Eric Kavanagh: Ladies and gentlemen, hello and welcome back once again to a very special edition of Hot Technologies. Folks, this is Eric Kavanagh, I’ll be your host for today’s show, “The CxO Playbook: The Future of Data and Analytics.” Yes, it’s a pretty big topic, I have to say. In fact, we’ve got a bit of a record-breaking crowd here today. We had over 540 people register for the webcast this morning. We’re doing it at a special time, as many of you know for our regular shows, we typically do these at 4:00 Eastern, but we wanted to accommodate the very special guest calling in from across the pond. Let me dive right in to the presentation today.
So this year is hot – it’s been a very tumultuous year in many ways, I think the cloud has a lot to do with that. The confluence of technologies that we’re witnessing in the marketplace is the main driver, and I’m taking of course about SMAC as they call it. We’re talking SMAC: social, mobile, analytics, cloud – and all that stuff comes together. Organizations can really change the way they do business. There are more channels for executing your business operations, there’s more data to be analyzed. It’s a really wild world out there and we’re going to be talking today about how things are changing in the C suite, so the chief executives, the top people in these organizations, well that whole world is changing right now and we’re going to talk about that.
There’s yours truly at the top. We have Jen Underwood from Impact Analytix and Nick Jewell, the lead technology evangelist from Alteryx on the line today. It’s very exciting stuff. I came up with this concept last night, folks, and I think it’s really kind of interesting. Of course, we all know musical chairs, the game for kids where you have all these chairs in a circle, you start the music, everyone starts walking around and one chair is pulled away; when the music stops everyone has to scramble to get a chair while one person loses out in their chair in that situation. It’s a very strange and compelling thing happening right now in the C suite, and if you notice in this image right here, you’ve got two empty chairs in the back. Typically, a chair disappears in musical chairs, and what we’re seeing these days, is there are two more chairs at the C level: the CAO and the CDO, chief analytics officer and chief data officer.
Both of them are taking off. Frankly the chief data officer is really taking off like wildfire these days, but what does that mean? It means something very significant. It means that the power of data and analytics is so significant that boardrooms, or executive rooms I should say, C suites are changing – they’re adding people into the C suite, whole new executives are filling out some of these new seats. If you think about how hard it is to change the culture of an organization, well that’s a pretty serious deal. Culture is a very hard thing to change, and typically positive change is fostered through good management and good ideas and that kind of thing. If you think about the opportunity that we have right now, by adding new executives in the C suite for analytics and for data, that’s a really big deal. It speaks to the opportunity for organizations to change trajectory, and let’s face it, the big, old companies really need to change because of how the marketplace is changing.
I usually give examples of Uber, for example, or Airbnb as organizations that have fundamentally disrupted whole industries, and that’s happening everywhere. What we’re going to talk about today is how your organization can adapt, how you folks out there can use this information, this insight, to change your business trajectory and to be successful in the information economy.
With that, I’m going to hand the keys of the WebEx over to Jen Underwood, and then Nick Jewell is going to chime in as well; he’s calling in from the U.K. Thanks to both of you, and Jen, with that, I’m going to hand it off to you. Take it away.
Jen Underwood: Thanks, Eric, sounds great. Good morning everyone. Today we’re going to talk about this CxO playbook; it’s the future of data and analytics and I’m going to dive right in. Eric already did a nice job of talking about why this is so important. Our speakers today, again, you’ve seen another slide with this information, but you’ll have myself and Nick Jewell conversing with you very interactively in this session today. We’re going to open up with describing what these roles are and the types of things that they’re on a mission to do. We’re going to look at the analytics industry, the outlook in general and some of the challenges these folks are going to be facing. The dynamics within organizations today as you’re preparing for the future, and then we’re going to talk about the next steps and give you guidance for planning, if you are going to be exploring some of these roles in your organization.
Talking about this CxO, the CAO for instance, that’s the chief analytics officer, that’s a job title for senior managers that are responsible for analysis of data within the organization. The CAO usually will report to a CEO and that fast-emerging position will be pivotal, when you think about the mass of transformation and its digital transformation that we’re having right now in the way that companies make and take their business decisions.
If you think about digital transformation and intelligence being that core of the digital transformation, this CAO is a very strategic role within an organization. They not only bring strong data science back to actual insights and that knowledge, but they own that resulting ROI and impact, so what are they measured on? How they’re bringing that ROI with the data that they have and some of the bottom-line numbers across an organization for leveraging data strategically. That position, along with the CIO, the chief information officer, has risen to prominence due to the rise in technology and digital transformation and the value of data.
For years now, data is gold in this particular world with monetization and intelligence and transforming this information. To be able to take these proactive actions and not just always be looking backwards, per se. The two positions are similar in that they both deal with information, but the CIO, per se, will focus on infrastructure where a CAO focuses on the infrastructure needed for the analyzing of the information. The similar position is the CDO and you do hear much more, we probably hear a bit more about CDO than you do about CAO today. The CDO focuses more on the data processing and the maintenance and those processes of governance throughout the whole life cycle of data management.
These folks are also going to be responsible for monetizing data and getting value from data and working across the maturity of the governance and security life cycles, across the whole span, I would say, of the life cycle. These are folks that would be very in tune, per se, or responsible for making sure GDPR – and we’ll talk about in a bit – the European Data Protection Act, making sure that those types of things are covered in their organizations. Now, we’re getting the structure and the future for disruptive dynamic data-intensive roles. These are the types of things the CDO will be responsible for and not just themselves – they’ll be building a cross-functional team, and I have some examples of some of the folks that would roll up to the, per se, in an organizational structure, from architects and governance folks, and even the analysts and the data scientists and engineers in an organization may roll up to them.
