We're sorry but this page doesn't work properly without JavaScript enabled. Please enable it to continue.
Feedback

The Role of Data in Institutional Innovation

00:00

Formal Metadata

Title
The Role of Data in Institutional Innovation
Title of Series
Part Number
164
Number of Parts
188
Author
License
CC Attribution - ShareAlike 3.0 Germany:
You are free to use, adapt and copy, distribute and transmit the work or content in adapted or unchanged form for any legal purpose as long as the work is attributed to the author in the manner specified by the author or licensor and the work or content is shared also in adapted form only under the conditions of this
Identifiers
Publisher
Release Date
Language

Content Metadata

Subject Area
Genre
Abstract
There is a new sheriff in town – and his title is Chief Data Officer, or CDO. Presented by Deutsche Bank.
2
Thumbnail
1:01:39
5
Thumbnail
57:37
14
52
Thumbnail
1:00:21
55
Thumbnail
1:02:36
96
102
Thumbnail
59:03
115
Thumbnail
1:01:49
128
148
162
176
185
HypermediaPole (complex analysis)WeightXMLComputer animationLecture/Conference
Power (physics)Network topologyMathematicsPoint (geometry)Goodness of fitElement (mathematics)MeasurementForestLecture/ConferenceMeeting/Interview
Asynchronous Transfer ModeBelegleserTowerSound effectOrder (biology)Physical systemProduct (business)Identity managementElectric generatorBiostatisticsTerm (mathematics)Service (economics)Arithmetic meanFeedbackOnline helpTheoryPersonal identification numberComputer animationLecture/Conference
ArmFeedbackMathematicsGene clusterDevice driverMacro (computer science)Service (economics)Group actionPay televisionEndliche ModelltheorieInternetworkingComputer animationLecture/Conference
ArmMetropolitan area networkTerm (mathematics)Row (database)SurfaceCharge carrierService (economics)Sound effectNichtlineares GleichungssystemGroup actionDrag (physics)Natural numberCohesion (computer science)Context awarenessResultantCoefficient of determinationState observerBus (computing)TrailRoutingComputer animationLecture/ConferenceMeeting/Interview
Category of beingMultiplication signRight angleChief information officerFrame problemSet (mathematics)Product (business)Self-organizationEndliche ModelltheorieSoftware testingQuicksortInternetworkingVideo gameDifferent (Kate Ryan album)MeasurementService (economics)Computer configurationAxiom of choiceProxy serverResultantLecture/ConferenceMeeting/Interview
Executive information systemPattern languagePerfect groupPoint (geometry)Goodness of fitFeedbackQuicksortMereologyConnected spaceDifferent (Kate Ryan album)Endliche ModelltheorieInheritance (object-oriented programming)TheoryIdentity managementCollaborative filteringCASE <Informatik>Form (programming)Computer animationLecture/Conference
Metropolitan area networkFamilyFeedbackForm (programming)Distribution (mathematics)Endliche ModelltheorieProfil (magazine)Product (business)Element (mathematics)Different (Kate Ryan album)OutlierMultiplicationMechanism designAreaLecture/ConferenceMeeting/Interview
Distribution (mathematics)Touch typingProduct (business)Point (geometry)NumberData conversionForm (programming)InformationLoop (music)Interactive televisionElectric generatorWordOffice suiteVapor barrierMobile appMereologyContinuum hypothesisSystem callKeyboard shortcutTablet computerPlastikkarteMobile WebFeedbackComputer animationLecture/Conference
Metropolitan area networkMaxima and minimaEvelyn PinchingFamilyComputer-assisted translationFeedbackPoint (geometry)Mobile appLecture/ConferenceMeeting/Interview
Curve fittingOffice suiteProduct (business)Game controllerMainframe computerMetadataQuicksortPower (physics)Level (video gaming)Propositional formulaMultiplication signAnalytic setMultiplicationElectric generatorEvoluteOffice suiteInformationMereologyBitEmailBookmark (World Wide Web)Total S.A.Semiconductor memoryControl flowTwitterAugmented realityOrder (biology)Computer fileArithmetic meanTerm (mathematics)Noise (electronics)Capillary actionMathematicsInfinityTouchscreenHypermediaNumberComputer-assisted translationLogic gateInformation overloadConnected spaceUniform resource locatorElement (mathematics)Series (mathematics)Attribute grammarComputer animationLecture/ConferenceMeeting/Interview
TimestampPlastikkarteIP addressToken ringVideo gameGraph (mathematics)Core dumpGodSurface of revolutionTerm (mathematics)Open setMereologySineOrder (biology)CodeLecture/ConferenceMeeting/Interview
InformationMereologyConstructor (object-oriented programming)Level (video gaming)Total S.A.Element (mathematics)Multiplication signPoint (geometry)Goodness of fitData miningScaling (geometry)Process (computing)Office suiteMoving averageConnected spaceWordComputer fileSource codeInsertion lossMobile WebPattern languageParameter (computer programming)Firewall (computing)Information privacyInternetworkingLecture/ConferenceMeeting/Interview
Game controllerWeb 2.0Row (database)Office suiteWebsiteClient (computing)Element (mathematics)Endliche ModelltheorieModal logicIntegrated development environmentBuilding2 (number)Computer animationLecture/Conference
FeedbackProduct (business)Multiplication signLecture/ConferenceMeeting/Interview
HypermediaJSONXML
Transcript: English(auto-generated)
Thank you. It's a pleasure to be here in Berlin.
