 Welcome. My name is Shannon Kemp and I'm the Chief Digital Manager at DataVercity. We'd like to thank you for joining the latest installment of the Monthly DataVercity Webinar series, Advanced Analytics with William McKnight, sponsored today by Looker. Today William will be discussing predictive versus prescriptive analytics. Due to just a couple of points to get us started, due to the large number of people that attend these sessions, he will be muted during the webinar. For questions, we will be collecting them via the Q&A in the bottom right-hand corner of your screen, or if you'd like to tweet, we encourage you to share highlights or questions via Twitter using hashtag ADVanalytics. If you'd like to chat with us or with each other, we certainly encourage you to do so. Just click the chat icon in the bottom middle of your screen for that feature. If you'd like to continue the conversation after the webinar, you can follow William and each other at community.databercity.net. As always, we will send a follow-up email within two business days containing links to the slides, the recording of the session, and additional information requested throughout the webinar. Now, let me turn it over to Elena for a word from our sponsor, Looker. And Elena, hello and welcome. Thank you so much for having me today, Shannon. I will just pop open my screen and we can get going. So thank you, everybody, again for joining us today. My name is Elena Rowell and I am on the marketing team at Looker. Digital transformation is a profound shift in the way that companies use technology to deliver value to their customers. It is without a doubt the number one priority for technology leaders today. Out of all the areas that can drive digital transformation, the number one area of investment is data and analytics. We're not really talking about the conventional understanding of data and analytics. According to Gartner, spending and focus on traditional query and reporting BI environments will shift dramatically to analytics that are integrated into business operations and digital solutions in the coming years. And with this, we're talking about a real fundamental change in the way we're used to thinking about data and working with data. Forster calls the companies that are doing this well, insight driven companies. And these are the companies that are using data to drive growth and create differentiating experiences, products, and services. The idea here is that data is not just something that we analyze. It's actually something that can be infused into everyday business processes to make employees smarter, to make optimizing operations easier, and even to drive new streams of revenue. We're talking really about the idea of expanding the idea of traditional BI and bringing it into that modern era. Drive growth, create these actually data driven experiences. Our mission at Looker has always been to empower people through the smarter use of data. Over 2,000 companies partner with Looker today to harness the power of their data and to empower their people to drive successful business outcomes and deliver better experiences for their customer. And we've seen that these data experiences are not just through analytics. Sites delivered to people where they are already and when they need them. So 50% of our customers are serving data to people outside of our tool. And really what this means is that it's finding them where they're already working and arguably more important, where they're already making decisions. To do this, a modern approach to data and analytics requires a modern technology foundation and Looker was designed to take advantage of evolving trends and data infrastructure. So the technology powering Looker is based on three main pillars. So obviously here on the screen is our in database architecture. And this really leverages the power of modern cloud MPP databases. I'm sure as you all know, are significantly more powerful, way faster and far cheaper than their predecessors. This live connection to the database provides a complete view of your data without having to move any of it to an application layer. So instead, we query the database in real time. So the data is always accurate and always fresh. It's this idea of ELT, not ETL. A semantic modeling layer is a centralized definition for all of your business rules and business logic separates the business logic from the physical data, which allows you to reliably apply consistent definition across KPIs. So both technical and non-technical people can work with trusted and reliable data. And this is all made possible and even made easy by our APIs for data that give us the tools they need to deliver any data experience anywhere. So be this custom apps, active rule based alerts, scheduled reports, automated workflows. It's really the beauty of the API is that it allows everybody to be creative. All these capabilities are built on a flexible multi-cloud architecture. We believe organizations need to modernize at their own pace. We have a strategic commitment to a flexible and multi-cloud approach because we believe it's critical for companies to choose the environments that work best for them. But then that's both for today and in the future. I've mentioned is data driven experiences. Let's dive a little bit more into what that means. Data driven experiences can represent any number of ways in which companies drive value from data. It could be internal value by increasing revenues or optimizing costs or external economic value by monetizing data and generating new revenue streams. Bottom line is that Looker's platform offers a unified surface to access the truest most up-to-date version of your company's data. With this unified view into the business, you can choose or design the experience that makes the most sense for what you need. We power a number of data driven experiences. First one is a modern BI experience. Looker's world-class BI experience puts trusted actionable insights into the hands of decision makers and helps to cultivate a data culture throughout the organization. BI is just one of the data driven experiences. Ultimately