Steps for Using Data & Artificial Intelligence to Make Business Decisions @DeborahLabsOW (Episode 42) #DataTalk

Listen to the podcast:

Every week, we talk about important data and analytics topics with data science leaders from around the world on Facebook Live.  You can subscribe to the DataTalk podcast on iTunes, Google Play, Stitcher, SoundCloud and Spotify.

This data science video and podcast series is part of Experian’s effort to help people understand how data-powered decisions can help organizations develop innovative solutions and drive more business.

To keep up with upcoming events, join our Data Science Community on Facebook or check out the archive of recent data science videos. To suggest future data science topics or guests, please contact Mike Delgado.

In this #DataTalk, we had a chance to talk with Deborah O’Neill at Oliver Wyman Labs about the ways to gain a competitive advantage with data-driven decision making — and steps to adopt artificial intelligence in your business.

Here is a full transcript of the interview:

Mike Delgado:  Hello and welcome to Experian’s weekly Data Talk, a show featuring some of the smartest people working in data science. Today we’re excited to feature Deborah O’Neill who’s a partner with Oliver Wyman Labs. Deborah has worked with a range of clients around the world in data insights, risk management, and customer centricity. Today we’re going to be talking about the ways that companies can better leverage their internal and external data to make smarter business decisions. Deborah, it’s an honor to have you on our chat today. I apologize for all the technical difficulties earlier.

Deborah O’Neill: Well, I suspect it might have been some crazy cyber security restrictions on my side. We obviously take privacy of data seriously, so I’m just glad that we finally made it work.

Mike Delgado: I know, this is great. So happy to finally connect with you, Deborah.

Deborah O’Neill: We’re platform agnostic, as it were.

Mike Delgado: That’s right. I thought it’d be great, Deborah, to kick this off if you shared a little bit about your background and what led you to what you do today.

Deborah O’Neill: I’m a career consultant so far, that’s the short part of my career. As I know, I’ve got a long way until retirement these days, but so far I’ve been a career consultant. I, as anyone here would know, joined a consulting firm in 2008, during the financial crisis. Spent a lot of time doing finance and risk, so traveling around the world helping banks figure out their losses around all the mortgage defaults, helping them figure out the new wave of regulation. Generally, it was a little bit of the time, and that was the topic of the time.

Mike Delgado: Yeah.

Deborah O’Neill: But also, I guess my physics background gave me kind of a love for rigor and data, and I was probably more suited to analytical consulting than I was every going to be to blue sky thinking. And I found I loved it. Being able to build interesting models that allowed you to work out the positions of complex products on a bank trading platform was really cool. But what I realized was I’m a little bit of a stickler for wanting things to be right and accurate.

So, I very much ended up in the world of spending a lot of time with IT departments looking at data and models, and sort of going, “Okay, how do we actually get these decisions embedded into the organizations.” So I actually became a little bit more of a change manager, as I got more senior, and still very much in the data legacy, but in that direction. And I think the way consulting was going, there was a much more technology delivery angle to it. And then I guess over the last three or four years, I’ve landed myself in our labs practice, which is the technology enablement capability for Oliver Wyman Consulting. It’s where all the great consulting work, and analytics that everybody’s working on get embedded into cool tools, and analytics platforms that actually live and breathe in these organizations.

It’s nice to move into a world where it’s sustainable. I think the new world we’re in of open source technology, and all these platforms is allowing that world to become very vivid and quite real for everybody, so it’s a cool place to be. It’s very fun. A lot of traveling still. I am definitely still in that up in the air kind of category of living on a plane and out of a suitcase. But it’s exciting times. I think it’s a great time to be in consulting. I think it’s a great time to be working in technology and financial services. I think we’re at a turning point. That’s me in a nutshell.

Mike Delgado:  Yeah, it’s awesome to hear. And what’s interesting Deborah, is as I’ve talked to different data scientists, hearing how their career has progressed. You were just mentioning how your background was in physics, and how that eventually led you into working on models, and working on algorithms. It’s just fascinating to hear how your career has progressed in data science.

