Episode Eight: Gaining Actionable Insight From Data

Using Data and Insights to Effect Change for Your Customers

In our latest episode, Paul Walker gives us an insight into his career at Goldman Sachs and his view on how data is foundational to all commercial, and non-commercial activities.

Listen here or subscribe.

Read full transcript

The Power of Data Podcast

Episode 8: Gaining Actionable Insight from Data

Guest: Paul Walker, Global Advisory Council Member, Motive Partners
Interviewer: Sam Tidswell-Norrish, International CMO Dun & Bradstreet

Sam 00:00
Welcome back to the Power of Data podcast and a huge welcome to Paul Walker.

Paul 00:04
Wonderful to be here Sam.

Sam 00:05
So Paul, you've had a wild career, you were a Goldman Sachs for a long time running technology. You're now an Industry Partner and Global Advisory Council member of Motive Partners. And you're an Advisor and Board Member. But most importantly, you're formerly retired.

Paul 00:21
Formally retired. Yeah, sure.

Sam 00:23
You parked it when you left Goldman, and then all of this stuff came along, in a space that you know so much about.

Paul 00:28
Well, you know, I like to say that, you know, I used to work really intense 80-hour weeks, and now I work really intense 18-hour weeks. And so, I still feel lucky to be able to have structured the life I have, but yeah, a lot of interesting projects going on, including ones with Motive.

Sam 00:41
Awesome. So let's get straight into it. Running technology at Goldman Sachs.

Paul 00:45

Sam 00:45
I mean, talk about a brand name on a CV, Goldman Sachs, the ultimate financiers, the ultimate in financial services and increasingly becoming a technology company with things like ‘Marcus’ coming to the market. What did you focus on when you were there?

Paul 00:58
Sure. First of all, I'm thrilled to have had the career I did with Goldman Sachs and I started working on the risk systems in the securities business in the fixed income derivatives and then expanded that to essentially run the risk practices for a large amount of the firm. I worked in the prime brokerage section for a while and ended up in the technology division where I was responsible for the entire suite of engineering activities. It was a fascinating experience, because there was never a moment where I knew everything I needed to know. And so, I was always learning and enjoyed that greatly. Look, the things I think I've always worked on in my career are the nature of software data analytics, you know, whether it was when I was doing my PhD using supercomputers to model black holes or whether it was when I was dealing with financial crisis trying to you know, understand the appropriate way to deal with our derivatives books or whether it was you know, now working with Motive portfolio companies on ways to do optimizations in capital markets or ways to use data assets in companies like Dun and Bradstreet.

Sam 01:56
So you talk about Marcus and you talk about Dun & Bradstreet, two firms that are leveraging data and leveraging machine intelligence, you advise many companies on AI. It's become a real niche of yours, and you're known as a leading expert. How much of a game changer actually is AI? And can you demystify a little bit for our listeners?