Moving further into the industry outlook for analytics, this has been a phenomenal – probably ten-year, even longer – ride in this particular industry. It’s been constantly growing, very exciting, even during the market crash years ago it was still in high demand. It’s just been a wonderful place and if you look at the CIO agenda from Gartner in 2017, BI and analytics is still within the top three rankings of what’s most important to an organization, and looking at the growth of software markets, we’re constantly seeing growth there. For as long as I’ve been in this space, it’s always been a really bright career.
When we look to this digital era and the transformation, what’s very, very interesting to me, is these processes that we have, and often it’s getting information and taking action from processes or during business processes. Now, Gartner has estimated by 2020, information that you’ve used will be reinvented, digitalized or even eliminated. Eighty percent of business processes and products that we had from ten years ago, and we’re starting to see that, right? We’re starting to see that with Amazon verses maybe some of the big box stores, the Ubers, the Airbnbs – these digital models are disrupting process and now folks are interacting. Even Black Friday – I don’t know how many folks really went to a store – a lot of folks are buying online, and how do you reach that customer? It takes intelligence to do that. It takes a very different way of interacting and personalizing the message and having that intelligence to present them the right offer at the right time, and now maybe it’s at a click of a button. It’s so easy for them to leave your online store. Things are really changing in this world, and I think Nick wanted to chat about this too.
Nick Jewell: Yeah, hi everybody, thanks very much. I’ll apologize in advance if there’s a slight delay on the audio coming in from London, I’ll do my best not to talk over you, Jen.
You’re absolutely right, that elimination of waste, that reinvention as part of the digital transformation, often comes about as organizations move from bespoke products, maybe disconnected applications and into more open and connected platforms. When your process is digital, it’s going to be a lot easier to see the end-to-end journey of your data. Really refine the steps you take, by using data to optimize that process.
Let’s move forward a slide, if we can. When it comes to digital transformation, what it means for organizations, I guess is either exciting or intimidating, depending on which side of the spectrum you’re sitting on. Take a look at the chart here, showing the lifespan of companies and how disruptive influences affect an organization’s fortunes. If you started a company in the 1920s, you have nearly 70 years on average, before another company disrupted you. A pretty easy life by today’s standards, because today, a company barely got 15 years until disruption threatens its existence. It’s predicted that around 40 percent of today’s Fortune 500 companies, so on the S&P 500, will no longer exist in 10 years’ time. By 2027, 75 percent of the S&P 500 is going to be replaced, so the half-life the organizations face today, before having to worry about disruption, is really shrinking. Successful companies need to stay ahead of that digital innovation race.
Today, no one really questions the analytics. It’s the centerpiece, that digital business transformation. In fact, organizations are putting digital innovation right at the head of their strategy. Those companies, they’re the top five most valuable companies in the world, representing two trillion dollars in market value, Jen.
Jen Underwood: Yeah, it’s amazing, it really is. It really is changing, and fast. The other dynamic that we have and we’ve been talking about this, now I think we’re finally seeing it and organizations are feeling this exponential growth of data sources, and it’s not even just analyzing data on structured data sources anymore. Again, we’re talking about, you only have a moment in some of these digital processes to make a decision and these things are coming in JSONs from REST APIs, we’re talking about unstructured data, whether the log files, there’s all sorts of different types of data, as well as the extreme constant growth.
Nick Jewell: Yeah, Jen, so as you pointed out, analytic leaders drowning in a sea of data. Getting to the high-value insight, maybe using a blend of existing or new analytic techniques, is really the end goal, but there’s a simple and fundamental problem that many organizations we work with, they’re really facing. We commissioned Harvard Business Review, we did the survey, talking to data analysts and business managers. They asked how many data sources they use in their organization to make a decision, and it’s pretty clear, there’s been a fundamental shift in just the last few years. IT used to blend data, push it to the data warehouse, but I guess despite all the excellent work that IT groups have done, creating centralized data management, analysts are still faced with task of creating that specific analytic data set, but they need to answer a business question. In fact, only 6 percent have got all of their data in one place, and the majority of analysts have to pull data from five or more sources – things like spreadsheets, cloud applications, social media and of course, not forgetting that data warehouse.
Now, most organizations recognize this, but what most organizations aren’t dealing with is the simple fact that data professionals are spending more of their time governing and searching for data, than they are in actually extracting value. These aren’t the high-profile strategic analytic problems the business execs want to hear about. But not addressing the fundamental issue is going to prevent organizations, really, from achieving value-driven insights. Jen?
Jen Underwood: That’s interesting. I’ve definitely seen different studies on this and it is this piece here, whether it’s the 80 percent of the time or trillions of dollars re-fixing the same data over and over again, very inefficiently in an organization. This does add up, these 37 and this 23 percent is very expensive waste of time. It’s amazing to me that more attention is not paid to that.
Looking at some of these, what I would call the market forces, and a lot of times when I talk about industry’s trends, I love following the industry and keeping a constant pulse on it. It’s important to understand when something’s more than a trend, when it’s really going to be a force that you need to pay attention to, and these are the top three right now, forces to pay attention to. It’s this rapid growth, number one is rapid growth of non-relational databases. I just mentioned this whole concept of not having much time in having to query, per se, a JSON, it’s these type of non-relational scenarios, that are growing quite – I think I have some stats in a moment here – rapidly.
The other thing is the ongoing shift to the cloud. Before the call I’d mentioned I was a worldwide product manager at one of the big tech firms and had difficult conversations three years ago with groups saying, “We will not put anything in the cloud. We will not move to the cloud.” And, it’s been very interesting seeing groups a year later, two years later, now I’m hearing from the same groups, that everyone has a cloud plan. I think everyone’s a very broad-brush statement extreme, but what I would say is, folks that have been anti-cloud, certainly the attitude has changed dramatically, within a very short period of time, even since I was talking to groups worldwide about these types of things.
Automation, this is an area that I’ve been fascinated with and an area that we’re certainly seeing a lot of activity and great activity. We talk about some of these things with having this wasted time and inefficient use of your time. Automation is certainly one of the areas that I’m most excited about when I think about bringing value to an organization.
