Jonas Christensen 2:45
Felipe Flores, welcome to Leaders of Analytics. It is so good to have you on the show.
Felipe Flores 2:52
Right, thank you so much for having me on the show. I am so excited to get to spend this time with you. It's a real honour. Thank you so much.
Jonas Christensen 3:00
The honour is all mine. And you and I are still buzzing from a conference that you and your team put on last week. The first in-person conference that we've been to for years, but also the first such event that Data Futurology has organised. So we're going to get to that in a minute. We have a lot to cover today, including your story, your background, what Data Futurology is, and also what you do in your day-to-day. But you're very much an applied practitioner still and we want to hear all about that. And what you do at Honeysuckle Health.
Felipe Flores 3:31
Amazing.
Jonas Christensen 3:32
So Felipe, I just alluded to it. We're still buzzing from this in-person event that you organised with your team last week, and we had there some fantastic speakers from all around the world, top leaders in AI. You have gotten to that point now but you've come from a completely different point. Could you tell us a bit about your story through this journey of becoming ultimately a big league conference organiser in Australia from where you started in the copper mines of the Atacama Desert?
Felipe Flores 4:03
Yeah.
Jonas Christensen 4:04
And over to you.
Felipe Flores 4:06
Mate, thank you. Thank you so much. So I'm originally from Chile, from South America. I grew up, as you said, in the north part of Chile. I grew up in the driest desert in the world. So over there it would rain once every seven years. When it did rain, we'd get about one centimetre worth of rain. And the town that I lived in was right next to one of the largest copper mines in the world. And we were about five kilometres from the mine. So close that when they did the blasting at 5am and 5pm, you could feel the earth moving as a result of the blast. And that's where I grew up. I started going to uni in Chile and then I was about 19 when I came to Australia as backpacker. I thought I was gonna be here for like six weeks and that was like 20 years ago. So it's been like the longest six weeks of my life. I loved it. Decided to stay and then to - When I got to Australia, I didn't speak any English and when I decided to stay, I had to figure it out. And that meant doing every job under the sun. The first job that I got, I was a door knocker selling home phone and internet services for one of the major providers in Australia. And they gave me a script, so I had to knock on 120 doors a day. And I would like bring up my little piece of paper and be like ''Hello, how are you?'' and just like the worst accent ever. So as you can expect, I made zero sales. But what I did find was an amazing and super welcoming country. So I was couchsurfing with friends and colleagues or like colleagues who became friends. I was staying at their places. And through the door knocking work, I was getting invited to lunches and barbecues. And people like, have some morning tea, afternoon tea, that was like, really, really nice and got to travel around Queensland for that job. And then did kind of like heaps of heaps of odd jobs. I was an AV Technician, and etc. And then eventually, I got a job more in IT, where I started in a small company, more as like their web developer. But in my first couple of days, in the first week that I was there, I could see that they had a database that crashed about five times a day, and I could hear - the company was like 12 or 15 people - and I could hear people cursing everywhere going ''Huh this, you know, terrible piece of whatever''. And so I was like, ''Oh, that's interesting. That's failing a lot''. So I went to the library, like back in the day, and I was doing some research. And essentially, I found you could build a database on Microsoft Access, especially for a company that was that size. And I got this book that it was like ''Learn Access one hour a day, or 30 days'' or something like that. And so, I started going through this, and I was like ''I can do this''. Then I went to the owner of the company and I said ''Oh, you know, you got this terrible, crappy database that crashes all the time. I can build you a better one''. He goes ''What do you know, you're a kid''. He's like ''I pay 50 grand to this software development company to build me this one and it crashes all the time. You think you can do better?'' And I was like ''Well, try me. Like, let's find out''. So I was able to convince him. I build a database and then the reporting and then the analytics, and then the analytics allowed the small company to start making better decisions. So the company sold other businesses, like restaurants and hairdressers, and things like that. They were the broker for those. So we started identifying who had better success with different types of businesses when they were selling or who could target certain demographics, particularly people of different countries of origin. And then through that, we started shrinking the time that it took to make a sale. We started getting better conversion rates. The sales guys were really happy. And it started to snowball. So from that point on, I was like ''Shit, this data stuff is pretty good. Pretty good''. And through that, I started, you know, going to uni. So I did computer systems engineering and business. So I did a dual degree, which I found super beneficial to have the both sides. So I kept working in data from a degree. And then when I was