Paul 02:15
So I personally think that the data analytics that are getting lumped together under the phrase AI are going to be really fundamental for the way we undertake almost every commercial activity and many of our non-commercial activities as well.
There’ll be a big change in medicine, a big change, I think, about the way we think about education if we do it right. And a big negative if we do it wrong, as well, by the way, the ability to introduce bias in AI is enormous, and especially in the educational context, a very scary one. But you know, to demystify it some because I think you're right, people think it's magic, and it's not magic. It's a collection of mathematical and computational techniques that let you find patterns that let you make predictions. And it works for two reasons. First of all, we've done a bunch of math and technology, the computers are faster, the math is better, and they really have been mathematical advances, fundamental theoretical advances. But also it works because the digitization of our data processes creates data. And that data allows us to run analytics on it. Right? So if you'd have had the world's best AI engine in 1973, even with like, you know, today's computers, you got the GPUs, you got the TensorFlow, you've got the whole shebang, you know, all the neural networks, and you've got the multi-layer, blah, blah, blah, in 1973, and you went out to apply it to a grocery store. The first question to the grocery store would be like, what were your sales last week? And they'd say, Oh, I don't know. We sold some lettuces. And if you go to a grocery store now, they'll say; this zip code bought this much lettuce on this date from a consumer who had this card who also did… right? The ability to digitize our business processes and extract data gives us a backbone upon which these analytic techniques can provide real actionable insights. So you can spend a lot of time talking about AI and you know, neural nets and, you know, reinforcement learning and whatever the current, you know, exciting new paper that just came out today. And those are all really cool and fun and I, you know, I would love to do a different podcast about that yet. But the upshot of it for those of us who are actually trying to build and run businesses is; have we built business processes that capture all of our data? And then can we feed that data to these algorithms, so that we can give ourselves answers to questions that are relevant, actionable insights? And then can we change your organization so that they ask those questions and act on them? You know, changing a Salesforce with AI requires three things; it requires your data, of course, your AI, and it requires your Salesforce to read the damn report. Yeah, excuse me for swearing a little bit. You know that the last mile of someone does something with your AI is very important. And that's why I think the consumer AI companies like Spotify, and it's weird to think of me calling Spotify a consumer AI company, right?

Sam 04:52

Paul 04:53
But if you use Spotify and you listen to an album, it says you might also like these albums, and it's kind of right. You know. That is an application of data analytics that affects my life in a way that's actionable, right? And if every time I listened to an album, it told me no matter what you should go listen to Exile on Main Street, which is fine album, I wouldn't like that recommendation engine because it was never helping me discover something new. Or if every time I listened to an album, it told me something I didn't like, I wouldn't use it. But packaging, that analytic that Spotify has – I actually don't know how it works internally – but you know, I'm sure we could think about how it works, packaging that analytic and something that's actionable for individuals. That last mile is I think, the key success for AI and that's why I'm so excited about what you're all doing with Dun & Bradstreet right now.

Sam 05:37
And that is a really, it's almost poetic, actually, as an explanation. Right, the digitization of industries is creating more data. And using math to help find patterns to create insights is the application of analytics to that data.

Paul 05:52
And then changing your business processes to use those insights to effect change for your customers that they care about.

Sam 05:57
Exactly. And that's the virtuous cycle.

Paul 05:59
Exactly. And that generates more data. And then you can apply more math, not maths because we're in America now!

Sam 06:06
Yeah. So let's talk about where the most opportunity is okay, there's a ton of industries out there that are using data and analytics every day to improve, to make better business decisions, to increase revenues, increase margins, be compliant. What industry do you think is not using it as much as it could yet?

Paul 06:24
Well, look, I think there's real privacy and bias concerns that make applying it in medicine hard, but applying it in medicine really useful when we get it right. So, I think a lot of people are trying and healthcare and medicine to think about AI and there could be some remarkable results. Getting that right is going to be an interesting journey in the next five or 10 years. And so I think was last year that Google published a result that, you know, human radiologists had a 90% hit rate and AI radiologists had a 91% hit rate, but the 10% they both missed were a different 10%. So, a human radiologist, plus an AI radiologist had a 97% hit rate, right? So I think, you know, what we're going to see is a lot of these human industries, where people think that “it's a human industry, I'm going to replace the people with AI”, actually not going to work that way, you're going to find a lot of opportunity for augmented human industries, right? You still like having a sales person, that sales person is going to magically be 10 times smarter. You still like having a doctor, of course, that doctor is going to make 3% fewer mistakes, 7% fewer mistakes, who knows? But that's a lot fewer mistakes. And so I think, you know, one of the things that a real opportunity is integrating the result of AI with human-centered workflows in these environments. And so is that an entire industry that's missing that? I mean, every industry is missing that. The Amazon approach was to not have a salesperson. It was a great approach for buying barbecues, but it's not a great approach for you know, high net worth private wealth. So how do we help the high net worth Private Wealth Advisors say smarter things with less preparation time. Right? And I think that feedback cycle will be very interesting. Look, I think there's an awful lot of analytics that we're just starting to see come to light and process automation and games and things like that and applying those two industries more broadly very useful. The last thing I'll say is, I think the industries that are struggling the most are the ones that are yet to actually digitize their data. So if we forget about AI, and we forget about actionable insights, just that work of building your plumbing and architecture, so that, you know, in 1973, you go to the grocery store, and you ask how many lettuces and they don't know. But now if you go to a grocery store, they do, what other industries have those bad data practices, and how do we help those industries get those better data practices? I think that's another opportunity that could be ripe for innovation.