Deborah O’Neill: Yeah. For me, what’s interesting about it is if I was honest, when I was leaving university, I always thought I was the non-heavy lifting side of physics. I enjoyed physics as a concept, but I was much more on the practical end. I wanted to be getting something landed. I bizarrely at the time, loved project management more than I loved math’s and things like that. I was very much a “let’s get stuff done.” Consulting’s been that happy medium of the people who really like numbers and models, but want to be doing something that’s perhaps a little bit more, has an end in sight. It’s not research that can go on for years. I think data, big data, computer science, machine learning, all of these topics, career light consulting’s a great place to do them.

Mike Delgado: Yeah. I know, definitely those field are hot right now. I just saw a chart of Google Trends, people searching for things like data science, artificial intelligence, machine learning. And we’re now seeing machine learning as one of those keywords that’s now surpassing big data. For a long time, big data was the big buzzword. And now machine learning has overtaken search queries for big data.

Deborah O’Neill: Yeah. I think it’s interesting, because I think big data, for a long time, I don’t think people really understood what big data was. It was a great buzz term, but most people’s data’s not big, in the real academic sense of what big data is. But machine learning is applicable in any field. It doesn’t matter how big your data is. And I think that’s the realization. I still find very few institutions that are working the realms of the big data as purists would call it.

Mike Delgado: How would you define big data, because it is a buzz word. People do throw it around a lot. I’m kind of curious. How do you define it? Coming from academia, now what you’re doing now?

Deborah O’Neill: I think there used to be a lot of theory that it was about size and volume. For me, big data is more of a complexity question. Is it different types of data that need to be stitched together and used in a coherent manner, that being is it speech to text data? Is it unstructured data? Is it fast stream data from trading platforms? Is it static customer data? That, to me, is sort of the realms of what I would call more complex data than big data. The bit I’m most interested in is how you use that to make decisions.

I think it doesn’t matter how much data you’ve got, there’s interesting things you can do with it, and there’s a richness in it, if you apply the right techniques. And that doesn’t always mean jumping to the most complex techniques in the world. It’s more about getting things in a sensible manner, to get that value out, I think is the most important way I would look at big data.

Mike Delgado: You just mentioned, I think one of the difficulties that a lot of businesses have, and all of us have, is dealing with the unstructured data, because there’s so much of it. As you advise clients, with the amount of unstructured data that they have, what do you suggest to them?

Deborah O’Neill: I think it depends where they are in the journey. There’s a lot of organizations out there that are very tempted to jump to the most sophisticated level of, “We should use all of our unstructured data. We should get an AI platform. We should have chat bots. We should be automating this, and automating that.” A lot of the time, they’ve been caught by the buzzword, or they’re scared of if they’ve not got a block chain project, then they’ll be not cool in their group of peers.

A lot of the work that I do is around just dialing that back a little bit, and saying, “What are the decisions you actually need to make? What is it that’s difficult today?” That could be an example of wealth and asset management firm, who, they’re still very client focused. They’re still having one on one discussions around trading and positions. It’s still a relationship based business, so shoving a chat bot in giving advice is not the right thing to be doing. But equally, they’re not. There is a theory that they could be using AI to generate really good insights around if the market moves in one direction, what does that mean for their consumer’s trades? Should they be hedging something? Should they be making a different decision, which I think is right. But to get to that point, you need to move away from the world that they’re currently in, which is 75%, 80% of their time could be spent manually going through different data sources to find the thing they want to talk to their clients about. And at this point, you’re talking about process digitalization.

How do you get that single view of a customer, just from the basic data you’ve got? Can you even pull up their latest positions, their latest contact details? Have you even got a view of if they changed jobs via LinkedIn? Can you even get that single view? And then you can start layering on the, let’s do some AI around what their next position should be, that they should place. Can you eventually outsource some of the advice to a robo-advisor, because sometimes it’s more time consuming for a client to ring up and ask you a question. They may like the idea of getting self-served. But it’s such a spectrum journey, and what I see so often is, arriving at a client and they, “Ah, we’ve just procured this platform,” and they’re just not ready for it. They haven’t even got the skills to even maintain it once the project’s over. They haven’t got their data either complete or accurate, so anything they churn out they won’t be able to rely on. And they’ve just got a basic process issue at the start of this journey, that they just need to understand the decisions they’re making, and why they need to make them. So, it depends on where they are on that spectrum when you walk in the room. But sometimes you do have to dial back the buzzwords a little bit to get to where is the kernel of value.

