are going to need different skills. The second reason is really one that comes directly out of our research, which is the way to get the best answer.
If your goal is to get the best answers, the best insights, then the way to do that is not to take the same approach or set of approaches and apply them repeatedly. All the research shows that you will get better answers by bringing in more heterogeneous data and more diverse analytical techniques.
So, most people find value in bringing in diverse skills and capabilities. It also helps to have fresh eyes on a problem. One of the things that we find among analytical leaders is that, even among the people they have internally, they try to develop rotational assignments for them.
If you're always in a supply chain, there's real benefit in going to work on HR analytics. It gives you a broader view of the company. One of the advantages we have as consultants is that we get to see how lots of companies do things in lots of functional areas, so that's maybe an advantage we have.
Q. Looking at Accenture's clients, how do you see the relationship between boards and corporate IT, and respect of this newfound imperative for analytics. It seems to be that one narrative has been that CEOs have become interested in big data because they read about it in Harvard Business Review, the top financial press, and so on. They then go to their CIOs, their IT directors, their IT function, and say: “I know there's gold in this data, can you get me my gold? And a lot of the pressure that IT is experiencing comes from that and there's a real tension there.
A. There is a real tension there, and it's interesting because it plays out differently at different companies. In some companies, they don't even look to IT for those kinds of answers. So, I suppose you should be glad they're coming to you and not someone else, but the truth of the matter is that the expectations and hopes of the C-suite really have been raised tremendously.
When I first started surveying companies around their attitudes towards data analytics as a strategic asset, I was surprised that, say back in 2006, we looked at 19 industry sectors in 35 countries back then, and in every single one we found at least one company that said either “we view data analytics as a strategic differentiator" or "we aspire to do so in the future”. Well, eight years later, there are very few industries where you wouldn't find multiple people viewing that as their strategy.
One of the challenges IT executives have is that, too often, the C-suite view this as a technology problem or a data issue, so it's down to the CIO to say: “We can fix this part, but there are a lot of other pieces on this chess board that need to be moved. We need to change how we make decisions, how we execute on those decisions, how insights get implemented into practice, how we can make real-time, operational decisions."
There's a host of operational leadership and organisational changes that need to take place, and you do see examples of CIOs stepping up and taking that role. In fact, there are examples of CIOs whose job titles have been redefined.
Q. There's a lot of discussion at the moment, and this is a broader question to do with the so-called "second machine age", about the robotisation of so many jobs. That basically you're going to have a digital elite and jobs at the bottom, but there's going to be a whole swathe of the middle class professions that are going to be decimated by digital technology and so on.
A. Yes, Erik Brynjolfsson and Andrew McAfee wrote a book about this. Whenever you're at the cusp of a transformation like this, some jobs are going to go away, but a lot of jobs are going to evolve.
Some 30 years ago, when I first started working for Accenture, the talk was about programmers. There was so much demand for programmers that if you forecast how many programmers you were going to need it was going to be more than there were people in the US.
What happened was that what we used to think as programming is just something that everybody does. Everybody can download an app. Everybody can run a spreadsheet. That used to be programmers' work, now it's everybody's work.
But, I do agree with the second machine age in the sense that there's a risk. We do have an imperative as a society to build up our science, technology, engineering and maths (Stem) skills in our workforce, because there's going to be an incredible demand for those skills, and people who don't have those skills are going to be challenged to retool and prepare themselves for the jobs of the future.
In the developed nations that's a real issue because we tend to send our children more to the social sciences and the humanities. It's a luxury we can afford as wealthy nations. Whereas in China, India or Korea, there's tremendous pressure to send your child to become an engineer, a technician or a scientist.
If you look at India versus the UK, there are just a lot more Stem skills graduates per capita of population. So, maybe that needs to change.
Q. And that fills into the data science skills shortage. Are you seeing that change in the last few years, because there have been a lot of people saying that kind of thing.
A. It's certainly true that there's a proliferation. There's tremendous demand. I suppose it's more fair to say that globally we have a mismatch of where the skills are versus where they need to be. The jobs are here, but perhaps there are people in India or China or someplace who could fill those jobs.
But in the long run, technology's going to solve that problem too, because we have to get better at taking these tools out of the ivory tower and putting them in the hands of business people.
Q. How would you advocate that companies develop the right mix of quantitative and qualitative skills? It seems to me there could be a danger at some point that you would have gone too far with this data-driven decision-making.
A. Absolutely. The former chief analytics officer of New York City [Michael Flowers] said, at first, when he was trying to implement and help foster change in the city of New York, he had a mandate from the mayor, so he thought he might just talk about this and say we all need to be more data-driven.
But he realised it's not just about data and analytics, but that intuition and data analysis are like two wings on a bird -- you need both to soar. I think that is the right imagery. It's not data versus expert judgement. It's both. You will always get a better answer if you use both.
So, expert judgement informed by data analysis is probably the way to think about it.
And so, going all the way back to when we first started, doing surveys and research on what it takes to have affective analytical talent, it's always been quantitative methods, technological knowledge, business and industry acumen, communication, coaching and relationship skills.
You can't have someone who just sits in a closet and ignores the business, because they have to have enough context to be able to do the analysis in a way that's going to be valuable and then they need to be able to communicate it someone else. So, I think that's kind of the core of what makes a good analyst.