finishing my degree, I did a thesis. And this was kind of like my first foray into data science. I was part of a research project where we were detecting the fatigue levels of truck drivers in mines. So it was a research project sponsored by the mining industry. And the way we did this was using EEG. There was three of us in the team. And there was an electrical engineer, very senior. And he created the sensors that went on inside of the forehead. And there were new sensors, because one of the constraints was that the whole thing had to fit within a baseball cap. So there was three sensors at the front. And then there was one close to the ear, behind the ear and that was the ground zero for comparison. And he made all the amplifiers and all the electrical engineering side and then gave me the signal. And then I designed the onboard computer that had to be on a flexible PCB to go around the inside of the hat. And then I had to find a way to solve the problem from going as an input having an EEG wave, and as an output having a number from one to five, where when people went from three to four, they couldn't drive anymore. And luckily, we were able to get thousands of hours of labelled data from sleep experts around the world, both the EEG and video footage. And they had told us when people were at different levels, and then that's how I discovered machine learning and ended up building a model to do that. And this was like early 2000s. So the time was completely different compared to what you can do now. But the onboard computer and Ryan ended up running a linear regression like crazy with obviously a potential transformation ahead of time to then be able to predict the level. But then much of compressed data was sent to a server where we had a neural net that was making the, kind of like, the final prediction. And then being fit back to the hat and the cabin, and also the organisation. So when people got to that level three, when they went from level three to level four, there was an alarms in the cabin. And that sort of meant that people had to go home, which was quite - the scene is quite invasive. So we had cases of really interesting change management of people taking off their hats and running them over with these, like 500 ton trucks. Yeah, and just saying, like ''Oh, you can't control me'' or whatever. But it was interesting that at the time, early 2000s, there was a mine, a very small mine in Chile, that had the first autonomous trucks and they were having teething issues. Like these trucks, they were a bit smaller back then, but they were about 200 tons and these trucks had been programmed essentially so well, that they were following the exact same path on these dirt roads that the trucks go on. And every truck was following exactly the same path with these 200 tons. So the trucks started sinking into the dirt because essentially, this has never been a problem before that when humans drive you go a little bit further to the left one side, a little bit further to the right, the next slide. So it was staying even for longer. So, we were having kind of like the change management and adoption challenges of this new technology, when humans drove the trucks and we were able to say ''Well look at this little mine that's solving the autonomous trucks and improving the teething issues''. So as part of the change management, obviously, that I wasn't intimately involved, I was kind of like a researcher on the team. But I could see that the decision was made to people to say ''Wear the hat, or you might be automated'', and you know, like change management and adoption in AI is still a problem that we're facing today. And it will continue to to be an area of us to focus on because that's how we get the impact of our work. So yeah, that's a little bit about how I got here, how I started in data, studied in AI. And it's almost 20 years later, and I love it. I love it.
Jonas Christensen 12:04
Yeah, well, that sounds like that's actually only half the story, because I'm sort of - you're 10 years into your 20 years there. But a couple of things I picked up on there was: I really liked that story of the trucks driving into the same track and actually digging a hole for themselves because that's really in essence, that is what this whole discussion around AI ethics is about. And if you swap trucks for people, you're actually keeping people in the same track and you got to be really cognizant of that. When you build models and perpetuate the past into the future that you are potentially perpetrating, either fairness or unfairness in there. Really good story. And Felipe, I'm sort of thinking throughout all this, there is some university mixed into it, but you're learning English as you go. Your English I haven't spotted any errors right now. Of course, I am also someone with a foreign accent but - I'm not an English teacher - but...
Felipe Flores 12:55
We gotta stick together, mate.
Jonas Christensen 12:56
There's no thick Spanish accent or Latino accent anymore. You've let it go. All of that you're doing at the same time, and you've gone to library and learning and so on. There's a lot of self taught knowledge throughout this whole process. What do you think is so unique about you here, right? Your career is these entrepreneurial endeavours, technical work, computer design, and so on. And you will hear in a bit that you've also done lots of senior leadership roles. It's rare to find someone who can straddle that full spectrum like that. What do you think is, in a nutshell, what you do that you do so well, and that you're better at than most?