One of the first things I had to think about was whether an audience like this would really enjoy a nice long half an hour worth of PowerPoint. And decided perhaps not so. So, all I'm going to do is talk to you about what I'm thinking about when it comes to the role of data in innovation, particularly institutions.
To do that, it makes sense to start with the very idea of innovation. See, innovation is different from invention in at least two respects.
One, it talks about a change, and you can't measure change unless you have some concept of measuring. The second element is that it talks about usage. As they used to say, you know, if a tree falls in the forest and there's no one to hear it, does it make a sound?
An invention is meaningless unless there is adoption. Innovation takes place to try and make adoption. When I was young, I remember, you know, I'm old enough to remember the first ATMs coming through. And people were building these huge monolithic systems to help deliver banking services through holes in the wall.
And the theorists turned around and said, we have these problems of people stealing identity, concepts of like the pin came through. But even in the early 1980s, the answer became biometrics.
And there were people in researched ivory towers who actually believed that when you wanted to get, you know, 30 marks out of a wall or whatever it was in those days, that you would actually be interested in putting your eyeball against a laser scanner.
No concept of understanding what the human being would actually want to do. Would you trust an early 1980s laser scanner as a slit in the wall in order to draw money? I mean, instinctively, if they talked to two people, they probably would have got the answer, no way.
But money was spent on these things. Similarly, people designed services to say, I know what people will do on a rainy Saturday afternoon. They will stand in a queue and want to pay their bills through an ATM. Yep, totally normal. Why don't you want to pay your bills while it's raining in the middle of a queue?
That's when innovation doesn't happen. Innovation begins when there is adoption. And we're lucky, you know, I live in a generation, we live in a generation where the idea of adoption,
in terms of being able to say who uses it, how is it used, what works, what doesn't work, these things start having an effect because you have an ability to have feedback loops. Any engineer would want to be able to design products with active feedback loops because it's through the learning that takes place in the feedback loop that one is able to assess what's working,
what's not working, what's getting used, what's not getting used. For the modern internet firm, particularly those with subscription services, adoption is not just important. It is critical. It is the meaning of whether you get your bills paid
because if someone isn't using your product, pretty soon they're not going to be paying the bill. It's going to be a decay. So being able to understand where adoption takes place is critical. Before we move into that and looking further at the role of data and adoption, it's worth trying to think about what innovation actually means.
I spoke about the fact that it is through use, it is through adoption, it is through change. But it's the change, use, adoption of an invention. And the drivers for invention themselves are things that you can sit down,
as a scientist you would have to say, you'd be able to classify what drives an invention, how it becomes an innovation. The first and commonest reason for an invention to exist and for innovation to take place through the adoption and usage of that invention is a perceived need.
How do you perceive a need? Well, one way of perceiving the need is to be able to have some model to be able to trace what a need is and how that need is exposed, how it is surfaced, how it's collected together to understand the grouping and clustering of the stimuli that make that need up.