ends up being a small slice of a company's data strategy. We see companies working with data in incredibly innovative ways. Leading it into every part of their business in ways that go far beyond reports and dashboards. These companies are really taking their data and starting to think what other parts of their business could be improved having accurate, liable, fresh data coming from the platform throughout their business. Finally, we see leading organizations deliver data driven experiences by building data products. To truly capture the immense value of digital transformation and get ahead of competitive markets, companies are even taking the next step and that's monetizing their data and even creating new revenue streams from the wealth of data that they're already sitting on. There's a number of case studies up here and they're really all just perfect examples of how leading companies are delivering data experiences and shifting their focus from these traditional reporting tools to platforms that integrate data into business operations and data products and it can really both the predictive and the prescriptive side analytics. Today it all comes back to bring value. So a simple core thing to connect with our customers is so powerful because that's at the end of the day what it is and that's what digital transformation is about. It's about driving better conversations and driving more growth through data. I'm going to pass it back to Shannon who is going to take over. Elena, thank you so much for kicking us off and thanks to Looker for helping make these webinars happen. Really appreciated and if you have any questions for Elena, feel free to submit them in the Q&A section in the bottom right hand corner of your screen as she will be joining us in the Q&A portion at the end of the webinar. Now let me introduce to you our speaker for this series, William McKnight. William is the president of McKnight Consulting Group. He takes corporate information and turns it into a bottom line producing asset. He's worked with major companies worldwide, 15 of the global 2000 and many others. When McKnight Consulting Group focuses on delivering business value and solving business problems, utilizing proven streamlined approaches in information management. His teams have won several best practice competitions for their implementations and he has been helping companies adopt big data solutions. And with that, I will go to Florida William to get today's webinar started. Hello and welcome. Hello and thank you, Shannon and thank you, Elena for the great lead in. Those were some very eye opening statistics and it's just really great to be sponsored today by a company that really embodies what I'm about to talk about, which is this whole idea of prescriptive analytics. I'm really excited about bringing you this topic because I think everything else really sort of builds up to this topic. Everything else is about establishing that foundation, that data foundation, that architecture foundation, the governance foundation, etc. for your analytics as an organization. And I'm referring to the many webinars that we've given here in this series one per month for this whole year and hard to believe, but we're getting to the end of the year. Two more left after this one in this year, but we are renewed for next year, so keep on coming back. We're going to keep talking about data architecture, keep talking about the data tools and so on to help you with your analytics environment. And one other thing I hope to see some of you next week in Chicago at the Data Architecture Summit. I know I'll see you there, Shannon. Hope to see some others of you there. It's bound to be a great opportunity to talk in more detail about these things. So, let's jump on in here and slide. Okay, that's a little bit about me. I think I've been introduced, so I'll move on. Yes, we do strategy. We do training and we do implementation. We should need any of that. But anyway, analytics is moving. Let's talk about analytics. Now, I'm going to give you a lot of examples here today because I think that the examples speak louder than definitions and things like that. We'll have some of that too, but analytics is kind of, in my walk anyway, it's been kind of a nebulous thing where people say they're doing it. Sometimes they're just doing reporting, in my view. They're not quite doing complex analysis to the level that we need to be doing it at. Okay, I think it's kind of a gray scale between your basic reporting and your analytics. So, I'm going to be pushing higher into that scale, into the analytics side of things today, which I think you need to be there. Now, I don't think everybody's there. I think quite a few companies are not utilizing analytics to the degree that they need to, I'd say most. And you might take some comfort in this, if you're one of them, but I wouldn't because to survive as an organization, you really need to be pulling in analytics. You need to be capturing analytics and utilizing them. So, for example, at Wellpoint, they are using analytics to optimize patient outcomes and transforming the entire industry. They'll knock on effect of what they're doing with ensuring that patients get the right care and get to the right providers for what they have so they have a healthier population. That's just one example. As Elena mentioned in her statistics here, some of mine, but this is moving from business initiative to business imperative. More and more companies are saying that analytics will create a competitive advantage for me. And indeed, those who are achieving with analytics are achieving are more likely to substantially outperform their industry peers. So, what is this thing? People are self-selecting their way into studies like this. You always have to take them with a grain of salt, but it does indicate that things are moving forward in analytics and we all must be doing that as well. It's not anymore about your everyday low prices, your great supply chain, your great customer service and all that. That's all expected. It's really not about utilizing your core data assets anymore