Felipe Flores 13:32
Thanks, mate. That is - Well, first of all, that's very, very kind. And that's a tough question. So yeah, as you mentioned, like, during my period of almost like growing up professionally, in the analytics world, I spent 12 years in consulting and the last five of those I had my own consulting business, where in the past, it's kind of like in the first seven, I spent some time as a freelancer and then working for large corporates and small ones too. But then the last five, I had my own consulting company, and grew it and then eventually sold my part. From there I went to ANZ Bank and I was the first Head of Data Science at the bank. Spent about five years in finance all up, between ANZ and Liberty Financial, another company. And then moved on to healthcare, which now I work at Honeysuckle Health, which is for the last couple of years. So that's a bit of an overview. And in terms of - I can tell you the things that I enjoy and then some things around my personality that I guess predispose me to different things. I really enjoy helping people with building teams and helping people improve and rise in this space. One of the things that I think is particularly important that is what sometimes is referred as soft skills. I've also heard them referred as power skills, which I like a lot more. And that I think that in the data science and AI space, we have so many ways to acquire technical knowledge, and it's definitely a space that's moving super, super quickly. But we have so many options and ways to acquire technical knowledge, the programming, the cloud skills, new platforms, like understand the algorithms and implementations and even use cases. Like, there's so much content there available online. And then when we think about the leadership side of analytics and data science, it seems to me very sparse in comparison. And I think that that's the area where you can make a huge difference in your career as a person, an individual, but also you can have a huge impact in your work and for the organisations that you're involved in. And it's a value multiplier for the technology, this wonderful technology that we have access to. So I saw that as an area of gap and definitely decided to stay as technical as I can and double down on these like power skills. And then I am a person that's quite impatient by nature, and like my teens will tell you that as probably like the first thing. That has made me, help me find ways of shrinking the time to value and to the point that like, in my teens, for example, to implement this approach that when we start a new project, before we look at any data or anything like that, we sit with the stakeholders, and we get them to draw out by hand the charts that they expect to see. And essentially through that process, we want to uncover any hypothesis that they have, their expectations of how the world works. And then we have a subset, which literally usually ends up being three to five or two to five charts that then we draw by hand, and we explore and discuss and then we go and replicate those in the data. And then we discuss where the differences are. So to say ''These are the ways that you thought it was totally right and then these are the ways where it's a little bit different. So let's explore that''. And taking an approach like that, it means that you're delivering value in day one. If you have the data and a bit of the infrastructure, you can meet with them in the morning to get the hand drawn charts and then you meet with them in the afternoon, when you have, you know, three or two to five charts that are ready to go. So then that allows you to hone in into what's relevant for the stakeholder, what's important for the organisation and from the get-go answer new or bring in new insights, which in this case, are the differences between the expectation and reality and then you start following that part. So yeah, I think those are some of the things that I focus on, and that I want our industry and our community to leverage more and more over time.
Jonas Christensen 17:42
Wonderful, I love that. I think there's a natural tendency for people in this line of work to just take problems at face value and then go and try and solve them because we're natural problem solvers. But that doesn't always mean that you've got the exact problem phrase right. Then you're just basically wasting your time. I love that and I'm going to steal that and bring that back to my own organisation. I look forward to seeing people's drawing skills as well improve in the process. Now, Felipe during this period, you decide to start '''Data Futurology''. Can you tell us what that is and why you started?
Felipe Flores 18:19
Right. I love it. Thank you so much. So Data Futurology was essentially born out of this almost like gap in the market and working with so many data scientists and analytics practitioners over the years and seeing that people wanted more from their careers, wanted to increase their impact and wanted to have more impact and create more value in organisations. I could see that a lot of the gap was in the soft skills and power skills that we're talking about before. In 2018, I was going to have a bit of a break. My wife and I were gonna go - we got married at the start of 2018 - and we're gonna go on a long honeymoon. And I remember saying to my wife, something like - which obviously went down terribly - but I said something like ''Oh, you know, we're gonna have six months off. You know, I've never had a break this long''. She had. But I said ''I've never had a break this long. What am I gonna do? I'm gonna get so bored''. And she was like ''What the hell? What do you mean?'. But being nice about it, she sort of said ''Well, yeah''. Like, understands that I might have needed a project and she said ''You love listening to podcast so much. What are you doing yourself?''