The next slide I’m going to talk about, this is a study by IDC, they look at the market segments and the growth and it’s really a wonderful way to take a pulse at what’s really growing, what are your peers buying? What types of things are they not interested in anymore? Those types of things and putting into their strategy.
The worldwide big data analytic software market has, according to IDC, 16 segments and in that segment sense we’re looking at even some name changes. There was an addition of continuous analytic software, cognitive AI software platforms, search systems, so there were some new categories even added into here. This market overview encompasses pretty much the horizontal tools, prepackaged applications as well as some decision support and decision automating use cases. Again, this is going to be the types of solutions, when you think about CDO, putting in a context of a CDO, their portfolio that may be managing from data integration to analysis visualization, machine learning and all these types of capabilities that they need to have in the digital era.
The worldwide market itself for these types of solutions grew 8.5 percent in current currency terms and the overall market grew 9.8 percent according to IDC. This was compared to – you look at currency fluctuations over a couple-year period and the degree of variation is minimal, but those top three segments that I highlighted, just to give you a feeling for those non-relational analytic data sources, 58 percent year-over-year growth, content analysis and search systems were 15 percent and some of the customer relationship applications, CRM-type things or the Salesforce Einstein, for instance, those are growing over 10 percent, they’re 12 percent right now. I think Nick wanted to add some commentary as well on this one.
Nick Jewell: Thanks, Jen. It’s a fantastic visual. I think at Alteryx we’ve always believed that data preparation and blending would always be a core competency, I guess, of any analytic system, but it’s really the foundation for any more advanced analytics. Now, for the past few years, let’s talk about the industry – it might have been a bit over-focused on some of the new interactive visualization capabilities. They look beautiful because they increase engagement, they drive insight, but they didn’t really move us beyond descriptive analytics.
But, I guess now that folks are setting their sights a bit higher, organizations starting to understand the business values is going to come from those more sophisticated analytics that are just now making their way into the mainstream. The question there becomes, how, or more specifically, who? This jumped to higher-value analytics; it’s really throwing the issue of analytic talent shortage into pretty sharp relief, would you agree?
Jen Underwood: Absolutely, and I had, I think I just tweeted, I saw a really fascinating comment last night from the vice president of Adobe saying, “Machine learning has become table stakes,” where folks used to be wary, now it’s become a need and it is interesting. Looking at this and just a tiny other little different angle, per se. A lot of folks, we’re starting to see this as a high-growth area with a non-relational analytic store and the cognitive AI, these machine learning, these high-value analytics. But still at the end of the day, right now the largest segment, so where most purchases are happening today, is still in this basic, what I would say, the query reporting, some of the visual analysis, and it’s still growing and that’s something a lot folks assume you already have it – not necessarily. It’s still growing 6.6 percent each year.
As a CDO – and I love showing this slide – basically just to say, when you’re walking into this new role or you’re looking across data in an organization, it is chaos, and I think that this particular slide really does a nice job of – these are all the different potential areas that you may have data. They may be on-prem, it may reside in the cloud, it may be hybrid, it’s everywhere and it is a big overwhelming – again, it’s a C-level type role now within an organization, and it’s not a simple task or simple – in this particular world to take on, it’s quite overwhelming at times. This is the world that this CDO needs to navigate, to be able to master, what I would say, maximizing the value of data.
Continuing on the challenge, maximizing the value of all those different sources and what we’re having is these closing windows of time, with these digital processes or the insight to action is closing. If you think about maybe five years ago, ten years ago, it may be you’d have reports that you would run to make some decisions with inventory or actions, those might run weekly, monthly, then they became daily or overnight, maybe it’s hourly.
Now, what we’re seeing are these intelligent machine learning embedded artificial intelligent offices, making decisions and corrections on the spot, so even things like the internet of things, IoT-embedded analytics at the edge, these systems are smart and these algorithms can self-tune and alter some of the decisions that they’re making on the spot at the right time. It’s been very interesting to see this particular dynamic with the digital revolutions and these touch points – even though they’ve increased, the time to action keeps decreasing and the technology then is evolving for these scenarios.
Nick Jewell: Yeah, Jen, I think one of those most interesting aspects of how the delivery of insight is changing, is where the analytics arrives to the end user. Are we asking users to jump into a dashboard when they make a critical decision, or are we saying that the insight, the next best action, is available directly within the process, in the flow, in order to drive that competitive advantage? And the analytic model that we’re talking about might need to take its inputs from a wealth of different sources – traditional data warehouses, geolocations, social media, sensors, clickstream – all of this data has a bearing on the decision and that actionable outcome.
Jen Underwood: Continuing on this theme of challenge and change, what we have right now, and the challenges the CEO needs to embrace and plan a way to conquer these, is essentially we’ve got too much data to efficiently manage and manually analyze. There’s long delays; we need to shorten these delays and we need to find a way to maximize the value of the data that we have. There is a shortage of data science talent in the world and to cover these insights and what we would call oceans as data. The good news is, there are some wonderful innovations that are happening to help in every area of this today, and it is getting exciting on seeing what, where technology is going to take us, to help us with these challenges.
As I continued to look at this, there’s a bit of confusion as I talked to customers or I talked to groups using some of these tools. Some of the classic challenges still exist today, it’s just getting a little more exacerbated with trying to find data to analyze. Some of the search tools, some of the catalogs out there are certainly helping things – now what we’re finding is which catalog to use when. There’s a couple different catalogs, so there’s different places that you can store and share data, so it’s matter of trying to find out one, maybe the catalog we should be looking in.