in terms of the basic reporting. Maybe you're required reporting. It's about something deeper. I'm going to get into that here today. So, first of all, let's start at the lower end of things. What's the difference between business intelligence and predictive analytics? Some people would call that lower tier of analytics usage just descriptive analytics. So, I'm okay with that too. Descriptive analytics or business intelligence. You're looking in the rear view mirror. It's showing you what happened, which, okay, that's great. That's nice that happened. Maybe we have to report on it. Okay, check that box. Great. That sort of thing is expected today. Yes, so many organizations are still having a hard time with that or kind of stuck there. And indeed, I think you have to be at least okay at that before you can move on into analytics as an organization. And the first thing you would move into is predictive analytics. I mean, really, when you're getting into analytics, predicting what's going to happen, predicting it in real time and utilizing, and I'll say this over and over again, summaries of data, summaries of data that are actionable, because you can't take terabytes of data and make a real time decision out of it. I don't care how fast the computer is today. You have to be proactive with your data. If you've heard this series, me talking about the various AI and the data foundation and so on this year, you've heard me say the data needs to be screaming out what to do with it. And that means we have to take a proactive hand in the data that we put into our leverageable platforms for our users to consume and our systems to consume today. The technology is different for predictive analytics. Elena mentioned that there's an earlier legacy generation of tools, and then there's the modern set of tools that are really embracing this idea of analytics. Kind of hard to get from one base to the other, so you want to choose your tools wisely. And the questions you're asking are very different. I'm going to get into some of those. Analytics, at least the analytic data part, not necessarily the analytical applications of this stuff, but analytics are formed from summaries of data. Summaries that become actionable. So different things like customer segmentation. What, what, uh, what decile is a customer in or what is their, uh, designation, you know, as a, as a B to C consumer customer, are they a senior citizen on a fixed income? Are they a young urban, blah, blah, blah, et cetera, et cetera. What are they will let the data speak? Let the data speak. The data will tell you, but you have to, you have to be constantly as the fourth bullet indicates reevaluating the data and determining what it's saying about these customers. And what's their profit level? Well, that changes with every purchase, doesn't it? So, and, and their predicted future would change with every purchase as well. So we have to be constantly looking at the data. Analytics should be tied to business actions because it's great to understand what may be about to happen, but if you can't change it, it's really not very good at all. It's just a step towards actual business value. So let me put it on the table right now. There's really no debate between, you know, uh, predictive analytics and prescriptive analytics. They're just levels of maturity with analytics and, uh, we need to be getting to the prescriptive kind. Okay. So if it came here thinking that I'd be kind of debating that maybe predictive is good enough. No, it's not, it's not good enough. It's just a step. And, and believe me, that's a big art form right there to be good at predictive. And it's the foundation for prescriptive. But let's bring the prescriptive, uh, element into whatever we do and bring the big data. And I'll show you how that adds to, uh, the, uh, possibilities as we go along here. So it's all about the future. It's not about the past. The past is done and do anything about it. It's about the future. It's about having trusted knowledge of an accurate future. It's undoubtedly the most useful knowledge to have. If you knew what the future was going to bring, you could get in front of that future and change it to what you want. That is certainly the most predictive, uh, and prescriptive, uh, antelake you could have. So that is what we want. The future is one that you would want to intervene into and tune who your preference. Okay. Here's my definition. Those of you looking for a definition. Analytics is the deep systemic examination of a company's information. So you can ask yourself, is the basic reporting that you're doing and what we all have to do that, right? But is that deep? Is that systemic? You know, usually, uh, there's, there's some level of that that's not in an organization. Don't let that be the only thing you do. The non deep, the non systemic, we've got to get into things that embrace those terms. Analytics are key to predicting the future. And again, I'll say being able to predict that future is what we are striving for with analytics. Here are some examples, some data examples. All right. These are some of that, you know, quote unquote summary of data that I keep talking about that you need. You need to summarize this stuff and make it available. Um, if you followed along, uh, the, the, the, uh, webinars here, you know, I might say at this point that a master data management hub would be great for you or some of this stuff. But wherever you're putting it, make sure it's leverageable. It's not one and done. It's not for a departmental, um, very finite type of application because to do this well is a lot of work and I'm not going to read through all these. You can see what some of them are and some of them would apply to you. Some of them would not, you know, maybe you're not a B to C. So you don't care about age and gender. Okay. I get that. But these, these, these concepts apply to you no matter what. And here's the thing about, about these analytics is you don't want to wait until you're until your hair, everybody's hair is on fire and you're in the middle of an application development cycle. You're getting towards