DN Meets... Roberto Maranca.

Roberto, great to meet you, we wanted to discuss the difference between the role of a Chief Data Officer in 2018 vs 2020, a really interesting topic, why do you feel passionate about this subject?

The first challenge you have when you start as a data function is to find a group of people that can think of Data as a home for their careers. People in finance want to be a CFO, if you are in IT you want to be a CTO/CIO, if you go into risk you want to be a CRO, but probably nobody in the last ten years joined a company wanting to be a CDO.

So my problem at the beginning was that I was going around, and this is as late as 2014, as the first of a need breed at GE, asking people if they wanted to come and work with me, but they didn’t completely buy into the idea of data as a full time job. The first problem was helping people jumping the barricade into something that was not incredibly defined. Luckily the role has evolved incredibly since then and people are way more aware.

Are you starting to see more of an understanding across businesses?

Yes, in some places, and suddenly everyone wants to be a Data Scientist. Quite frankly, I have never seen myself as a Data Scientist, Data Science, in the big world of Data, is quite narrow. But the problem is not only to find an insight “playing” with data, it is make it relevant and actionable. Some people who have heard me talking at conferences might recall that the joke that I can probably demonstrate, a relationship, or correlation between the shoe size of the CEO of a company, and their propensity to buy our product, but that’s hardly actionable insight. If you are not careful you can end up producing what I call “data entropy”, degrading energy, and it’s very easy: you take massive dataset, throw it into a data lake, add some code, and expect that the machine will give you something in return that is unbiased, clear, acceptable and beneficial. I am of course exaggerating, I have loads of Data Scientist friends!

But you ought to be incredibly intimate with the business, and incredibly careful when posing questions, because when you think about data science, or in science in general, the way you ask your question, could lead you into the answer, which is a risk. Your internal bias might lead you down a certain path.

To remain unbiased, must be a skill in itself?

It really is. Truly Artificial Intelligent generated insight has not been done yet. The data should not tell us: “Oh, yes, I found that...”. but rather: “Did you know that...?”

The second biggest risk is doing something unethical. Because we didn’t tell people yet that playing with data is probably not the right thing to do. We have a lot of young and smart kids who can do wonders with coding, and we are keen to them access to the crown jewels of your company so they could find for us the next data monetisation nugget. However, have we ever told them, that if you have access to large data set related to individuals, an innocent search about your friends’ details or a simple algorithm that searches for suitable soulmates in their post code based on some purchased product feature, is really an unethical thing to do

Logically, you could say “it goes without saying” but I really think that we should explicitly say that, because I don’t believe we did.

Are you of the belief therefore that AI should be regulated, in the same way we are starting to see regulation of data with policies such as GDPR?

I believe regulation is something that is imposed. Ethical approach is something that is born into you. It is your pledge. It your commitment. I often talk about this pledge at events, at this year’s Cognition X event (CogX2018), I pulled up a bit of a stunt having people pledging to a “Data Practitioner Pentalogue”, five statements like a sort of Hippocrates Oath, where the last one is the most powerful one, where you pledge: Never infringe human rights, or use data against humanity in general.

And you mentioned GDPR, that has got a set of recitals, recitals are not exactly regulation, they are setting the context of the regulation, its intended purpose and spirit. And the GDPR recital 4 says something along the lines of : “The processing of private data should be performed for the benefit of mankind”. Find me another regulation that mentions “mankind”!

We are playing with very tricky and sensitive things. And people are getting their heads around the fact that their data is something that is part of reality, but something that people expect increasingly to be handled in a correct way.

Do you think that one challenge is that ‘what is ethical’, is to some extent subjective?

You are correct. Ethics is alignment of interests hence relative to the actors involved. So in ethics we need to find a place of common ground. In a Venn diagram, the intersection of what we all believe is ethical, has to be the one we deem ethical. It is tricky, I find it fascinating.

For me the fascination isn’t implementing the next tool of data management, the allure is to resolve the problems like intolerance on Facebook, or proliferation of fake news.

Critical to solve the data problem is to establish a common language. And that for me is what is the most attractive as of very deceptive difficulty. If you think about it ambiguity of definition is behind 99% of the data problems, because when me and you are talking about this thing, you might call it a telephone, I might call It a mobile. And if you don’t believe it is a phone because you call it something different, we are at an impasse. So for me data has never been a technical problem, it is a language problem. It is a communication problem.