. And I was like ''Shit, maybe I should''. Like I hadn't really - the idea hadn't crossed my mind before that. And then I started thinking ''Well, what would I do? What am I passionate about? And what is the change that I think needs to happen in the industry? And based on the conversations that I had with professionals before like, what areas I felt like I could help them and things''. And that's how Data Futurology was born with the initial aim: The initial aim was to help analysts and data scientists get to C level executives. That was how I distilled it. Like what are they skills, the mindset, the approaches or strategies that people need to have to develop their career to get to a C level executive? And that doesn't necessarily have to be a chief data officer or chief data analytics officer because I think that the more people that come up in our industry with the power skills, they'll be able to move literally within a C level executive. So we've already seen people that were data scientists or data analysts by background that they become chief marketing officers, chief financial officers, chief operating officers, and often like CEOs as well. So I think that there's a whole world that is available for people that have our background, like the quant background, and that add these power skills and then are able to develop into professionals that can definitely lead organisations and business lines. And I really want to encourage the industry to think more broadly, like that. And that was the initial intent behind Data Futurology, which stays to today. So I'm four years in, and it's core to what we're trying to do. And during the journey, we've learned from the audience that there's a lot of people that are non technical, and that they see the value and power of AI and that they want to be really good stakeholders, and that they want to help bring this value to reality. We've increased our approach and our community. And we've become a bit more inclusive to say ''how can we help people from that background?'', like non technical people to better work with AI and AI teams, and help create the impact that is available from this technology. And those are kind of like the two major markets that we serve through the community. And it's been great. And as you said, like before, we've spent about four years doing virtual events, and we just had our first in-person event in Melbourne. And I was really happy with that. We had a few 100 people there. And I was trying to talk to as many people as I could, trying to get their feedback. And overall, the audience seemed to be really happy. The speakers did really well. They were excellent speakers and covered really interesting topics and approaches. We had really good sponsors as well. And I couldn't have been happier with the event. One of the things that was really important for me was the quality of the catering: to have like good food and enough food. Like nobody wants to go to these conferences where they bring out like three muffins for 300 people. And so like the fact that there was good amount and good quality food, on that side too was great. So yeah, super happy with how the Data Futurology is going so far. And when I first started it, I remember thinking, well, at the very beginning, I was like ''I'll do this while we're in a honeymoon''. So I'll do this for six months, and then call it. And then when we got back to Australia, or during those six months, we started getting sponsors, about month three. And that obviously helped. And then when we got back to Australia, there was like two weeks that I didn't release any episodes. And then I started getting messages and emails from people saying ''Hey, when is the next episode coming back?''. And I was like ''Shit, I'm stuck''. Like, I'll have to go back and keep doing that. At that point. I was like ''Well, maybe this is something that is valued or needed in the community. And if it is, then I'm happy to do it''. Like, I don't know. Ten years was the timeframe that I was thinking. And I'm happy to continue to give it a go, as long as people find value in it. And then, yeah, we'll see kind of like - we'll see what happens. And we've had tough times, particularly during COVID, where we lost all our sponsors and the company like- it's essentially a company becasue we charge sponsors, but it's been - the team always tells me that I should have incorporate it as a non-profit, because there's always been $0 left at the end of every year. And then like during COVID, we lost all our sponsors. And I was like ''Well, if this is the time to like scale it back heaps, I was like, I could keep doing a podcast or an interview maybe once a month. With the team, we can do a lot more. But if it's time to like really bring it down, then so be it''. But lucky enough, we were able to get sponsors again and then sort of growing over 2021. And then that brought us to our first big in-person event at the start of 2022 and we're looking forward to doing a couple more this year. So we'll be in Sydney in August and then we're planning an MLOps event in about November. And then we'll continue to do the podcasts and the videos and from there we'll see. Like as long as people enjoy it and we know what people are interested in, we'll keep covering that.
Jonas Christensen 24:38
Brilliant and I can really see how you got the ultimate customer feedback there, which is when you disappear, people start asking about you and where are you. So that couldn't be any better. The event was a really good event. It wasn't just the food. We had brilliant speakers from around the world and knock-on-wood Felipe, you and I we still don't have any COVID symptoms. So that also a good outcome. Hopefully, it remains like that. Do you have a long term vision for what Data Futurology will turn into?
Felipe Flores 25:09
That's a good question. I think that basically, I guess in summary: Whatever the community needs and is not receiving, we want to be able to provide that. So we definitely started on these like power skills and soft skills and helping people rise in their careers and have executive skills that are hard to even find places on how and where you acquire, especially when you combine that with AI. And I think in general, more broadly, what the industry and the community needs, and they feel like they're underserved. That's what we want to bring for them. So that can be, I don't know, books, or use cases. If it's conferences, if it's podcasts, if it's videos, I'm flexible from that perspective. And I'm keen to help people in their journey and to help them add value where they are, and kind of like be better at the things that they want to do. So if I can help a little bit on that, great. And now we have a wonderful team that obviously, they work incredibly hard to put this conference together. And I was telling everyone at the conference, I was like ''I'm terrible at this stuff''. I contributed embarrassingly little to all the amazing event that happened and it was all the team. So now that there's this engine that can help make things happen for the community, I'm keen for us to continue to find the areas that the community wants and needs, and then be able to provide those through this engine.