Sam 08:41
Yeah, there's a lot lots of take on board there.

Paul 08:42
Yeah, sorry, that was a very long answer. I if you're listening to this on the train, and, you know, you can press the rewind button, I suppose.

Sam 08:49
So a lot of what you just said, you know, we talked about the Power of Data, actually, a large part of it is the power of people. And how do you integrate the two together? I think as technology becomes somewhat commoditized, EQ, the emotional intelligence that human beings have, to apply reason to things is going to become increasingly important. What's your thoughts around, tou know, you think about Elon Musk who talks about you typing a text message on your phone, but you only being able to move as fast as your thumbs can move. And he's building a company called Neurolink to help do that stuff faster, move as quick as your brain is thinking, do you think that that type of integration is going to happen?

Paul 09:26
Sorry, I haven't spent a lot of time seeing what Elon Musk is thinking about human brain connective interfaces. So I'm not sure particularly what's going on there.

Sam 09:50
You can be forgiven.

Paul 09:51
Yeah, you know, I think the folks who are listening to this podcast the more actionable thing to do is think; what places are actual judgment and actually EQ and relationship and what's not? Right. So we used to think that recommending what to buy in a store was a real job that required insight to get it right. We used to think that recommending a piece of music was a real job that required insight to get it right. And recommending a piece of music sometimes is, but you can get 90% of the way there with a data driven process. So, you know, we've learned that the grouchy record store clerk who always yells at you for not knowing that 1979 album or whatever is not actually a useful way to go and discover new pieces of music. We make a mistake as technologists to think therefore, everything that is computable could be stripped of human kind. Right. I think it's a huge mistake. The social interactions people have, I talked about education earlier, there’s a lot of research into people having different learning styles, and I think this applies in education, but also applies in the corporate environment. Some people are fine to be given a book and a quiz and to teach themselves. Some people are fine to just be pure autodidacts. Some people find that social learning – a teacher, a group project – are ways that they learn much more quickly. And, you know, education researchers have theories on why people learn differently, but they do. And so to think that the way we're servicing our customers could be replaced with a single piece of technology, I think is probably foolish. So, what I would do is, I would think, sure, you know, maybe Elon Musk has a brain to computer interface, and then he can stop interacting with human kind at all. He can just sit in a tank and tweets emerge or something like that. That doesn't sound to me like something he wants or something any of us want. And I haven't looked at what he's doing. But I think that most of our people who interact at a bank, at a wealth management moment at a decision when they're allocating capital to their company, at a point where they want to sell an asset that they find valuable, are going to want to have a human interaction. So how do we make those human interactions richer and more robust for all participants? And how do we make them more efficient? That's, I think, the real way that you should think about human-computer interfaces that are relevant.

Sam 11:50
There you go, I couldn't have asked for a better answer to my ridiculous question.

Paul 11:54
Well, you know, and then maybe I'm wrong and we'll all be brain tweeting in two years.

Sam 11:59
Looking forward to it, if Twitter is still around in a few years.

Paul 12:02
And brains.

Sam 12:05
We're going to end on a couple of larger questions. You're a philanthropist at heart. I know you're constantly wanting to give back and you spend a lot of your 18 hours a week on philanthropic endeavors or maybe don't include those in your working hours.

Paul 12:18
Yeah, a separate project for them.