At a macro sense, you could turn around and say the jet engine was invented to fill a perceived need of wanting to be able to fly further. Some way of changing from propeller style use in terms of fuel through to the jet engine,
the design of the jet engine was to turn around and say, well, I need to fly at a higher sort of altitude, but within that altitude I'm going to get less air resistance, how am I going to be able to deal with these to get the trade-offs?
But the perceived need was to be able to fly further and someone like me wouldn't be able to be here except for that invention. The more common way is not just perceived need, but observed effect. And the observed effect route to creating an invention and then to getting that invention adopted to be able to innovate,
that comes from being able to look at something, record the results, track the results, understand the results. My favourite example of an observed effect becoming an invention is that of Velcro, right? The idea of saying somebody observes a dog running through bushes,
picking up burrs, and then finding that if you could design one surface that looked like the dog fur and another surface that looked like the tendrils on a burr, you got something with high cohesion and loose coupling that would not damage either surface or create pain
and yet at the same time be able to have the cohesion to be able to carry and drive through. And over decades it starts getting adopted, increased use takes place because the context within which you can do that adopting starts increasing once you start learning.
But if you look at something like that, the very nature of invention and then of innovation is actually based on the ability to observe, the ability to record, the ability to assess and create synthetics of what you observe and then to be able to build and adapt services as a result.
For human beings we just call that learning, right? We learn ourselves from the time we are babies by constantly putting forward some mental model, testing it, seeing what works and doesn't work and then adjusting our behavior as a result. Well that is as true for any invention
and it's particularly true for innovation because being able to listen to what the customer is saying about what works and what doesn't work is the difference between seeing adoption for a product or service and not seeing it. This is the way life has always been
and many of the failures that we see in innovation, many of the inventions that never went off the ground are where it has not been possible to be able to do that. I remember many years ago when I was CIO at a major institution,
I was trying to work something out with Apple at the time and it looked like I was going to spend some time with Steve Jobs. He came into the room and then he walked out. Didn't say a word. Then I asked one of the other people, you know,
doesn't he actually like doing business, what's the deal? He said he doesn't like CIOs, they are proxies for the end customer. He's really interested in hearing what the end customer wants to do. And that's really important now when I look at an institutional role,
always making sure that the voice of the customer is heard. What works, what doesn't work, when it works, when it doesn't work. If you look at, you know, modern sort of internet-era organisations, one of the things they do best is A-B testing, right? And an A-B is just almost a frame, a choice.
You actually can make that A-B-C testing for that matter, but being precise in saying I'm going to give a set of options, I'm going to drive them through, I'm going to watch what people do, and having watched, having observed, I'm going to allow that to track back
to be able to figure out how best to serve the customer. Now, that's easy to say and, you know, I remember when we do things like building recommendation engines, which many of you would have been used to, creating collaborative filters, you know, people who did A also did B.
Those kind of models are completely based on good feedback loops about customer usage. But, unfortunately, we still haven't quite got to the world where perfection can exist. So you take my case, you know, I'm a grandfather, I have three children between 18 and 30, and over the years, they've all understood that, you know,
dads are good for at least one thing, usually it looks like this, okay, and the wallet is always a useful parental sort of connection point to have. So what happens is that my Amazon account gets used by all my children,
which is fine, except that when I get recommendations that are based on what they buy, it's completely useless to me. So I then represent some hypothetical person, which is the conglomerate of four people of different ages and all our buying patterns.
So the very idea of identity, who it is at the other end of what you're watching, whether it's one person or not, in some parts of India and some parts of the Far East, you know, a PC may represent a village, okay, a mobile phone may represent a family. So the idea of how you actually take that data,
allow for the weaknesses in it, understand how you're going to be able to create some way of filtering the outliers. There is a way of doing it, right, and the best way of doing it is to actually empower the customer to be able to say, why don't you discard those things that aren't meaningful to your profile or behaviour?
So in somewhere like Amazon, I can actually do it obliquely by saying I already have that book or I'm not interested in this, so there is a learning that takes place because I'm able to make some feedback on recommendations.