Jonas Christensen 26:42
Yeah, you and the team do provide so much value to the global data analytics community, and you really can feel that ethos of ''it's not about profiting or making money''. It's sort of a very giving organisation. So, I and everyone else who interacts with you really appreciate that. And it's also hard to predict what the needs are of the future, of course. Who knows, maybe in 10 years, you can make us all a Chief Data and Analytics officers in the metaverse or something like that. You know? You don't know what's going to happen there.
Felipe Flores 27:10
I'll take it. That'd be great. Thanks.
Jonas Christensen 27:13
Yeah, let's build that. But Felipe, we want to learn from you about all the wealth of knowledge that you've created on your journey, your personal journey, but also, you have interviewed hundreds and hundreds of people who have deep knowledge in this space. You are really a global thought leader, a practitioner, and a facilitator of all this data science knowledge for all levels of experience, in different types of organisations and industries. So, I think of you almost as a walking encyclopaedia when it comes to understanding what it really takes to succeed with data. Could you perhaps distil that down into what do you think the biggest trends in data science that we all need to watch out for and learn about and learn how to take advantage of in the next, say, one to three years?
Felipe Flores 28:01
Amazing. Thank you. Well, thank you so much. That is very, very kind, very kind of you. So thank you. There's a lot, obviously, that's happening in our industry and a lot that is required of us. And I think that the expectations on analytics professionals is always, kind of like, increasing. And so, I think that number one, I would say: understand your strengths. What you're good at and what the landscape requires. Because the landscape requires such diverse skills, that obviously they're not going to be in one person. So if you can understand what part of the diverse skillset, what part you are good at and passionate about, then double down on those in terms of your own skills, and understand what other supporting infrastructure you need in order to maximise the impact that we can get from these teams and this technology. So, in some cases, I've seen organisations that have an analytics leader, which could be a GM or a C or C level exec and sometimes higher. Sometimes the leader is very, very technical by nature. And as a result, they're - almost like their first level of people that report to them is people that are great at account management and sales, project management and things like that. So to say that, if a leader is very technical and they understand the analytics and the ML, then they might be weaker on some of these other areas that are important for an organisation to create the change and get value from the organisation, then do that. And the mirror image of that is organisations that are being - where the analytics capability is led by leaders that are not technical, that they are business people, that they can manage the complex relationships in organisations and they can open the right doors at the right time. So, the impact is realised. They often have people that are quite technical reporting to them as a result. So there's different models that can be implemented to maximise success in this space. And you'll see like heaps of different models in different places, and there's no kind of like - there's no one model. And that's because the model needs to be built based on the strengths and the weaknesses of the leader and the people that are there. So that's a great place to start. I think additionally, the area of adoption is something that we continue to find a challenge. As a challenge in our industry, that we want to be creating a lot of impact in organisations. The expectation from business continues to increase and they want to see more and more and better return on investment, but also a greater likelihood of probability from our projects and deliverables that we know that the overwhelming majority of AI projects and products fail. So from a business perspective, that's quite a risky investment. They want something more predictable, which is obviously out of our game. So focusing on the book ends of projects and products, and particularly - and sorry, what I mean, when I say book ends, I mean, the stock is around - as you were saying before and I was like understanding the business, understanding the problem, understanding context, the domain knowledge, and getting deeper, deep, deeper into that. And I think the specialisation on domain knowledge is something that I'm seeing more and more happening in data sciences. We continue to make sure that people are having domain knowledge and data science knowledge in the one brain, which makes them exponentially more and more valuable in organisations. And then at the other end of this sort of book end is the last mile as it's sometimes called, which is both the delivery of the AI product or service to the user of it. Which sometimes can be in the analogue space, but - the delivery of that analytical model or insight, and then the adoption and the usage of it, i.e people need to change practices, need to change the way that they work in order to get value from that. And that's a very, sort of, human element, where you need to understand the psychology and the mindset of people that need to make this change. That's an area where, you know, change managers and relationship managers are having a great welcoming to our industry when they come with those skills. And then if they learn AI, we can make them part of our ecosystem to help us improve those practices. But I have to tell you, one of the really good story, like interesting stories that we've had in the podcast - and obviously we've heard heaps - but in this space, I remember speaking with the people at Woodside that they have natural gas plants. And they did an IOT project. And they had about 20,000 sensors feeding data into the cloud. And then they did some optimization. And then they wanted to feed the recommendations to the operators of the plants to tell them how to improve the use of the plant. So essentially, how to get more out of the plant. And the way that they did it was excellent. So they took their time and they leveraged human curiosity. So in the first year, the operators were getting a message to say - it was personalised and it was comparing them versus them. So only comparing you to you. So it said ''Hi, Jonas, two weeks ago, you were doing the same shift. You were operating the plant at 2% higher efficiency. Would you like to know what you did?''