the end and oh, we need to know the average balance of customers by geography. Okay. You know, and we just whip that up. Um, yeah, you probably could, um, but it's going to take time and it's not going to be something that the corporation has bought into. Maybe it's not being done by the right people. Maybe it's not being put in a leverageable place. So these are things you want to understand up front. Understand that a lot of these analytics for the applications you're doing today are really the long poles in the tech. Get ahead of that. Okay. So big, big data, I mentioned you need to bring into your analytics. Adding that to analytics gives you some big value. So we go up exponentially in terms of the possibilities. Again, I'm not trying to give you examples here that are way out there that most people, most companies are not going to be able to aspire to in the next year. These are things I think that you all should be aspiring to at the least some of the things that you see here. These are very accessible things. Netflix with its personalized recommendations based on history, progressive with its custom auto premiums based on actual driving habits via sensors and so on. You see examples here from insurance. Plenty in healthcare as well as plenty in manufacturing. Like for example, Vestas optimizing the sit siting of wind turbines by mining larger volumes of data. So optimizing things in our business. If we're not optimizing these things, our competitors surely are and that will make or break us as companies. So we do want to be sure that we're optimizing, not just winging it. And let me drill in here. This is one I had the opportunity to be a part of a commodity purchasing application example, streaming data solution. So this company was always deciding whether to take up offers to purchase supplies from vendors based upon promotions that were just constantly going on or not. So there's a lot of factors involved here and they were winging it and they were getting suboptimal decisions. And by bringing more analytics to bear and turning more of that, at least those initial triage of it all over to the computer is something that was very ideal for this company. So they have to look at things like the quality and condition of the commodity. What are they buying? Minimization of risk. Supply and demand. Is this going to sit on shelves or is this going to move? Is this needed by our customers? Are we doing things as an organization that will require more usage of this particular supply or not? What's going on in the business? Storage costs and availability, especially if you're like storing up pharmaceuticals, for example, for eventual sale. Well, you know, there's a lot of temperature requirements for this, space requirements for this and so on. And that's true for a lot of things, right? So you have to take that into consideration because you have a finite set of that. And then the impact of weather on the supply chain, how will that change things? Will that make us increase the air conditioning? Will that make us, you know, have to move the product a lot quicker and so on? So there's a lot of factors involved. When that's true, whoops, we need analytics. And here's one from a children's hospital trying to predict baby crashing, which you can imagine that's pretty important. So a lot of factors that go into that and deciding when to act and how to act upon the conditions that you're presented with is quite important to this hospital. So this is out of my book, Information Management. And this is going to try to highlight for you the, from me to you, the difference between predictive and prescriptive. So in figure two, as it's labeled here, the, now we're trying to decide upon churn. Now, if a customer has a high probability to churn and they have a high customer lifetime value, we might want to get in front of that situation and do something about it. But if a customer has a low probability to churn and a low customer lifetime value, we don't necessarily want to do something about it. See, it's a kind of a problem if you are saying, well, this customer has a, you know, they might churn, you know, I don't know if it's 1% or 20% that they're going to churn in the next month, but they might churn. And there's this one thing that we do to prevent churn. We send them an email with a promotion or blah, blah, blah. That is not really taking advantage of analytics here. So I have a very basic model. I show you where there may be something you do at a high degree, which might be in this case, email them and give them a call. Right, there might be something that you might want to do this lower intervention, like just send them an email, maybe with a promotion, you know, this can get pretty, pretty complicated. And I want it to get complicated. I want it to get well beyond my capabilities as a, you know, data professional. I want the marketing team and the product team involved here working with me or taking the ball and actually figuring out what to do with the data. But on my point here is that there's a lot of no intervention. You don't have to act on everything that you see in the data. Now what you might want to do is, for example, if there's greater than blah, blah, blah probability to purchase and they're a top three desktop customer, they're a top customer. We don't want to lose them. We want to do all the actions that we possibly can. So for example, a telecom might offer a free phone, might offer a free month, might offer a free phone. Free additional family member, et cetera. So whatever it is, it's the max you can do to keep that customer. That's what you do there. But at a lower tier of probability to purchase and a top five decimal customer, you might just send them an email and say, hey, how's it going? We care about you, blah, blah, blah. And, you know, do something to try to keep them involved, but not something that's so expensive is sending them a phone necessarily. But if they're unlikely to purchase and they're a low-decile customer, they're not moving forward with us and, yeah, they're probably going to churn, well, there may not be anything