But the premise is what is important, that any form of learning to be able to create invention, becoming innovation, becoming adoption, because it's that sustainable adoption that makes the difference in any form of business, and it's not just, you know, as Peter Drucker said,
the purpose of business is to create a customer. Having that customer relationship is what drives it, and knowing that customer becomes an important element of it. Now, many of the institutions we have built over the years actually work on a multi-tier model
where so much has been focused on production and distribution. The touch point with the customer disappears. So you have, you know, whether it's in the motor industry, the pharmaceutical industry, the soft drinks industries. I remember talking some years ago to the guy who looked after technology at Philips,
and why he was so excited about mobile devices, smart mobile devices, the coming together of the app and notification generation. And part and parcel of that was almost overnight, they could move from knowing 0.5% of their customers
to knowing 99.5% of their customers, right? Suddenly, there was a conversation. And like the guy said in the Cluetrain Manifesto in 1999-2000, markets are conversations, and conversations are at least bilateral, if not multilateral, right?
The feedback, it's the word backing that's important. The loop is a continuum. So we have the situation where, at least as long as I've been involved with any form of invention and innovation, I've had to concentrate on adoption. That adoption takes place because the voice of the customer is heard.
The voice of the customer is heard because I'm trying institutionally to pay attention to what the customer is saying. I then come on to the role of the Chief Data Officer. What has changed dramatically in the last 20-odd years
has been the number of people that are connected, the number of devices that exist, the number of ways we're able to make that conversation take place. When I look at an iPad, when I look at a tablet, one of the first things that comes to mind
is we have removed a huge barrier to interaction that has existed for perhaps a century. The QWERTY keyboard, or in Germany, I guess I'd call it the azerty keyboard, but the premise is that we had created an artificial intermediator
between the ability of a human being to engage with many information tools by adopting a keyboard. So now my one-year-old grandson has some basic understanding of what to do with an iPad because point, click, zoom, pinch are not things they need to go to school to figure out.
In fact, why my grandson? My cat can do that. And, yes, there are apps for cats, okay? The premise over there is once again saying feedback loops are possible because we started removing some of the control points,
some of the bottlenecks that came in the way of our ability to listen. In some ways, it's insane for us not to be able to get adoption of products because the customer's voice is heard in ways that were not possible beforehand, right? We talk about moving from mainframes to mid-range to desktop to mobile
to smart mobile to sort of now onto the wearables and embeddeds, but one of the key things that's changing and all that is not just affordability but the sheer number, okay? Everyone and everything is connected if my cat's going to have an iPad.
What happens when everyone and everything is connected is we get what lots of people refer to as information overload. In order to make sense, somehow you have to be able to extract signal from the noise because if everyone and everything is connected, you really have a challenge in terms of how to extract valuable information from it.
But one of my favorite professors, a guy called Clay Shirky, he said there is no such thing as information overload. There is only filter failure. And one of the reasons why people have gone on to a lot of modern social media tools
because something like email is, what, 45 years old or thereabouts, over 40 years old. And the reason why people are coming on to these is that there is a switch from publisher power to subscriber power. I have the ability not to listen.
I don't have to follow. I don't have to subscribe. So as soon as you go from thinking that control and filtering happens at publisher level to control and filtering happens at subscriber level, you're much more able to figure out what to listen to because the power is in you to choose what to listen to.
So one of the first things that happens when you have an infinite array of connections is that you start being able to listen to the right things by that choice because you're not having to deal with a firehose. Tools like Twitter work because you don't have a firehose. You're able to segment the firehose into sort of capillary chunks.
I can follow a person. And that becomes valuable. So again, part of what you start thinking about in the evolution of data and why an institution would even have a data officer is to be able to understand these things, to be able to know that while we can, you know,
I think a few years ago people said the total amount of data that existed since time immemorial now takes a few years to be created. And then it became a few days to be created. Right? But it is meaningless because none of us can actually consume that level of data.
How does that data get consumed? Because it's filterable. How does it get filtered? Because it has been tagged. It has been classified. It has a series of metadata, of attributes that allow me to select what I want. That's how consumption takes place. So institutionally, when you start dealing with a world where customers really are empowered to have a voice,
when they don't just have a voice but they work with multiple screens, multiple devices, where location has now become just an additional critical means to be able to augment information, where time has become another critical augmented means.
The way you need to think about it is, in the modern world, all data is just bits in that sense. You know? And the bits are themselves agnostic of anything else. And the first generation of value comes when you're able to associate that data with time and place.