The other thing is collaboratively sharing. We talked about one of the studies from that Harvard Business Review, how much time is spent, basically doing non-value-add tasks, wasting time and how expensive that can be. If you’re collaboratively able to share and use common data sources, the scripts have already been developed, the logic is already in there, you can govern them effectively, so balancing governance with analytics agility, that’s really what you want to strive to do and navigate this world of what I would call, we have the niche tools, we have automated workflow tools, we have classic Excel, the data catalogs, self-service BI, data science tools. As that one picture showed, there’s many, many tools and lots of overlaps between them.
Nick Jewell: Yeah, perfect, Jen, and I think the window of insight, as you mentioned, it’s most definitely shrinking, but the time it takes to actually deploy models isn’t keeping up. Predictive model deployment continues to be a major challenge for many companies. We’ve been talking to Carl Rexer who’s the President of Rexer Analytics, and in Carl’s 2017 data science survey, he found that only 13 percent of data scientists say their models always get deployed, and this deployment ratio is just not improving, so we go back with each previous survey. In fact, going back to 2009, when the question was first asked, and we see almost identical results, so we’ve got a real gap.
Jen Underwood: When we look at analytics maturity, it is rapidly progressing. Again, two, three years ago, we were very excited to have visual self-service analysis and finally being flexible and expanding BI to the masses, per se. When I say masses, probably still power users within an organization. Now we’re seeing optimization, predictive analytics, deep learning, natural language, many other technologies that really, as they’re embedded into everyday processes, will finally truly democratize analytics very seamlessly for the masses, for the true masses to use within the existing business processes that they already have.
Nick Jewell: Yeah, Jen, let’s talk a quick story around that last category, if I can. Most listeners on the call today are going to be familiar with Google DeepMind’s AlphaGo software, defeated some of the best Go players in the world over the last couple of years. AlphaGo learned to play the game by studying enormous volumes of previously recorded matches. So much so that the commentators of the AlphaGo tournament claimed that the software played in the style of a Japanese Grand Master, believe it or not.
But, over the last of month, an almost more astonishing result was reported. This was AlphaGo Zero, deep learning, neural network, armed with no more than the simple rules of the game and an optimized function. It taught itself to become the strongest Go player in the world, with no supervised training, and it did all this in around 40 days. This so-called reinforcement learning, where humans define the challenge, let the deep learning system explore, improve, really could yield the biggest impact in the analytic space yet. So, I guess, stay tuned.
Jen Underwood: Yeah, that’s really interesting you mentioned that. Can you imagine the exclusions? And this is what I’m starting to see. Really, when I talk about automation, very exciting for the solutions to be smart enough to clean air, to learn from systems automatically, plug and play and just know what to do next based on some of the past decisions that have been or other decisions that have been made within the organization and having managed some of these systems, the ETL systems and cared for them, and had way back in the day beepers and phones calling me with alerts when processes weren’t running, it’s so exciting to think, “Wow, now it’s smart enough to probably self-heal.”
My husband manages a self-healing grid, we’ll have self-healing data integration, self-healing analytics and where it gets better and better, it’s really exciting. As a CDO, when you start thinking about people process technology, we’re going to take a look at, right now we’re looking at technology, then we’re going to look at people and how to approach building your team and building the skills. If you look at modern analytics platform, I’ll tell you right off, not everybody is going to have everything on here, although the largest organizations might have all these different components, per se, some groups may only have two or three little boxes on here, so I didn’t want to overwhelm folks with this. But a modern BI platform does not require necessarily an IT build, predefined reporting semantic layer.
The users and experts should really just be empowered to just prepare data for analytical speed and agility, and if you think about the rise of what we would say user and expert-led analytics, letting the subject matter experts have the agility, they need to make quick decisions. We are seeing an increased adoption of what we would say, the personal data preparation tools, the data wrangling, the enrichment, the cleansing, the types of activities that Alteryx does as well as some of the data science-type activities that they offer as well. The modern preparation solution, they do offer that intelligent, automated joins, air resolutions, shifting of data, when you have big data pipeline it’s very, very cool. This is probably, again, one of the areas that I love and really enjoy testing as well in the industry.
Unlike the traditional IT-led BI, IT today is really focusing on enabling the business and you’re having folks like the CDOs and putting together or choosing the right solutions to orchestrate, organize and unify this data and make sure, of course, it’s governed, right? One thing that’s very interesting to me and certainly I think we’ve inferred to this, but I don’t think we’ve just straight out said it, the days of a one-size-fits-all data warehouse and that being the end-all be-all, are certainly over. Data is everywhere, you need to make – data lakes have come into the picture, there’s streaming and live data, there’s so many different data sources now, it’s really more of a use case-based, “What do you need?” verses this “We have to get everything into a data warehouse.” I’m not sure, Nick, did you want to comment on this one? I don’t recall.
Nick Jewell: I’ll just say one thing and it’s just, watch the evolution of the component. What experts did five to ten years ago, is now in the hands of the user, so the things on the right-hand side there, are going to be more prevalent for user in a drag-and-drop code-free form very, very shortly. It will move faster and faster, so just keep an eye on that.
Jen Underwood: Yeah, that’s really good point. I love thinking about that. The different data science, it’s finally becoming a reality and the tools are getting so much better. Thinking about technology, now we need to have the skills and the people and what do we need to do? Right now the best jobs, they include titles like data scientists, data engineer and business analysts, yet what we’re finding is that employers themselves find it really tough to make a match. Even in the data prep space, I’ll say, “Is it data prep, is it data wrangling, what terms do people call it?” It’s been very interesting to find.
The business doesn’t know what they need and there’s this whole new emerging field that will span many different areas. If you look at everybody now needs to be a master of their data, business analytics, IT project managers, my husband that’s managing a power grid and a portfolio of projects, he needs to be able to analyze this. It’s not just finance and the data analysis anymore, it’s really expanded much wider, to other areas of the organization. I think I saw a study about how many data sources marketing uses, and it was overwhelming. Again, when you think about the study that was done by Harvard Business Review, it’s not just one data source anymore that people have to mash and merge together and find an insight from, it’s many data sources and it takes skill to do that.