Mike Delgado: Deborah, as you’re talking with these different companies around the world, so many of them are wanting to start to use artificial intelligence and machine learning. But like you said, many organizations aren’t ready for it, because they’re not even using data analytics correctly in their own organization. What are some of the questions you’re asking them to figure out where they are in the journey, before you begin advising them on what AI tools they should be using?

Deborah O’Neill: It’s usually around what are the key value drivers in your business? Where do you make money? Is it pricing? Is it on cross sal? Is it on not churning customers? Sometimes just retaining the customers you’ve got is value in itself, particularly in this competitive world of startups coming in to threaten the playing space that banks have traditionally dominated. And once you can identify that kind of value driver, you can then start saying, “Okay, what does that process look like in your organization? How do you address that value driver? Is it that you’re in a great position where you’ve got a view of all the interactions your customers have with your organization, and you can, you’re already looking at where there were points of failure. Making sure you address those, both for an ongoing process, but maybe having that client on a better watch list, to make sure they don’t feel that again.”

If they’re saying to me, “Oh, we look at KPIs on a quarterly basis,” and you’ve all been through an airport where you see their red, amber, green of how they’re doing on delays, and things like that. But if they’re at that point, then churn is a big issue, because it doesn’t matter whether 90% of the flights were left on time, but if you travel every week, and you were on the 10% that were always late, you’re a churn risk. But they don’t know that, because they’re looking at that overall number. So, it’s trying to assess what data they’re using and how they’re using it, and what they do with it, to understand how sophisticated they are.

Mike Delgado: Yeah. That’s fascinating. I think that because companies have that combination of all this internal data, data that they’re collecting from their clients. But then there’s also that external data, maybe that they’re purchasing, and then, to your point, how are they leveraging it right now? Where are they in that journey? And then from then, at that point, you’re going to figure out are they ready to begin adopting some AI technologies?

I think one of the difficulties, we’ve all heard that saying about garbage in, garbage out, when it comes to data, and the art of data science. You don’t have the right data. You may have a ton of data, but if you’re not looking at the right data, or finding the right data, it’s not going to be providing you much help. So, I’m curious about what advice to you have for brands, for companies that want to use AI effectively, that want to be finding the right data, but maybe are not sure where to look for it.

Deborah O’Neill: Yeah, so it’s two things for me. I think increasingly, we’re seeing more people focus on data strategy as an actual strategic goal. Previously, you would have your brand strategy and your product strategy. Data strategy is a real thing now. How do you measure your data quality? How can you certify a piece of data as being suitable to make decisions on?

We’re talking to a lot of people around what are those measures? A great example would be if you are trying to use AI to make credit decisions. You’re trying to decide whether to lend someone some money. You run your algorithm, and one of the big factors in there is clearly someone’s credit bureau score. What is their credit rating score? Now, it doesn’t take much for that to miss. That data is not refreshed on a frequent basis, because maybe the batch file that came in from the credit bureau was corrupted, or you didn’t update the system, and suddenly your algorithm’s telling you this person’s a great credit bet. And then you look at the score, and it says they’re an A rated, or a 600, or whatever number. And then you drill into that, and you realize that’s two years out of date. A lot can change to someone’s credit score in two years.