that we really want to do about that. So you see, that's the difference between predictive and prescriptive. Prescriptive is you're doing something about the prediction. So how much can predictive analytics truly predict? Well, this is a very difficult question. I get asked this question and hopefully it's in the context of understanding the client's situation and the data that they have available. But I kind of go to the data on this one. The more data you have that's in a ready state, that's, you know, it's in a leverageable platform, it's trusted, it's well-performing, et cetera, the better your predictions are going to be. Because, let's face it, you're not going to spend forever trying to come to a prediction. You're going to spend a window of time, 10 minutes, 20 minutes, whatever, 30 minutes. But it's a window and you're going to do the best you can with it. So if your data infrastructure is great, yeah, you're halfway there when you start thinking about it. But if it's not, oh boy, I got to go to that data warehouse slash outhouse and pull data off into Excel spreadsheets and do this and that and oh, the clock is ticking and let me just throw something out there. It's not going to be very accurate. And the accuracy is very important. But there's no guarantee of 100% correct prediction. Just move forward. Keep moving forward with this. Don't say, well, I'm not perfect, so I can't move forward. After all, Las Vegas is built on 51%, as I like to say. So let's get into some of the topics around prescriptive analytics. Real time, yes. Artificial intelligence, how that adds to it. The data architecture foundation for it. And something very important in this is self-service, getting intermediate parties out of the way. Okay, the time value curve. This is a slide from Richard Hackathon. I pull out every now and again because it just demonstrates the various latencies that there are. And when we see this, we can see that, oh yeah, I might be heavy on this latency, but not on that one. The capture might be the thing that's really holding us back. But you've got capture latency. You've got analysis latency and you've got decision latency. But hopefully they're really small and you're able to actually act while the customer's standing there in the checkout line, while the offer is being made because they're going to back up if you don't act on the offer right now or while the patient is laying in the hospital bed. You have to act in real time. I always encourage my clients to set up a real-time architecture. And sometimes frequently, I will hear back, well, we don't need that. This is batch stuff. Well, today it is, but it's going to be harder to get there down the road. So let's build it in. Again, one of the common themes that you'll hear me say is what do I say? I say it doesn't take more effort and budget to get it right. It just takes the knowledge and the focus. And that's what I hope to be giving you just a little bit of in this series. AI enhances analytics. Artificial intelligence is key to predicting the future. You can take the value of it up exponentially where you can intervene into that future. I'll give you some examples in a minute. This is a deeper level of analytics, more than what a human could sit there and do. Self-service data discovery, intelligent recommendation of new data. Another thing artificial intelligence can do is suggest and identify data that can be brought into the analysis. Bustering that data up, like photo tagging it, can help the analysis go a long way. For example, we have some leaders in AI, as I call them, across the bottom here doing some of the things that you see here. Now, improving financial fraud, that's at an all-time low. Of course, it's a spy versus spy game, but it is at an all-time low, thanks to the algorithms that are in place. We've all interfaced for better, for worse, I might add, with chatbots, but that is the way of the future. And we can do some things to make those chatbots even better. We're all familiar with smart cars and we're all familiar with driverless cars, which they're taking in a ton of information and having to make real-time decisions and they're using algorithms and probability and so on, and it's just something to behold. Reducing the cost of handling misplaced items, automating, blah, blah, blah, automating anything that you possibly can. I have some clients that will ask about how to get started with artificial intelligence. Well, if we don't know, one of the things to do is look at what can you automate. And there's probably a lot, if you haven't looked at things that way. And then predictions, maintenance. Predictive maintenance is a huge AI function, combine that with IoT maybe to fill out the architecture for it, but that's huge. So these are some AI-based analytics in action. Hopefully there's something in here that's accessible to you. Now, I hate to keep saying that I keep saying things, but I do keep saying get your data under management in a leverageable platform, in an appropriate platform for the data. And this is key to a lot of my consulting is getting the right platforms, getting the right data into the right platforms, okay? And it depends on the usage and how leverageable it needs to be and we want to make it as leverageable as possible. So that means data warehouses, lakes, master data management hubs, etc. Used effectively by multiple business groups, not one, but by multiple business groups. Make sure you're leveraging what you're doing and you're not just hairs on fire and trying to do what it takes to get through the day. It doesn't take much more effort to make sure that it goes a long way. Get your high NFRs, non-functional requirements, okay? Availability, performance scalability, stability, durability, and security, I guess. Make sure they're in place. Get your data captured at a granular level. You're going to need it. The more data, the merrier, the better for artificial intelligence. The more data you can trust, just any data, hopefully it's out of standard where you can actually utilize it in predicting the future with the algorithms. And make sure your