The other observation is asking yourself what makes a company great today? If you represent the typical company as a bubble, with four bubbles on top representing the typical attributes of a company: people, data, technology, process and the size of the bubble is relative to the company’s maturity or dexterity in managing that area, you will see that, unlike in the past, thriving companies have their Data and People bubbles out of proportion and merging in to one: in the 21st century neither process excellence nor technological prowess are the differentiators anymore. With how technology is accessible and commoditised now, I could BE Google or Amazon tomorrow morning. Tomorrow morning, I can open a shop, and my architecture can be built that morning with architectural components and functionalities that in the old days only the more mature large corporation could have mastered, and only after multiyear multimillion projects.

Today’s successful companies are successful because of how they connect their data and their people internally and especially externally. The more the technology advances, the more that technology is weaved into the fabric of reality and you don’t see it anymore, and people and quality of data is what is pushed into the foreground.

Is the data industry doing enough to prevent those ethical breaches?

Most of us are still bogged down in the harsh reality of fragmented system, fragmented teams, broken data, silo mentality, actually ..tribal menthality, because the silos that are impeding our success not physical. I can freely walk to the next office, but the problem is, that when I do it and try to sell my great idea on data, I will almost certainly be shown me the door, if I hadn’t previously convinced the “tribe” and most importantly the “chieftain” that what I do is aligned to their traditions (processes) and beliefs (goals) and it is expressed in their language (data).

Moreover, as we are bogged down in a physicality of problems and a complexity of relationship, my instinct is to push away from this and trying to create some formalisation layer.

Does that include formalisation layer for the general public?

Yes, because often we are bringing something with data that people cannot manage very well: the Truth, you cannot handle the truth. You cannot see the truth because there is a mismatch of expectations and language. If you could go back in time and give a caveman an iPhone, that would hardly accelerate human evolution and allow us to be the coloniser of the Galaxy thousands of years ahead of schedule, more likely he will either bite it, or use it as a weapon. Same thing here, I could give you this fantastic insight, but you have to be able to consume it, our communication tools are still based on parameters that are 50 years old and that impedes the emerging of new paradigms.

So is establishing the communications lines and processes a key part of the CDO role in 2018?

If in 2018 a Chief Data Officers hasn’t thought about proper communication plans, and they are just looking at patching infrastructure and centralising data, they will struggle. How you create the conducive context, and how that allows the people around you to really be bought into the journey, to get them better at understanding data, is by all mean a lot of graft.

So in 2018 the strategy should be to mobilise the company around data. True communication, true engagement. I have to have a media plan. I have to know the audiences. I have to follow a cadence, I have to deliver a clear message, with the delivery methodology tailored on the audience. I know the message to the top is a slightly different message to the one to everyone else, and different again to your key data leaders.

Probably the very first thing you want to explain is that Data is not IT. One of the things I had to clarify early on at Lloyds was tell them to stop thinking about data as the thing that comes out of your screen. That is not your data. Your data is the words you use to sell your product. Your data is what you communicate to customers, it is what we use internally to decide if we are doing good or bad. What comes out of your screen is a representation of that. So there is an element of soft skills that CDOs need to develop to ride this wave we see coming. It is almost a perfect storm. What used to be Chief Digital Officers are starting to become more data officers, because digital without data doesn’t go anywhere. The next thing that will happen is they realise without culture they go nowhere. So it could be soon Digital, Data, Culture to be unified.

How do you see the CDO role evolving, what might this role look like is 2020?

If in 2020, a CDO has got hundreds of people reporting into them, they will have failed. Because this is not something you do gathering legions, gathering groups of people to solve problems. I believe in 2020 a successful CDO will only ever have a one digit percentage of head count of the company reporting into them, if you want a number maybe 5-7%. If that is not the size, something has gone wrong.