. And then getting that curiosity and then people had to opt in to say ''Yes, I would love to know what I did two weeks ago, or a month ago''. And they did that for a long time and people got used to opting in. Then the second year, they did ''Hi, Jonas, people that have done this shift - and it was always anonymous - people that have done this shift were able to get 3% higher effectiveness from the plant. Would you like to know what they did?''. And then you opt in and you go ''Yes, definitely!''. And then once people, enough people opted in and they were comfortable, then they open it up to say ''Hey, here's how you can run this plant much better now''. And that was kind of like year three horizon. And I think taking that mindset of taking time to implement the change, it obviously would have started slower than people would ideally want. But the by end that you get from the workforce that needs to change from these technologies, you can increase the buy end drastically if you take a human approach and you leverage the psychology and in this case, a curiosity on giving people the same message enough times that they can't help but want to know how to increase that adoption. So I think cases like that are going to be continually important in our space. And then besides that, the other specialisation that we've seen is on the technical front. So, the mixture of engineering and data science, and the engineering could be data engineering or cloud engineering. But people can become more and more technical. They can help us build the platforms and the infrastructure that we need to create and deliver these models at scale in a repeatable way that's improving over time. That will continue to be a really hot area. So those are some of the ones that I see that are super valuable now and into the future.
Jonas Christensen 35:22
There's so much in that comment, Felipe. There's really a call out to everyone that's listening to this show: That you cannot just be technical, you need to have the depth and breadth around technical and soft skills. Again it comes up that I've said on this show many times that: to me, it's one of the hardest fields of work to be in at the moment because you need to have deep technical knowledge, typically. But you also need to be a marketing or a salesperson. Internally, you need to teach your stakeholders something that they don't know. You need to forge a path. You need to push your agenda upwards in an organisation, where they might not have had someone like you being a data science expert, or working in the data space at that level before. It's really hard. And it requires you to outperform, but it also requires you to get help from the organisation, C level, and especially your CEO, if you can. That is a huge game changer, if you can get them on board and make them help you along the journey.
Felipe Flores 36:18
Totally agree. And I think that's one of the areas where the fact that we are a new specialisation or a newish area for businesses, I think that that's that's where it hurts us a little bit. That a lot of businesses have established patterns on how other areas deliver value to the organisation. Other areas being IT or HR or marketing, that they have kind of like the supporting structures and engagement patterns of those areas with the rest of the business. And sometimes that a business partner, an engagement manager or project manager, whatever it may be, there's a lot of that supporting infrastructure that typically doesn't exist in analytics teams today. The majority of time, we're focusing on what would be the delivery teams, but without enough connections to the rest of the organisation to help the information flow and use the time of the specialists and in the most impactful way. And I think that that's one of the lessons that we're learning at the moment as an industry as well.
Jonas Christensen 37:21
Yeah, I often compare where we are now to where IT was. Probably sort of early 90s, where the IT professionals were the ones that were stuck in the proverbial basement, the nerds and those nerds got pulled out and elevated to CIOs when we started having PCs on mass, in businesses but also people started having at home. All that we are saying is that actually it takes that senior leadership. It takes for them to really interact and understand the new technology that you bring in, for the adoption to happen. So, you know, when we saw that personal computing push come in, in the 90s, all of a sudden, that space got elevated. When everyone started using internet and everyone became internet native, all of a sudden, we saw that explosion. And people who've grown up being either computer literate or internet literate, if I may call it that, when they end up in senior executive roles, then it really takes off. Picks up pace. The only exceptions I can think of - this is my analogy, so it's not proven that Felipe, but you can probably see where I'm going with it - you can also have someone who is very young, like Mark Zuckerberg, who is the CEO of something that pushes up some of these companies, elevates them much quicker than it would have been otherwise, with a 50 year old CEO. To be in the space in the data and analytics world, where we have a big selling job still to do to make the top echelon of your organisation really, really - they might appreciate what we do, but they don't really understand exactly what it is and the technical difficulty of it and how it works. But when we get to them, things really take off. So what is it that you can do today to make that happen quicker in your organisation? That your job. It's a big one.
Hi there, dear listener, I just want to quickly let you know that I have recently published a book with six other authors called ''Demystifying AI for the Enterprise, A Playbook for Digital Transformation''. If you'd like to learn more about the book, then head over to www.leadersofanalytics.com/ai. Now back to the show.