When you look at essentially the bigger picture here, most new hires will be in this pink bubble towards the bottom, when you talk about these business analysts to the data mining analysts, the HR managers, this area, just regular roles within in the line of business using data. The fastest growing roles will have less jobs, but certainly what we hear about the most in the market today, the data scientist and the data engineer. As a CDO, they’re looking ahead and you’re planning talent, you need to factor in some of the automation of routine tasks and the types of skills that will be more strategic, and again, add value with your organization, for both those in analytics enabled but also for the data science and data engineer folks there. Consider how your unposted positions and even some of the freelance economy might change when you think about that to compete for the best and brightest.
And, always be thinking about your talent pipeline as well, helping candidates navigate the market or looking for things that might be a little different and not exactly what you want and creating in-house analytics courses, that may not really be the fastest, most cost-effective strategy for you to keep up. Consider looking at folks that are dedicated to training on this or different groups, and I believe Alteryx has a recommended course at the end of the session today as a call to action, that you can leverage for some of these things and help your team leverage some of the existing resources that are already available.
Nick Jewell: Absolutely. There are so many ways of filling that talent gap without getting caught up in an arms race. Couple of slides back, I don’t know if you’re able to flip a couple there. Kaggle, the data science competition site, they just released a survey with 17,000 responses around the state of data science and there was a really interesting response from the survey around the skills that people had, and the majority of respondents didn’t have a PhD, it’s just not a prerequisite anymore.
The idea that the next-generation analytics experts, that major bubble you were just showing, they can gain the knowledge they need from nano-degree courses. They can go to sites like Udacity and they can deploy this knowledge immediately, directly in the business, short-focused delivery cycles makes them an immediate source of competitive advance for their companies. So something to watch out for, I think.
Jen Underwood: No, I agree. Even if I think about it, it’s certainly come a long way since I took a two-year program at UCSD. This was back in, I think, in 2009, 2010 timeframe and there were really maybe a handful in the country that allowed you to do that. There’s generally many more options now, as well as specialized programs, whether it’s through the vendors, lots of resources available today with loops and all these different online resources, it’s just amazing, it’s really the time. Making time and budgeting that and scheduling yourself to keep up. What is it that you want to learn? And then following that path that you want to learn.
Speaking about looking at this and putting together your own plan of skills and from a CDO’s prospective, making sure that they have folks in areas covered, from what I would say a competency framework per se, looking at skills or looking at things like domain knowledge is still really key, even though these solutions can self-train and self-learn, it is really a business subject matter expert that will guide and make sure that the results make sense.
There’s always something and I like to use the example of when I was doing critical analytics for an insurance company and one of the findings that the algorithms had was not to hire anyone from New York. Well, no, we’re not going to not hire anyone from New York – we had to find out why was the algorithm giving us this information. It was because the legal, one of the laws had changed and so we were having a lot of churn in that particular segment. A business subject matter expert needed to be brought in to decipher that, and I don’t see that changing, I don’t see that kind of guiding it, making sure that the results look accurate, does something look off – it is still, there’s something said to be the human mind, the beauty of that combined with the power of machine, is really where we are going.
The other types of things when you’re looking at skills, visualization, telling an effective story in the data, telling an effective story to whether it’s even machine learning output. Putting together and looking at what is the impact it makes, understanding the human nature of decision making, those types of things are very important regardless of technology. Governance is really important, ethics is becoming more and more important. Having social scientists involved, that understand and they’re trained to look at if there’s biases in your data that you don’t even realize or don’t have anyone in the organization that might not even recognize that, even bringing them into the expert, having those types of things.
And again, of course having the infrastructure for engineering and the hardware and making sure you can scale and it’s developed and making sure you are using the right cloud provider, maybe that you’re not locked in or that you have options to move or that you understand the pricing on how much these are going to cost you. It’s these types of skills and when you look at this, we would call it skills by different areas, whether it’s data-driven frontline decision makers – where most of these roles will be – all the way to those data engineers and data scientists that will be massaging and working in these oceans of data. These are the types of things that you’ll want to put together a framework for.
Looking at competency frameworks, you look at an organization in general, you want to consider competency, not just skills. There’s a little nuance there in the wording as you’re looking at this. A competency framework for your organization is a clear signal. War policy makers, education providers, while skills would be say, typed under R, you think about those types of things, you have a competent coder, but you want to want to have more than just those skills. When you understand competency, what a person must be able to and understand the framework, that’s the important, there’s a little bit of a nuance there.
As you are building this, you want to diagnose what you would call capacities that do have a positive impact on the business and highlight those high-potential areas, so you’re prioritizing what are the competencies that you want to elevate in your organization and then align those again, with the business objectives. The CDO that is responsible for maximizing the value of data, they’ll look at, and their CAO, that’s going to use analytics to maximize the value of data. They’ll look at those competencies and those different areas, on the past grid that I had there, but then they’re also going to look at high potential of staff. You’re going to cross-reference that with your staff for data and analytics work and invest in them, provide them learning opportunities and not just training, essentially real-world opportunities working on real business problems.
There’s nothing better – even though I went to school for a couple years, it wasn’t until I went and applied some of these algorithms or learned about check fraud, learned about some of these things I’d never thought about before, and you start putting together in the real world and that is where you really learn. Giving people that opportunity to gain the experience in these areas. The companies that are best able to build strong capabilities, that systematically identify, objective assessments and looking at where are the gaps within my organization for learning and putting some metrics in place for goals for folks, those are the ones that are going to be able to deliver.