data is at a data quality standard as defined by data governance. Now, all of these things taken together, that's the definition of data under management. But you have to ask yourself, how much of my data, my enterprise data, is under management versus not under management? And the more, the better that is under management. Nobody's 100%. Okay, let's be clear on that. But I think grade A performers or level five maturity clients have at least 80% of their data under management. So they have a very pronounced data warehouse there but pronounced data lake and a master data management hub and maybe an operational data hub which has become quite interesting lately. Now, people love to hear about this, so I wanted to include it. This is the analytic workflow of data. And data is going to come from source systems, so-called source systems, right? They run the business, they're pretty important. But we ETL that stuff or ELT, as Elena pointed out, into our analytic database, which if it's leverageable, it might be our data warehouse or one of our data warehouses, as the case may be. We are now putting semi-structured data into a so-called data lake, mostly in cloud storage. This is where we want to, quote unquote, capture all data. In our analytic database, we're capturing, that's more of a relational database, okay? Capturing analytic structured data. And I'm not showing you the myriad of data marks that are probably proliferated throughout this nice clean picture. But anyway, the lake and the data warehouse are sharing data. Each is creating its own analytical data that's being shared to the other. And the data lake, ideally for me, it's actually a staging area for the data warehouse. So it's actually pushing a lot of data, a lot of that what I call multi-use data into the data warehouse, where it can be queried by the user out here, or if the user is a data scientist, they're exploring the data and mining that data out of the data lake. And anybody might be actually federating queries with data virtualization across both or more. The data doesn't have to all reside in one structure for you to access it in one query. Okay, so some other things that are important here as you make that move to prescriptive analytics. Self-service, just intermediating the intermediaries, okay? Making the tools easy to use, making the results easy to consume and enhance, making it easy to access data, and making solutions fast to deploy and easy to manage. Okay, does that sound like your data access environment? That's what it should sound like. And modern tools are going to go a long way towards this. And some of the legacy tools maybe weren't built for all of this and maybe won't. So the tooling is very important when it comes down to it. So what about those tools? What tools are fit for self-service analytics? They work with SQL and no SQL, because you're going to have that eventually, if not now. They provide that data virtualization that I talked about. They accept the results of and participate in data governance. They're not walled off from data governance, okay? Data governance influences the use of these tools. They provide secure data access. They adhere to security policy. They provide collaboration functions, which I haven't really mentioned much of, but analytics to be effective really, I found that if the analysts and the data scientists and the users can collaborate over the results they're seeing in the various data stores, that can help out quite a bit. And they can be up and running quickly and can pivot with agility. Those are some of my definitions for a tool that's fit for self-service analytics. Be sure yours is. Okay, an analytics manager. I'm going to give you this as a mid-level maturity with analytics. So maybe you've had a data scientist on staff less than six months. I've already lost half of you, I know. I am putting some aspirational goals out here for you, all right? If I didn't do that, I wouldn't be doing much. I want to push you ahead with your data. I want to push you ahead with your analytics. So I got to give you some challenge goals. Some of you are okay with this slide already. You're like, yeah, we do that. Okay, hang on. Got one more slide for the high-level maturity coming up. This is mid-level maturity. You got to get here. If you're not here, I think your business viability dictates you must get at least to this level within the next year. And really, I like to say that you got to get to my high-level maturity by the end of next year if you really want to be a standout competitor in your field. Anyway, so much about that. Let's get into the details. So we got concerted efforts to plan the analytics that will benefit the company. Okay. And this must be, obviously, this must be embraced at a high level. In the architecture space, there's a basic understanding of it. The data architecture is satisfying some demand, but it's still imperfect and misunderstood. There's some black boxness to the models where we don't quite know what they're doing. They may be harboring bias. We can't necessarily repeat and get the same results. So models have some prediction bias. They're dependent on other models, which is a bad practice. Annihilated systems with mixed signals make improvement cumbersome and so on. So there's the good news here, and I'm dwelling on the negative on the slide, but the good news is you have models. Or you've become a model creating and sharing organization. And then you have analytic processes. But unfortunately, they're still at a level of amateurish development where the systems are not developed by analytic professionals and unintended consequences result. Okay, but it's a maturity level. You've got to go through it. So I'm not digging it. It's just something you have to go through. Hopefully you go through it quickly though and get to the next level. No plateaus are comfortable for long. So if you get here, you have to keep moving forward. And ethics, you're not considering ethics yet at peril, I might add, but we want to get to here. Where you have multiple data scientists, they're up to speed quickly. You have a great