The algorithms can run very effectively, but before we get there, the strategy around what are the KPIs? What’s the reliance people can put on that data? And that might firm timeliness, accuracy, certainty. We see a lot of models now where people are using external data sources to try and triangulate. So, if you’re trying to triangulate someone’s credit score from their transaction data, you may not have all their transaction data, because they may have multiple bank accounts. From their credit history, that’s great, but if they’ve moved house, and not linked their address, that can take a while to catch up with you. Setting these measures so that when your data science team, or the team that’s using analytics to make decisions, they have something that say. “This is an amber field. It’s only 75% reliable, so if it’s coming up as a very high factor in your model, treat with caution.” And that strategy is a strategy that must be rolled out across an organization, because often you find data repeated in multiple systems. Which is the right data? Who is the person who makes that decision? And that links a lot to how do you govern your analytics teams? Do you have a centralized analytics team in your organization that churn all the analytics for the entire organization? Do you have analytics embedded in each of your business units, so each business unit has their own dedicated team that will be smaller, but more linked with the business goals?

There’s different answers for different organizations, but it’s a real choice now. It’s a strategy. It can’t just happen by accident, because otherwise, these algorithms work so fast, as you all know. They churn out results. They’re not necessarily easy to understand once you get into the deep neural nets. Actually, explaining the reasoning for that becomes difficult. So, it must be a deliberate choice the way you go. You don’t have that pace to have time to catch mistakes before big things are done.

Mike Delgado: That reminds me of something you wrote in a recent Harvard Business Review article about data analytics, and people working data analytics sometimes have blind spots. Can you talk a little bit about those blind spots?

Deborah O’Neill: Yeah. So, it comes in two flavors. One can be people get a little bit too attached to what’s happened in the past. Your data is historic data for a reason, and the algorithm can be faultless in terms of its determination. My view is AI doesn’t risk people’s jobs. I think there’s more opportunity for people to spend time thinking, and making decisions, rather than having to do manual stuff. It should be seen as a bonus, not a threat. And you still need humans to overlay what the machine doesn’t know. A great example, that I’ve seen a lot, is people drawing conclusions about what may happen in the future, and making decisions off it. If you were the buyer for a supermarket, and you’re in charge of stocking products for special events in that area. You may be looking at a supermarket which, for the last ten years, they’ve always sold a lot of beer on one particular weekend. They’ve got a lot of people buying beer. Seems like a sensible idea to go and put more beer in for that weekend. It’s a trend that you’re seeing. Everyone’s going to buy beer.

Mike Delgado: Yeah.

Deborah O’Neill: But what you end up doing, if you drill in, and you sense check it, you find out that the reason that that beer sale was really popular was that there was a local event or a festival around the corner. And guess what? The festival’s not happening this year. Especially in Europe, where festivals happen in rural areas, if you just then buy all the beer without thinking about it, you’ve suddenly got a rural supermarket, with a load of expensive beer, because beer’s an expensive product. But they’ve had to forego their regular groceries to make space for it. And suddenly, your local residents, who use your shop every day, are pretty fed up. They haven’t got any milk, or bread, or anything else. And so you have these detrimental facts which, if you overlaid the human touch of, “Oh, let me just go and double check if that event’s data field is up to date,” or, “Have I read and checked whether that’s still happening,” you catch those things.

The data can make you quite blind. That’s why you don’t want a complete robot just choosing all the buying for all the supermarkets, with no overlay of checking. It can have some silly outcomes.

Mike Delgado:  That reminds me. I don’t know if you heard the story, but Walmart was looking at the purchases of those strawberry pop tarts and noticed that dramatic increase in sales, in Florida. But it was because of hurricane season.

Deborah O’Neill: Yeah.

Mike Delgado: And for some reason, during stressful weather conditions, people were buying, in bulk, strawberry pop tarts.

Deborah O’Neill: Well you also hear the stories about items for sale online, where the two buyers match the prices. I want to be 95% cheaper of the price of the other one. The other one says we’re out of stock, so let’s rack the price up to 120% of the other one, and you get this great outcome where they just ratchet each other up. I saw a great blog where someone had ended up tracking this, and some book had got to a million dollars or something.

So the data can make you blind, because although history’s a great help, and let’s not underestimate it. There are a lot of day to day things and habits, especially tracking habits of people, because people don’t understand their own habits, so checking buying habits. Particularly if you look at the global population, if you’re looking for trends. That noise disappears. If you’re making big decisions for individual, individual supermarkets, or individual places, you’ve got to be a little bit more careful. And that’s what we were getting at with that data can make you blind.