When you think about training adults, again, usually it’s a time starved – we’re all time starved – but looking at what works for each. I personally have books, so if you were to come into my office today, you would see tons of books, even though lots of folks like videos. So it’s a matter of finding out, how does someone in your organization like to learn – to motivate them to learn – but also providing them some time to do that and goal of some sort of – what is an effective to reach that and usually that’s blended, it’s not just, take that course to check that mark off on a score card, per se, it’s blending that with real goal project and what did you learn from that project and what do you want to do next? What’s a stretch? Stretching your team or motivating your team to take it further.
Those learning objectives, again, if you’re doing that, it shouldn’t really be, it should be easy for the business essentially because those objectives should align with the strategic business interests. These are great projects. They’re experimental projects. They’re projects that will move the needle forward.
Nick, did you want to add anything? I’m not sure.
Nick Jewell: No, I was going to jump into a case study, if that’s OK, on the next screen. A little bit more detail of a specific organization. I guess they’ve put a lot of what you’re saying into practice, into reality. The Ford Motor Company relied on data analysis for decades, just like many companies, but it did so in pockets of the business, with probably very little oversight across the entire corporation to ensure consistency and coordination. Their problems were probably fairly typical for an organization of their scale, so analytics expertise contained – as we say – within pockets, data management and governance practices being inconsistent, even to the point where some business units lacked access to basic analytics expertise.
Again, we’ve talked today about lots of different types of data sources, they had over 4,600 data sources. That meant even starting the journey and finding the data that they needed was a real impediment to analytical insight. I see you’re laughing, but it’s a terrible thing, right?
Jen Underwood: 4,600, oh my gosh, yeah.
Nick Jewell: So, Ford formed the global insights and analytics unit and this was centralized – you can call it a center of excellence – consisting of team of data scientists and analysts, organized to share that analytical best practice and help spread optimized data-driven data making across the business. The unit selected the best-in-class tools, not only on capability but also on their ability to integrate well together, so that’s quite important. The focus of their democratization was actually around reports and descriptive analytics, before moving up that pyramid of needs that we’ve talked about.
Now, democratization doesn’t just make somebody a data scientist overnight; staff need to know when and where to get help, and there’s training, governance, methodologies available to help with all of this. Also, it’s not just about tool training, but also data science training, to bridge that skills gap that we’ve mentioned. So, a real-world use case at Ford, optimizing a logistics network, so was Ford paying the right amount to move materials from point A to point B? Their legacy analytics really didn’t highlight actionable opportunities; this made them very reactionary in the market. Now, a lot of complexity for that process was locked away inside the analysts’ heads and they made a huge breakthrough when the self-service workflow was actually iterated with the business, and the analytic experts sitting down together and being co-located.
This moved the analysis from multiyear to quarterly, and even down to near-real-time, so huge, huge benefit to the business. That impact of self-service analytics on business value, there’s been that Ford can quickly plan and establish corporate-wide data-driven strategies, to respond to emerging trends, help shape new services, and basically head off threats from the competition, without just having to look in that rearview mirror.
Now, if we take a look for a moment at how another customer has really moved analytics from maybe a vertical priority in a single division of the firm to being a horizontal stripe across all divisions, we’ll talk about Shell. Shell runs a center of excellence that reports into the chief digital officer – so there’s another D for our CxO playbook – responsible for digital transformation and sustainability. These guys, they understood that their environment contained several layers and the technology stack, storage, data processing and it all featured technologies that you’ll all be familiar with. Things like SAP HANA, Databricks, Spark, and they leveraged public cloud to reach those right economies of scale.
Now, they selected Alteryx as an analytics wrapper for a lot of their R code, feeding into technologies like Spotfire, Power BI and more. But now they see the adoption tying much more closely with data processing and visualization. Jen, just calling back to your slide of all those capabilities, this kind of thing spreads as we start to enable more analysts to have access. You know, they were hugely successful in delivering this capability and the COE, looking to deliver future capabilities now, some of those deep learning things we talked about – machine vision, natural language processing – and half of their mission is delivery, half of it is about explaining and catalyzing these ideas across business units. It’s part of the journey; the COE is always looking out to different ways to communicate with their business audience.
Taking into account on one side the skeptics who say, “Well, this black box will never be as good as my analyst,” all the way through to the fanboy or the enthusiast who sees correlations everywhere, maybe less in the way of causal relationships, but you need to be careful on both sides. It’s a fascinating middle ground, when you have this horizontal stripe across an entire organization, that hybrid skill set that’s needed to persuade both sides of the spectrum.
Nick Jewell: OK, Jen, are you there?
Jen Underwood: I am.
Nick Jewell: I guess what we’re trying to say here with this Clayton Christensen quote is that for many organizations, I guess, unifying the analytics agenda in order to drive the digital transformation that we’ve been talking about today, is going to be a challenge. More often than not, we find analytic teams starting with a weak hand. Attempting to innovate with legacy holdovers of analytic processes, technologies, team structures and holding on to these relics is going to be biggest barrier for analytic alignment and for analytic innovation. Do you have any thoughts on that, Jen?
Nick Jewell: Absolutely. So, if we move onto the next slide, we think there is a better way. I guess first of all, using something that’s akin to Google-like search, to quickly find all of your data assets that are most relevant. Understanding their context, understanding dependency, factoring in really simple things like business glossaries authored by experts in your communities, kept alive by all of that tribal knowledge of the heads of your coworkers.
Getting smart with data discovery. Think about the ability to hold conversations with report owners and experts. Uploading, do a little bit of Trip Advisor or Yelp, uploading the assets that are most useful, certifying those that the organization thinks is most valuable and then all of this feeding back into the search results and ultimately the search rankings, making it better for the next user. Once you find what you’re looking for, moving into that rapid, code-free, user friendly, preparation and analysis phase to develop your perfect data set, from which to publish repeatable processes.
Back to our automation conversation, building up user-friendly apps. Whatever is needed to build analytical models. Speaking of models, we’ve supported open-source technologies such as R for a number of years, allows us to build up a really advanced analytic capability that covers descriptive, but also predictive, prescriptive analytics, in a simple, drag-and-drop way.