analytics architecture. We have a central catalog tracking all the models. You're modeling. It's hard to make manual errors. It's kind of error prone. It's got a great SDLC around it or what we call ML Ops around it. Analytics processes are tight. It's easy to, for example, specify a configuration as a small change from a previous configuration. Okay, your scoring runs on a pretty periodic basis. You know that it's not one and done. When it comes to analytics, the scoring should be continually run on data. And you get that and you're doing that here. And ethics are a part of your analytics at this point. There are good faith attempts to remove the bias from the models. And there's potential for malicious use of analytics considered in the analytics lifecycle. I'm not saying you've stomped it all out, but you are considering it and you're trying to plug your holes. Maybe you have a few left, but you're plugging your holes in the ethics. You're plugging your holes in terms of bias that might creep into the data. If you understand, you can't just let a model run on data and accept its results, yet because the model may not understand some of the regulatory environment that you exist in and so on. Either that or you make the models understand that part of it, which will be a challenge. So I still got the human element to this, but you have quite a bit, as you can see, of advancement in terms of your analytics across the board. I'll mention once again, your strategy, your architecture, your modeling, your processes, and your ethics. Those are my five pillars for you to think about. Try to advance yourselves across all those pillars. And by the way, you can't be far ahead on one versus the others. It just doesn't work that way. They have to go in concert. So maybe you're not excited about analytics strategy. You want to roll up the sleeves and get to work. Okay, but the strategy has to be there to support the architecture and the modeling and the processes. And you can't be great at modeling and you don't have great processes in place to advance the modeling, you see. So it all kind of works together. Make sure that you're moving them all forward and you're moving forward with prescriptive analytics. So when it comes to prescriptive analytics, you might say, well, okay, where do we go from here? Now you're all doing this, but you may not be doing it with prescriptive analytics. Growing, satisfying, or retaining and satisfying your customers. We all need to do this, whoever our customers are. Increasing our operational efficiency, we all need to do this. Transform our financial processes, yeah. Manage the risk fraud and regulatory compliance, yeah. We all need to do all of this. And there are some examples given on the slide of things where prescriptive analytics help you out quite a bit. What was I going to say? So yeah, easier said than done, I know. But really think about how prescriptive analytics can help with all these initiatives that to some degree or another are probably already on your table as a company. But I want to add at this point that I would like everybody that hears my voice in these webinars, that gets my consulting, that hears me speaking and so on. I want them to think about being leaders themselves, not sitting back and accepting, wait for my order to come in from the users and et cetera. First of all, do you believe that data is the new gold? We say this in this industry a lot, right? Well, if you believe that, then you should feel kind of not only proud, but you should feel a sense of obligation there to actually make sure that it's treated like gold and make sure that the initiatives for the company reflect the fact that data is the new gold and data is how we're going to be competitive in the future. And once you do that, you start from a position of data leadership, analytic leadership. We start putting the initiatives on the table for the company because really companies are being run by analytics today. And this is working its way right into the executive leadership. So some executives are getting this to more degrees than others. And I think that's going to fold well for them. So top six considerations, if we begin to summarize here, for taking advantage of prescriptive analytics, simplify data environment that includes big data. So your data environment, get it under control, get the data in leverageable platforms, making sure it all fits together with data integration and data virtualization used in sensible ways. As a matter of fact, data virtualization is second here because people leave that off and shouldn't, if you embrace my ideas that you need to have data into the right fitting platforms for the data, you're going to have a lot of platforms. You're going to have some NoSQL, some graphs, some in memory, some columnar, some data warehouses, some Mars, some MDM, some streaming, et cetera, et cetera. Oh yeah, the data lake too and other big data platforms. You're going to have a lot of that. You're going to need data virtualization for those edge queries at the least. Data governance, data governance, so that the data can be trusted so it reflects something that people don't have to get the data and then think about it and go debate whether it's good or not, crawl into the details which are not evident in regards to how the data came to be, et cetera. The collaboration functions that I mentioned before, take advantage of those in your tool of choice, always be looking to shorten the time to value, and this means getting to more and more of a real-time environment, selectively, of course, but with forethought, okay? So you're thinking about not only today and the immediate requirements, but the future. You're trying to get that data into its platforms, available while performing and managed, et cetera, on a fast pace. And self-service, getting intermediary parties out of the way. I've been saying all along in this series that there's enough work for us as data and analytics professionals to do on the platform side of things, especially if we're trying to make the data scream out what you need to be doing with it. If we're trying to