Because if you feel your data needs to be centralised under you, then the mission has created a bubble that will soon be burst by the next cost reduction target. You will have missed embedding the data culture into everyone. This doesn’t mean that the CDO will disappear, think about the Head of Compliance, compliance is everyone’s duty and in good companies you don’t need to be told to be compliant, and yet the role still exists. Similarly, in a data driven company, looking after data is everyone’s duties, and the CDO team will still be there holding an overall strategic and design authority role, like loking after meta data, that’s fundamental, and looking after the standards and how you do things.

A picture I’m having of the role of the CDO is like a personal trainer. A data trainer. You tell the personal trainer what you want to achieve, maybe you want to run a half marathon or race in a triathlon, the personal trainer gives you the regime. But they don’t do it for you, you are the one who needs to build the right muscles and agility. Data should be the same.

So I hate when people come to me and ask for my data strategy. My response is usually “you give me your business strategy, and I’ll show you how to get there”. I can’t give you a data strategy in a vacuum as must be tailored to you.

So culturally that’s where I see it evolving. Realising that whatever you do from a data point of view can be destroyed by lack of proper adoption puts you in a different state of mind. You appreciate that change is perceived as pain by human beings: when we change habits our brain neuropaths need to be rewired and that’s uncomfortable, learning something new is painful and the tendency will be always to go back to the old way. That happens because genetically we are programmed to be animal of habits, remembering a safe path to the watering hole and following it is safe. Implementing a strategy, though, at the end of the day, is implementing a change: you are at point A and you want to be at point B, so if you are not securing that everyone really learnt how to go to point B, you will always be in danger of failing in your strategy (assuming that your data is good enough to tell you that you actually got to point B!)

Do you think a good Chief Data Officer can influence that change in mindset across a business?

Yes. Absolutely. A good Chief Data Officer must be an agent of change. To be a good one, you need to understand the forces that act when changing and try to make change more sustainable, cost effective and beneficial, as if you don’t, anything you implemented would be wiped out in 6 months anyway. All of your controls, and trust, will be gone.

Is the importance of data, and culture around data being accepted at board level yet?

No, I don’t believe so. Data is still seen as a technical thing. There are a few enlightened CEO’s, but overall, if you ask most of the top management, they will thing if data is wrong you could issue a policy and that makes it good. But usually the only people reading the policies are the people writing the policies. Policy are not change agents, winning heart and minds is.

What is it about this role that you love?

I have done IT for 20 years before I moved into data. I realised that it was where I wanted to be. It taught me there are things you cannot do from a technology point of view, these are things that are necessarily linked, and the more I studied data, the more I found People. What I love about this job, is that we are in an incredible moment in time. I can feel that we are in an age where we are going to work out how humanity will cross this chasm between physical and digital world and prosper over the next hundreds of years. And you cannot do that just with technology. Human sciences, once again in history, have a huge role to play in this. At high school I studied a philosophy and history, while my university years where in engineering, so I feel I can really see the cross over better than others. Our brain is 500,000 years old and hadn’t got the chance to evolve at the speed its surroundings have changed. Finding yourself together with billions of others, exposed to a barrage of data, it is stressful as we haven’t got the mental frames to cope, so a solution has to be found, but it won’t be a technical one.

Facebook can have thousands people reviewing fake news or taking down “inappropriate or offensive material”, but something must change as humans, in our approach, to prevent this continuing. Our reality has been extended with the additional dimension of data and we have to learn how to behave within it.

Look at how your generation absorbs reality versus my generation, it is mainly through social media. The way we learned in the past is gone. And going back to the CDO role, if the change is predicated on people learning something new, people like you will learn in a completely different way, so we need to make sure the communication is suitable for all demographics, back to the communication we talked about earlier. The must for CDOs in 2020 will be to make governance disappear. It must be weaved into the reality of the enterprise. Ethical choices are made based on a governance framework that is woven into your enterprise. It is not a decision, it is embedded. We should be so good at metadata, so good at access control, data classification, you get to your PC, you get it because it’s you. You don’t need a friend in IT. That data is available to you. You have the data you need, for what you need to do. More does not equal better.

That allows enterprises to have the data they need, and they will ask themselves: “If I use this data this way, am I depriving humans of free will?”. So the Chief Data Officer team of 2020 will probably be comprised of psychologists, visual artists and, why not, philosophers. We need to think differently about how we create products, and how we become a data driven company. And this is never be down to a new system, it is going to be driven by cultural change.