Okay, so they're the big trends that you see there. You had an example there, of Woodside petroleum. Do you have any other great examples from all your years of podcasting, where someone's really nailed this embedding of data driven decision-making in organisations that you would like to highlight?
Felipe Flores 39:43
Yeah, that's a great question. So definitely, I love that from a human perspective, there's been ones that have sort of knocked it out of the park from a business perspective. And one of the ones that comes to mind was from F1, that they focused on what they call fan analytics. So, special analytics for the fans and being able to do a lot of personalization on the communications and the merchandise, the offers for the fans, and they were able to grow that community. I think it was something like 4x the community in a very short period of time as a result of applying the lessons from analytics being, sort of, customer obsessed. And those areas, I always love. And something else that I love is: we've had people that have reduced the invisible barrier to value. And what I mean by that is that sometimes, as professionals, we are perfectionists and we want to build something that's excellent from an engineering perspective. That is the best that we can do: the gold plated version before it starts adding value. And we put that pressure on ourselves. And it's something that we can reduce, sometimes significantly, and lowering those barriers makes it a lot easier to get value, get impact, get started and to win support in the organisation. We've had cases where people were, like almost, doubling their sales through personalised offers through email marketing, right? And you go like ''Alright, fine, email marketing, like might be seen as like run of the mill''. And then in this case, the algorithm wasn't productionized from an MLOps perspective. It was something that was running on a laptop, on a desktop, where the data was being read from the warehouse and then written back from the warehouse and through the warehouse. Then the marketing system picked it up from there to do the personalised emails. So in terms of maturity, technical maturity, it could be seen as like quite low in the technical maturity space when you think about an MLOps lifecycle. But in terms of the value and the impact that it was getting, it was knocking it out of the park. So things like that I love where we are kind to ourselves. And we focus on starting with what we can, and improving from there, instead of shooting for the stars, where it just creates that extra pressure on us. And it takes a lot longer to get to that value. So, anything that brings value earlier in the piece, I definitely love them. And the last one that I'll quickly mention - because it like blew my mind - is we had the the CTO from NASA, from the Jet Propulsion Lab in NASA. From JPL, this guy called Chris Mattman and he was telling us how they built a deep learning model for the Mars Rovers and that it had to do online learning, because obviously the data transfers back to Earth had such low bandwidth. So they were so limited, that the algorithm had to be learning on the fly. When you compare that to, I guess, the barriers that we have for value in earth, we can get away with like offline learning and batching things. And in their case, they were going the full shebang. So you're in production in a Mars in a different planet, in a vehicle, online learning, self improving. I was like standing ovation. Like I was like ''That's amazing''.
Jonas Christensen 43:11
That is literally from a different planet, that use case. Okay, Felipe. So here's a fun question but it's a hard question. If we package all this up and put it into one place in one organisation, what would it look like? So really, the question is: if you were to design the perfect data driven organisation, what would it look like? And why?
Felipe Flores 43:31
Oh, great, great, great question. One thing, which we were discussing before we started recording, but one of the key pieces is the data as an input. And obviously, every organisation is going to be doing data collection and usually in the operational systems. Something that's key that I've seen turn around the quality of data that feeds the algorithms and the AI that organisations have is: By having the area that's responsible for entering the information - so, for capturing the information - have that area also responsible for the quality of the information. Which is like, obviously, like, that's easily said, but then it's kinda like, how do you do it? One of the ways that I've seen it work really well is for the analyst, almost to infiltrate that division, that business line and use the data that's collected within there to have better business results for that area. So, for that business line. And get the leaders hooked on the data for the decision making, and you have to start with the areas that have higher quality, and then expand to the ones that have lower quality, highlighting the fact that there's lower quality, but that things can be done better. And examples of that is things like call centres or using forms that people might get, where you want to be working with the business and getting them to align. The way to get started is find the KPIs that executives for those businesses are responsible for, and then come up with analytical projects or AI projects that can support those KPIs directly. And that they use the data from that area. Because for them, for the executive to get the result that they want, they need to have, obviously your help but additionally better quality data in that space. So, it'll be a direct mandate within the organisational structure that has the power to change that. As a double down additional benefit, you're gonna get better quality data for all of your endeavours that depend on data from that perspective. So that's something that I think is key. There's a lot of the technical components of having good data engineering, good data science, good analytics, being able to apply things at the right time. One of the shortcuts that I really like when you're doing a data science project or product is to go through stages of maturity of analytical maturity, that sometimes get used at an organisation level. Apply those to your project. So starting by looking at the historical information and say ''This is what happened''. Then you say like ''This is why it happened''. So, a bit of diagnostic. Then the next stage of maturity is the predictive: ''This is what's likely to happen''. And then the last stage, the prescriptive, is the recommendation. So it's kind of like ''What do I do now as a result of this?''. And sometimes the machine can close that gap. Sometimes is the human that needs to close that gap. And that's where I think as analytics professionals, we need to lean into it and say ''You know, like, the results of this experiment: this is good''. It means that we should scale and getting the organisation to be comfortable to take that leap. It's often like the responsibility of the analysts. Because, say if it's one of the first experiments that they've seen, they might not know what good or bad is, and what is a good enough signal to back and you got to be there providing that input. So I think doing that, obviously very helpful and helping on the adoption side, as we spoke about before. And then if you can do that on an industry level, or if you can provide analytical or AI services to an industry in kind of like a utility type model, I think is also really, really good. Because then, as an organisation, you'd be able to learn from many more examples. You get to help multiple organisations become data driven in their decision making. But on the flip side, you're not only keeping the learnings to yourself and multiple people can benefit from it. So I think, yeah, some of the ones that come to mind straight away.