Sam 12:20
A lot of what you're doing is about giving young people opportunities to fulfill their potential. What advice would you give a young adult today? I mean, the world is changing every second. Phones are creating anxiety amongst children that is unfathomable. What advice would you give them?

Paul 12:34
I think it would depend on the young person as well, of course, but I think there's a couple of things that are super important. When I look at the next 20 years, the rate of change is going to continue to be enormous. And so if you haven't trained yourself to accept mental flexibility, if you haven't trained yourself, whatever your learning techniques, to be able to learn. I think you're going to have a hard time finding a way. That's probably more useful than any particular skill. There’re still fundaments, there is still knowing calculus or still speaking Spanish, whatever, right? But the adaptable learner is going to be a learner who's going to do well. I think one of the things that's a constant of the future is people who are willing to work hard are going to continue to succeed. It's not a sufficient condition, but it's a necessary condition to success. I think the other thing though I would say and I would say this to the young people, but also say to the older people who are bringing those younger people into their workplace, is that we think very inefficiently about the value of social capital, about the value of privilege. And so, you know, my advice to a young person is great. My advice to an older person like myself is how are you distributing the social capital and the opportunity you have in an equitable fashion, so that a broader swath of the best members of society to solve your problem can be brought to bear on your problem. It's unlikely that the best x y z also happens to play lacrosse with you, or also happens to play Dungeons and Dragons with you, or also happens to be into this indie band you are, or also happens to play ice hockey or also likes Star Trek or whatever it is, you're using as a filter of acceptability and not realizing it. So to the young people, I would say, you know, have confidence, work hard, be adaptable learners, and expect people to give you opportunities that you can meet. But to the people who are doling those opportunities out today, I would invite you all to think are you doing it in a way that's equitable and reasonable?

Sam 14:25
Final question, it's a random one, but I've always wondered, what’s on Paul Walker’s bucket list?

Paul 14:30
What's on my bucket list?

Sam 14:31
What's at the top of your bucket list?

Paul 14:33
You mean like things I want to do before I die? Other than don't die?

Sam 14:36
Yeah, I wasn't gonna use the D word.

Paul 14:38
Yeah, you know.

Sam 14:39
Stuff you haven't done.

Paul 14:40
Well, look, I'm lucky enough to have found myself in a situation where many of the things that I want to do I can do. And I don't really think about unrequited desires all that much. I'm lucky to have had the path I did and found my way to the projects I think are interesting and people who I like spending time with. There's more of the world I would like to see, sure. You know, I've never been to Vietnam, Vietnam seems cool. But I wouldn't put that like on my bucket list of some defining thing, that it's some sort of massive, overarching goal. You know, I hope that when my daughter is 10 years older, we can have dinner and still make fun of each other and laugh, right? You know, I hope that I'm still enjoying doing things with my wife, I hope that I have professional relationships where, you know, I find them engaging and people find me useful. But I don't have some big thing that I've always wanted to do, because if I did, I do it.

Sam 15:27
I was half expecting the Paul Walker restaurant chain where you order with your thoughts or something, but that didn't happen.

Paul 15:33
You know, I don't know a lot about every business in the world. But I do know never start a restaurant is good advice.

Sam 15:41
Certainly not for financial gain.

Paul 15:42
Exactly. Exactly. Exactly. Or for lifestyle. Well, Sam, it's been a real pleasure being on the podcast today.

Sam 15:47
Thank you very much. It's usually my role to end. Yeah, you just ended it for me.

Paul 15:51
I thought, was it the last question the final question, though? You know, I know that our listeners have only so much time on their iPads and iPhones to listen to us.

Sam 15:58
Exactly. There's not many 45 minutes Tube commutes.

Paul 16:02
That's right.

Sam 16:02
All right, Paul. Thank you. It's always a pleasure and a privilege and we look forward to speaking to you soon.

Paul 16:07
Anytime. Thank you.