Mike Delgado: Not only can data make you blind, but also, for those of us who have gut instincts, or certain biases that we might not even be aware of, that can also mislead us. Can you talk a little bit about those gut instincts?

Deborah O’Neill: Yeah. This is always a tricky one, especially if you’re working in a specialist area like trading or relationship management for wealth and asset, gut’s quite important. The last thing you want to do is tell everyone that they’re not allowed to have a gut instinct. Having a relationship manager that’s known you for 25 years is still going to be the best general feeling of how you’re going to react to something. But there’s lots of ways that I’ve seen organizations deal with this via some work around making things more suggestive.

Let’s give people suggestions of things they may want to talk to their clients about so you’re not telling them they must do something. You’re not de incentivizing them to say, “You must do this.” You still give them the override of, “No. I really believe …” But you surface the information in a way that’s interesting to them, maybe not the way they’ve seen it before, and will just give them that feeling of, “Oh, there’s something there,” and enough backup information for them to challenge their gut to say, “Oh, I hadn’t looked at it that way. That makes sense.” That’s one way.

And then the other way is around the statistical tools that you put in your data science team. There’s a real importance that, or a real fair for me, in some organizations where they’re still getting to grips with data science. It’s still a little bit experimental. Each data scientist will have their favorite statistical test that they like to run, their favorite sample tests. And just standardizing those a little bit, not necessarily taking away the creativity, but just saying you must have at least done these four or five things, because then it means everyone in the organization knows those things, and then they understand the metrics they’re getting around, or anything else, is something that’s standardized enough for them to have that conversation.

And a lot of it comes, again, to governance and the desire to change. Leaders also must be willing, in these organizations, to admit they’re wrong, and the data’s saying something new. Having that leading from the top of, we’ve had the organizations who do the interesting data thing I found this week. They send an email top of the house and say, “I didn’t know this about the organization this week. This is where the data helped me.” And it can drive that information, I don’t have to pretend I’m the expert, and that the data can’t possibly know more than me.

There’s lots of little tips and tricks, but it does have to want to be led from the top. I think it’s important that the whole organization believes in this, because otherwise people can feel quite victimized, if they’re the person that’s brave to say, “I’m gonna try something different, because I think that’s what the data says.”

Mike Delgado: Deborah, I’d never thought about what you just said about even choosing a particular statistical model could be a bias.

Deborah O’Neill: Yeah.

Mike Delgado: And your talk about standardizing. I never thought about that.

Deborah O’Neill: Yeah, but we all do it. I mean, even in the old days of doing stuff in Excel. When I was looking at work of my colleagues and things, I would always go in and do the same few keystrokes. Let me sum that column, and let me divide it by that. Your natural reactions to how you test that. And that can be quite dangerous in itself. And we’re all guilty of it. “Let me go and pick up my three pieces of core code that I always run.”

Mike Delgado: Yeah.

Deborah O’Neill: That’s why we should all work in a world of being open and sharing things. But again, that’s a cultural change, but open source is helping that I think. I think it’s a real positive step forwards.

Mike Delgado: You mentioned block chain a minute ago. Many financial companies are investing in FinTrack.

Deborah O’Neill: Yeah.

Mike Delgado: I was reading about Citi, Goldman Saks, JP Morgan, Morgan Stanley, Wells Fargo. They all have huge investments into block chain, and I’m curious on your perspective on how important FinTech investments are for financial firms, and any advice you have.

Deborah O’Neill: Yeah. I think FinTrack has a place in this world. I think innovation is generally important and I think we’d all agree that that’s why we’re making so many strides forward. I do think they’re a great opportunity for firms to get access to new technologies, trying new things, in a way that’s a little bit quicker, a little bit less cumbersome than trying to start something internally. It’s also lower risk. You can try a POC, a proof of concepts. You can have a go at it and see how it fits.