Now, over to the right-hand side, actually getting that insight out into interactive visualizations, models and scoring being pushed down inside data platforms, or most recently, making that insight available instantly and directly within a business process. I think it’s this range of capabilities across the whole platform that’s allowed us to be recognized as the Gold Award winner in this year’s Gartner Peer Insights Customer Choice Survey, which is a fantastic accomplishment. I strongly recommend you visit the Gartner site to find out more and add your own votes and add your own commentary.
Cool, so, Jen, if we skip forward one more slide – I guess as we conclude, I’d like to give you all some next steps. First of all, please visit Alteryx.com to download a complimentary copy of our most recent research brief, done in coordination with the International Institute of Analytics (IIA), around breaking down analytic obstacles. You can also visit udacity.com/alteryx to learn more about how to enable your teams, to take the next step in their journey, with that advanced analytics nano-degree and then finally experience Alteryx for yourself. Visit the homepage, download a fully featured evaluation and get onboard with the thrill of solving.
Jen, over to you. We might have some time for some Q&A.
Eric Kavanagh: I’ll just chime in really quick. We do have a couple of questions. I’ll throw one, I guess, over to you first, Nick, and then Jen, if you want to comment on it, but it certainly has more applicability to the EU and that is the infamous GDPR, the Global Data Protection Regulations. How is that affecting Alteryx and your roadmap and what you guys are focused on?
Nick Jewell: It’s very much a boogieman, I guess, that’s out there right now. A lot of people talking about it, a lot of people quite worried, but it’s really just the first in a long series of regulations that are going to come into the data and analytics world. Really, from our point of view, it’s about understanding and classifying your data. Making sure as a CxO, of any particular flavor, you know where your assets are, you know their context and you know you can trust them as a first step to really just governing and managing data in a wider context.
Eric Kavanagh: I guess I’ll throw another question over to you before we bring Jen back in, Nick, and that is, the training data, if someone requests that their data be removed from your enterprise, that effects not just their name, address and so forth, not just their contact information, but also, if an algorithm uses training data that includes your data, you’re supposed to retrain the algorithm, isn’t that right?
Nick Jewell: It’s particularly complex. I think that the idea that not only databases as being a source of some of this personally identifiable information, but also the analytic workflows, the apps, the visualizations. This data gets everywhere with an organization, so having that context: absolutely vital.
Eric Kavanagh: And Jen, what’s your thought? Obviously, it’s not that big of a deal in the U.S. and we don’t see too many companies fretting over it now, even though technically it does apply here. If a U.S. company has data of an EU citizen, what’s your take on significance of GDPR and how big a deal it is?
Jen Underwood: Well, I certainly think it requires responsible treatment of data. I’ve written about this a few times and have some guidelines on some of these things. I think the question that you asked about algorithms is interesting. Certainly, some of the solutions that I’m looking at today, some of their product teams have designed features so that you can see how they’re making the decisions and what personal data was used to decide the outcome of that algorithm. We are seeing some impacts in the product designs here in the United States.
A lot of the technology companies have very large offices here and development teams here in the States as well as worldwide, so we’re seeing it on the product development. I am seeing more data catalogs being invested in. More governments’ initiatives being spun up so that folks understand, and they understand where all that data is in the chaos. Trying to get their arms around at least organizing it, being able to find it and do something with it.
Eric Kavanagh: I’m going to push this slide that we talked about earlier and throw this over to you, Nick. I think this is a fantastic slide because, to me, it really speaks to the immediacy of a need for analytics. What do you think about this changing dynamic? I mean, the bottom line is that companies must be agile and I see analytics as leading that charge. What do you think?
Nick Jewell: This is fascinating. I think there’s always – companies and technologies always exist in three states, so it’s either going to be war, peace or wonder. The war is going to be about that heavy level of competition. Wonder is all the great new stuff you build on top of a platform. Then peace before the competition and the war starts again. I think there’s always this battle going on.
Before today’s call, we talked about some of the other conference and key notes that are going on around the world today. Some of the big cloud vendors, they’ve reached a point where they’ve built up this platform and now they’re building wonderful new things on top of it. Companies have to keep a really close eye on this and make sure they’re going with something that has a coherent platform that will deliver that value for the future. They’re going to be the ones that are going to survive this disruption.
Eric Kavanagh: Yeah, that’s a good point, and you know, Jen, you commented earlier, in fact before the show, about cloud strategy and how a lot of the folks you know in the industry are saying that big companies, even banks, all now have a cloud strategy. I’ve been kind of surprised at how long it has taken for that to materialize, and I guess maybe some of them went to the AWS Reinvent Conference and realized how massive it is and drew the conclusion that the time has come. What do you think about the awareness among big business executives about the import of cloud and how that’s changing their planning?
Jen Underwood: When I think about this world of massive-scale data, being able to manage it, I think on some levels there’s some peace of mind with having one of the very large firms take responsibility for some of the security aspects, so there’s some peace of mind there. You know there’s some limited scale with cloud.
The other thing is, and I saw it, I was on a team that redeveloped a product in the cloud and it was certainly an underdog product and no one payed any attention to it, and within two years, because of weekly releases and even, I would say, it’s almost to the point of daily release in cloud. I know that Amazon says that they release multiple times per day. When you have that threat, when your competitors can release and improve daily, whatever it is that they’re doing, at least in the software industry – and everybody’s really in the software industry when you start looking at digital transformation – it’s a whole other ballgame and anybody can spin up a cloud and scale and become large.
Again, it’s going to be the data that they’re leveraging that’s going to make the difference and the intelligence in their algorithms, and that’s why folks are talking about data being the new oil or data being gold. When I look at cloud, it’s the game changer, it really enables very, very rapid development and scale. It’s amazing.