Jonas Christensen 47:42
So many good points in there, Felipe. And I can definitely subscribe to the getting your stakeholders involved in data quality. As head of data science in a legal firm, a lot of our data is collected by frontline in a very contextual way. This is the challenge that we face every day. It's how do we capture the right building materials and then servise it up to the frontline, so that they can see what the data is useful on and what it's about.
Felipe Flores 48:06
I think that's the key, right? That the data is captured in a contextual way for a particular purpose. So that through analytics, you're expanding the purpose of why that data is captured, and hopefully increasing the quality as a result.
Jonas Christensen 48:19
The analogy I use for my stakeholders in the business there, I say ''The data science team: we bake the bread, to then mill the flour when they put it in the data warehouse. But the people actually planting the seeds and harvesting the grain: that is the frontline staff. You need to make sure you fertilise that data and make sure that it has the right quality. No weeds in there. No insects, no bugs. So Felipe, we're almost at the end of this episode. There's so much more we could cover. Maybe we could do a part two, if you're up for it.
Felipe Flores 48:50
Alright, definitely, mate. I would love to do a part two. This has been excellent.
Jonas Christensen 48:54
Okay, we will do that. Now, one question I will ask today that I won't ask next time in the Part Two is for you to pay it forward. So that's something we always do here on Leaders of Analytics. So Felipe, who would you like to see as the next guest on Leaders of Analytics and why?
Felipe Flores 49:10
Right, great, great, great. So, so many people, and I know that you've had like amazing, amazing people in your show from the get-go. So I'm like super, super impressed by that. Which obviously, it gives people less options to suggest because you're getting such high quality tests. So there's a few that I love. I'll give you one from industry and then one that's more of a consultant. You can pick. So one from industry. There's a guy, his name is Dr. Jazeck Kowalski. He's the chief data scientist at Australian unity. I find him amazing. Fascinating. He has a wealth of experience in heaps of different industries, different countries, and he's pragmatic and his teams and him are able to create really high quality analytics and AI and have had amazing impact. And discuss with them optimization of workforce across Australia with constraints like time windows and travel times and how they do that in near real time. So that's, that's amazing. And then more from a consulting area, there is Dr. Eugene Dubossarsky, that he is a bit of a legend in at least, at a minimum in Australian AI and somebody that I always learn, whenever I speak with him and he is, kind of like, a horse of the industry. So he does, obviously, the consulting and training, but also he has a few businesses on the side, but they're all very AI analytics driven. And they're always interesting to explore the applications and use cases that they're exploring,
Jonas Christensen 50:46
Felipe, brilliant recommendations. I will definitely be reaching out to Jazeck and Eugene. And last but not least, where can people find out more about you and get a hold of your content?
Felipe Flores 50:56
Right, thank you so much. So datafuturology.com is where we host everything. So, all the podcasts that we've done, the virtual events, public events, it's all there. And do also look me up on LinkedIn, and connect with me there. That would be excellent as well. So, thank you so much.
Jonas Christensen 51:16
Brilliant. It is not goodbye. It is ''See you soon''. We will make sure we do part two to this one, Felipe. So until then, keep doing what you're doing. And thanks for being part of the show today.
Felipe Flores 51:26
Right, a true honour to be on your show. Thank you so much for the opportunity and really looking forward to part two. Thanks so much, mate.