However, I do think you still must go back to the basics. I keep saying it. You need to focus on the problem. What are you trying to solve, and find solutions that might help that, not just get dragged along, and it’s the hot new thing, cause unfortunately the hot new thing changes relatively regularly, and actually it can be quite draining for the organization, because you need to see something through. Not necessarily to the end. If it’s the wrong thing, you need to have the no fear of failure. But you do need to give it a chance. Everything has a little bit of a realization cycle. It’s not overnight. So just always step back and say, “What are we solving for,” rather than saying, “We must need a block chain program.”

Mike Delgado: To me, it’s exciting to see this, because, especially in these highly regulated industries, like financial firms, oftentimes, those are the ones that are slowest to adopt newest technologies, because of regulation and compliance. Do you think that these adoptions of block chain and AI are going to force a lot of these financial firms, or just companies that are highly regulated to be more nimble?

Deborah O’Neill: I think there’s a few things going on. It’s an interesting time in financial services, as it has been for the last 10 years, but I think there is, obviously, a lot of these cases now where the value of AI and machine learning, particularly, are being seen as important. So fraud protection, being able to very quickly scan a lot of data, say does this look like an outlier? Can you predict churn, because you can see all the interactions someone’s had across all the different channels that you operate in. Also, people who’ve churned like them. So, I think they’re seeing a lot of value in this, so I think there is a risk of them not moving into this space and someone gets there first.

But they do, at the moment, and it’s going to be a fleeting moment, still have an advantage. If they can get this stuff up and running, they have a lot of data. They have the history. Some of these banks are a hundred years old. Not saying they’re going to get a hundred years’ worth of data, because it’s probably in a bin somewhere, or it’s destroyed. But they have got significantly more than some of the FinTracks and things, or the startup banks, and all sort of things. But that is changing as well, and that is a real risk. Particularly in the new year, we’ve got the general data protection regulation, GDPR coming around the horizon, and that’s going to put a lot of power into consumer hands. I mean, GDPR requires one, that you have the right to be forgotten, which means you can ring your bank or your supermarket, who you might have a loyalty card with, and you say, “I want you to forget me.” And that doesn’t mean just delete my customer stuff. I want you to go get rid of all my purchasing information. I want you to get rid of any product I’ve applied for and been declined for. All that credit rating history. You need to be able to delete the law. But the other thing it gives you is the right to take your data with you, which is an interesting concept.

If you, as an organization, let me down and I decide not only do I not want you to have my data anymore, but I want you to give me all my information. I want you to give me all my information, and I can take that to another organization who have got better customer service, better personalization, better product mix, and give it all to them. And suddenly, they’re getting data that’s valuable, that I’m happy to give it to them, because they’re giving me a service that I want. And you can imagine what the dynamic of the arms race around this is going to be. You’ve got supermarkets suddenly swapping people’s buying histories. That’s interesting in itself. It’s going to be an exciting year when this comes in. And the fines are not insignificant either.

So I do see that banks are going to be getting going on that. And by the way, GDPR, you could have a whole talk on that. It’s fascinating in its own right. It’s fascinating as regulation can be to some people. But it is a risk, and I think they are waking up to it, but is it too little too late? Let’s wait and see. But I think the FinTracks have a place to play in helping them out.

Mike Delgado: Yeah, and what you just said, giving more power to the consumer, what’s going to be interesting to see is how are these financial companies going to react? Are they going to start treating or being more personable with their customers? Are they going to be quicker to respond by phone calls? It’ll be interesting to see how things change to better serve their customers.

Deborah O’Neill: Yeah. Me and some colleagues did some research earlier in the year about recorded digital families. In the UK, we did a big survey around people’s relationships with data, and technology in general. We asked them things about how long could they go without their smart phone? How did they feel about banks looking after their data? How did they feel about the government? How did they feel about all these different things? Did they feel technology was giving them an advantage in life, or were they feeling left behind? When we analyzed all this information, we got six groups, or digital families as we called them, and that’s quite an interesting consumer story as well, because it’s not just everyone wants personalized service.