Eric Kavanagh: I’ll bring you back in, Nick, for another question – we’ll go just a minute long here if we can to get to some of these questions, but, as I recall, five and six and maybe even seven years ago, Alteryx was really an innovator in leveraging third-party data – so bringing in data from sources like Experian, for example, or geospatial data. I’m thinking that’s probably a strategic advantage because that kind of thing is in the DNA at Alteryx, right? As companies move toward cloud, I think you guys have a lot of experience in being able to bridge those worlds. The worlds of on-prem verses third-party and cloud-based data, what do you think?
Nick Jewell: Yeah, absolutely. Ultimate connectivity is going to be such a power play to any company that’s going to be working in this cloud-based environment. But I will say, when we talk about something like infonomics, the idea that information and data should be considered an asset in your company. Most of the value that you’re going to be bringing in is taking external data sources, blending them and enriching them with your internal sources, to create and monetize more value in the process. It’s absolutely critical to work with internal and external data equally.
Eric Kavanagh: Yeah, that’s a good point. I think this whole world of hybrid cloud is here to stay. Jen, I’m just going to throw this over to you for some closing comments, perhaps. To me, having that strategic view and being able to unify as the new term is describing data across the sources, that’s going to be a critical success factor going forward, right?
Jen Underwood: No, absolutely, and it’s funny, I was hearing this hybrid, hybrid, hybrid. You heard about this and four years ago you think about Hadoop, Hadoop and big data and then you started to hear hybrid, hybrid, so certainly been there, we’re not necessarily, this is the year of machine learning, bar none. I mean, artificial intelligence, the machine learning has taken the stage this year, but in order to really function in an organization today that’s on the way to the cloud or that has to deal with all these different cloud data sources, maybe it’s Salesforce or Workday, all these different types of sources that live in the cloud, the only way you can handle it is to be hybrid. You can’t possibly copy data everywhere, so you do need to be able to directly connect and you do need to find a way to work with data located everywhere, find data everywhere, because that is the reality of where we’re at right now.
Eric Kavanagh: I think I’d be remiss if I didn’t bring back machine learning into the conversation, so, Nick, I’ll just throw it over to you. I know that you folks are focused on that now – can you kind of talk about where you see machine learning aligning with analytics and with the kind of systems that we use to understand our business and our data?
Nick Jewell: Yeah, sure. So, very briefly, then, let’s just quickly go back to our skills gap. The idea that we’ve got organizations absolutely chock-full with power Excel users. We’ve got data scientists coming through, but not growing at the same rate. There is a massive gap between the two. Think about where machine learning is today. How many algorithms do we have on our phone or our watch that incorporate machine learning techniques? It’s a commodity, it’s everywhere. We need to enable these power users in the simplest possible way to make sure machine is applied successfully across the business.
Eric Kavanagh: I’ll throw one last one over to you, perhaps. We got a couple questions coming in late, here. Jen, I’ll ask you this one. An attendee is commenting on this whole concept of unsupervised learning and the fact is you do need training data to do that stuff and typically that training data needs to be specific to the company. Even though in industries there are lots of correlations, there are lots of way in which organizations are similar. Nonetheless, every company is unique, whether that be its business model or its approach to marketing or sales, or whatever the case may be, product development.
The question becomes, will these algorithms be able to use third-party data for training? It seems to me you’re always going to need to use your own data to train these algorithms, even if that cycle time collapses from six months – which has been the case in some cases – down to 40 days or 20 days, whatever the case may be. You really have to use your own data and you have to make sure that data is pretty clean, right?
Jen Underwood: It’s really a blend. You’re going to want to have external context. In fact, I’m booked today back to back and my next webinar is talking about preparing and cleansing data, ironically for machine learning. What’s really key is you are putting together external context with your organization, and I love that you asked about the data prep and cleansing, because honestly, some of the tools are getting very, very good – they can handle some aspects of it, but the human mind, or being able to decipher the problem and look and make sure that they haven’t omitted – say that we have some kind of omission bias. The way that you’re looking at the problem and the way you’re choosing to design the problem that you’re automating or decisions that you’re automating, there’s an art to that and making sure that it accurately reflects that business process.
Going back to my example with the insurance company, when we were modeling churn and who to hire to go through this sponsored training to sell insurance; in the model itself wasn’t the legal climate, different laws for different states. There’s always going to be some aspect, where you’re going to have to have that external data with your internal data and, again, the human mind. There’s going to be different components there.
Eric Kavanagh: I think you brought up a really good point here. We keep hearing about robots and machines and machine learning taking over. To me, this is a very disruptive trend – there’s no doubt about it – but I don’t ever see the need for human beings in the mix going away, especially with analytics on data, on enterprise data.
Nick, one final question for you. To me, no matter how good the algorithms get, you’re always going to need people monitoring what’s happening, injecting themselves at the appointed times and really synthesizing the big picture of what’s out there. I don’t think any algorithm is ever going to be able to synthesize the big picture for a Fortune 2000 company, but what do you think?
Nick Jewell: Well, let’s take a completely non-Alteryx example, let’s talk about Uber from last year. Uber, during the terrorist incident over in Australia, people trying to flee from the area, they suddenly put on surge pricing, ‘cause that’s what the algorithm said to do, caused huge reputational damage. Immediately after that, they implemented humans and algorithms working together. Anytime this was about to happen, a human had to have oversight of the process. That partnership of human and algorithm, that’s the way forward.
Eric Kavanagh: Wow, that’s a great example, thank you so much. Well, folks, we’ve burned through more than an hour here on our webcast. Very big thanks to Jen Underwood of Impact Analytics. Of course big thanks to Nick Jewell and the Alteryx Team for their time and attention and to all of you for your time and attention. We appreciate these great questions. We do archive all these webcasts for later viewing, feel free to share them with your friends and colleagues. With that, we’ll bid you farewell. Excellent webcast today. Thank you so much again, we’ll catch up to you next time, folks. Take care. Bye bye.