Because suddenly, whenever people want to talk about analytics or big data or whatever, the proposition changes dramatically, the filtering capacity, because you're able to associate an element of data with time and place at almost no cost.
Okay? That's one of the things that the mobile device has done to us. We take it for granted, but we are auto-geolocating, auto-time stamping lots of things that we didn't before. After that, even if it's an IP address or a SIM card or some proxy token for what a person is,
the next thing that's happening is that we are associating data with a person or people, and quite often with the relationship graph of that person or people, if you have the permission to be able to do so, and if that has been exposed. Now suddenly, you don't just have knowledge of where and when, you also have an understanding of whom.
And those contextual pieces start refining the quality of the data available in order for you to be able to make sense in ways you couldn't before. And all this has been happening silently.
Beyond that comes a revolution that's much harder. That's why you see so much being spoken about in open data, which is that we use labels, and these labels are critical. You know, when I come to Berlin, I have to know that I'm actually flying to an airport
that in 1937 probably was called TXL in terms of a three-character ICAO name that nobody else would use except the airline industry. And in a strange way, it's fun, because I was born in Calcutta, and probably the only part of modern life that Calcutta still exists is in an airport code.
Peking exists in an airport code. Bombay exists in an airport code. You can't go there anymore, but CCUPKMAABOM, they all exist, because these labels are very hard to get rid of. They become part of the way people share information with each other,
almost adjectival in their construct. So today, we have these millions of devices, billions of devices. There are arguments about whether the total of connected elements is going to be between 20 and 50 billion by 2020. They're all capable of being sensors and actuators.
They collect information, and they send information on. They're made available in file hoses that tell people what people like or don't like, what's being used, what's not being used, how it's being used, what time it's being used, where it's being used, what switches people off, what switches people on.
But being able to make sense of it needs to have some level of discipline. How do you understand the where? How do you understand the when? How do you understand the who? And how do you do all this in a way that you protect the customer's privacy?
Okay, lots of people come and ask, you know, JP, what do you think? Who owns customer data? And the only answer I've had ever since I've been asked that question is, guys, there's a clue in the name. It's called customer data, okay? The word customer is there for a reason. But when we come to today, and the topic that I promised to speak to you about
was to do with institutional innovation and the role of the CDO, I wanted to build up to it to say, first understand what innovation is. To do that, understand what invention is. Then look at why data has become valuable or an issue.
Come to a point where you understand the scale at which that data is available. And then finally, you will be able to see what the role of a chief data officer is. So that's my role, I get called a CDO. What do I do?
Every CDO in a role like mine has four discrete jobs to do. One is to be a governor, okay? You have to sit down and establish policy, usage, patterns, rules. It's boring, it's block and tackle work, but it needs to be done, right? Good data is created because there is discipline.
The way I like to think of it is, in the old days, every large institution had a trust level on data that looked like mother's home cooking, okay? You trusted whatever came to you because you knew that your mother knew the recipes.
She'd always buy stuff from the right places. You could trust the food and the cooking. Data was like mother's home cooking because all good data was surfaced from inside the institution. Roll forward the clock 40 years. Now we have connected devices outside the firewall.
We have mobility, we have partners, we have alliances, we have the internet, we have the web. Most of the data that enters the firm is no longer created in the control of the firm. And if we behave as if it was built like mother's home cooking, then guess what? That's like having street food with a blindfold on and then wondering why you get ill, okay?
First role of the chief data officer. Create the governance model that people start learning how to consume street food, which means recognizing it, tagging it, cleansing it, having the necessary protections on it. Second role, data is useless if you can't get to it.
And we have created for ourselves an environment where getting to the data is not easy. So there is an engineering element to it. How do you make data accessible, consistently accessible, reusable, archived, brought forward on need, requested on demand with the same answer when you ask the same question?
There is a third role, science. All this talk about big data is really important, but get the basics first. You know, good data is important. Small data is important. The big comes because you've done the hard work to be able to extract insights. Insights for what? To serve your customers better.
Insights for whom? For your customers. And finally, finally, you can talk about building something that becomes a partnership to allow your businesses to prosper and your clients to prosper in partnership. And this happens because we understand we live in a completely new world with more data publishers than ever existed,
with feedback loops that allow us to get much better adoption of our products. Learning to listen to the customer can never be a bad thing, but remember it's the customer. Thank you for your time.