Some people want to be private. Actually, they would be very upset if you started pushing them geolocation messages, because they’d be like, “Wait a minute. That’s none of your business where I am.” And some of the commentary around this were people saying, “I’m willing to be off the grid to protect my data.” So I think people’s digital strategies are going to have to get more nuanced as well, because I think everyone’s been going, “We must get more digital. We must get more online. We’ve got to give everyone customer personalization.” It depends who your customer base is. If you are a Mom and Dad organization who’s got a lot of history in just being safe, maybe your digital strategy needs to be much more toned down, and a little bit more, “It’s there if you want it, but we’re not going to push it at you,” whereas obviously, a bigger technology company or something that’s appealing to the young generation who are stuck to their phones all the time, would probably like you to push them deals. But you’re going to have to understand your mix quite carefully, to just not completely miss your customer base.

And it comes to diversity and teams. Let’s face it, the news has been full of lack of diversity on every angle that you can possibly be diverse on. But you see it, these disastrous product launches, where you can tell that they just didn’t have the diverse people in the team that were developing it, because they’ve just missed the angle. And we’ve seen it in adverts. We’ve seen it in product launches. Diversity in your team to reflect the diversity of the people you’re dealing with; I think is a important topic as well.

Mike Delgado: I think that’s really smart. I’m kind of curious with that study that you did, where you identified these different personas of digital families. Did you find that the differences were more about the generations. Were older generations were more fearful, or not approving of people being targeted, versus millennial and Gen Y?

Deborah O’Neill: I think any person would go in with that assumption.

Mike Delgado: Yeah.

Deborah O’Neill: There was a little bit of bias towards that, but actually it was much more across the demographic than I expected around age. And there was an interesting gender split as well. There were some people that are feeling like they are less advantaged if they’re young now, because technology is threatening them, and the skills that they’re required to have, they haven’t been given. We had a family that just were like, they just said that it was the left behinds basically. It wasn’t that they were fearful of data and technology. They just didn’t know how to get involved. And that group was a varied cross the age group. It wasn’t old. It wasn’t young. It was just a whole cross section who said, “Actually, I’m a bit nervous that my opportunities have diminished because of technology,” which I think is an interesting social conversation on many levels.

Mike Delgado: Deborah, it’s been a blast talking with you today. I want to thank you so much for hanging out with us. Before we go, do you want to share any kind of last minute tips for businesses that want to better utilize their data before leveraging machine learning?

Deborah O’Neill: Yeah. I think it’s quite simple. Make sure you’ve got a clear decision or problem you’re trying to solve. Understand what insights you need to answer that problem. And make sure you’ve got your manual processes cleaned up, so you’re not wasting time on the basics. Because once you’ve got that cleaner, your people will have more time to think and play around with the more interesting machine learning and AI insights. And if you use that, and you do some quick iteration cycles, test and fail, learn, move on, I think it’s a good place to start.

Mike Delgado: Wonderful. Deborah, where can everyone learn more about you, and find you online?

Deborah O’Neill: I have a Twitter account, deborahlabsow, and also on the Oliver Wyman Labs page, we’ve got some interesting insights. The digital family stuff is on there. You can take a quiz, and see which family you most associate with. But equally, reach out to me on Twitter. Happy to answer any questions.

Mike Delgado: Wonderful, and for those who are watching live, I’ll be having the links to Deborah’s Twitter account, as well as the links to Oliver Wyman labs in the about section of this YouTube video, also on Facebook. And Deborah, again, thank you so much for being our guest today, and looking forward to keeping in touch. And thank you so much for sharing your insights with our community.

About Deborah O’Neill

Deborah O’Neill is a Partner within Oliver Wyman Labs focusing on financial services.

During her time at Oliver Wyman, she has worked for a range of clients around the globe to deliver change and impact within retail and investment banks. Deborah’s particular interests lie in the areas of insight and reporting, risk management, and customer centricity.

Prior to her role in OW Labs, Deborah worked in Oliver Wyman’s Finance and Risk Practice, where she contributed to a number of high-profile programs with central banks in Europe and further afield.

Make sure to follow our Deborah O’Neill on Twitter and LinkedIn.

Check out our upcoming live video big data discussions.