do that, there's a lot of work there. And I want to make it easy on the users, especially for the data warehouse and the data monitor, all those analytical structures. I want to make it easy on them to do what they need to do from a business perspective. And by the way, I'm speaking here to... I used to speak to IT audiences, right? But IT is fairly disintermediated. If you're a technology professional, but you're in a pocket of a department or whatever, that's fine. That's actually fairly normal today. You're still doing all of this, maybe for a departmental level or whatnot, but you want to be getting together and making sure that at the enterprise level, whatever advantages the enterprise can bring to it, or you can bring to the enterprise from your perspective, that happens. That was a lot of words. Sorry about that. Okay, what are the challenges to prescriptive analytics? Okay, these are your challenges if you want to get on this journey, right? Access to diverse, massive-scale data, corporation non-relational data with relational data. Okay, some of you are just getting there. You're laggard if you are just doing that now. Sorry to say, but let's get a move on with that. Get that under control. Get that embedded in analytics. Get that direct access that I talked about to full datasets, not limited to samples, not limited to just summary data, and get immediate access to fresh data without complex data pipelines. Yeah, ability to apply diverse analytics at scale. So the second challenge is scale. One thing to do it in a POC-tight environment, but will it scale to the enterprise needs as the data grows, as you want to bring more variables to bear on your models, make sure what you're doing will scale, and access a broad variety of languages that are ultimately going to be needed upon that data. And you want to finally enable on-the-fly exploration and analysis. So get the right tools to accelerate development and testing, performance and scalability to support rapid iterative analysis, and enable easy reuse across multiple use cases. So I've given you the beginnings of a roadmap, I'll say, to prescriptive analytics, hopefully drawn out a little bit of the distinction between predictive and prescriptive, giving you some maturity factors, so that's a little more detail that you can go by to see where you are. I don't think anybody's in a situation, I shouldn't say this. I don't think most are in a situation where you're too far gone, it's time to give up. I know some companies are kind of riding it out right now, that's fine, that's one strategy, but if you're not, you're trying to be a great company of the future, you really got to get your head around your analytics and start not only predicting the future, getting in front of that future and doing the right thing with it. So with that, I'll turn it back to Shannon to see if you have any Q&A for myself and Elena. William, thank you so much for this great presentation and just to answer the most commonly asked questions, or if you have questions for William and Elena, feel free to submit them in the bottom right-hand corner of your screen in the Q&A section, and again, just to answer the most commonly asked questions, just a reminder, I will send a follow-up email by end of Monday to all registrants with links to the slides and links to the recording of this session along with anything else requested throughout. And to diving in here, at the base level, William, what is your definition of a data lake? A data lake, okay. So a data lake is a staging area for the analytical environment, but it's also a data science laboratory or a data scientist to come in and have exploratory analysis capabilities to, I don't want to say all, because that's pushing it way too much, but a lot of the enterprise data, as much as we can get in there, both from a structured and an unstructured perspective. The data is not highly curated because it's going to be used by only a handful of people in the organization but to a very deep high degree. They're going to have a different set of tools they're going to want to use on it, like TensorFlow and so on. They're doing some pretty advanced things with the data, and these are some of the things that really are very important to the company, defining competitive advantage. These aren't checkbox items. Some of this is some deep thought stuff, and we want to make those possibilities available. And now that our data science is catching up with the possibilities in enterprises, we have to make sure that the data stays on par with that, and the Data Lake is a very important part of that. Elena, anything you want to add? Come visit. And either of you have any suggestions on Data Lake versus Data Warehouse or both? Well, I'm trying to think, didn't I give a whole webinar on that topic? I mean, there's a lot there. Most of my clients need both, and when I say need, I mean, they can take advantage of each of these structures from a price performance and ROI perspective. So the Data Warehouse undoubtedly is foundational. There's no point in everybody building their own. There's so much data that is so common that it just makes sense. And I think we've crossed that bridge quite a few years ago. But the Data Lake, yeah, it's coming on. I hope I just gave not only a description of the Data Lake, but a little bit of my advocacy reasons for Data Lake. So they're both necessary. The Data Lake will probably have more data in it or will have more data in it. And you've got to pick a place where you're going to keep historical data for all time. And it might as well be in the cheaper storage of the Data Lake. So that's one function it might do. We might be able to age off some data out of the Data Warehouse. Maybe not. I'm not hell bent on that for sure because if we're using the data, great storage is relatively cheap. So you need them both at a certain level of maturity. But keep in mind, the data you're going to put in the Data Lake is not going to have to be as highly curated. You're not going to have to know as much about its usage in order to be effective as a data manager. So that's a little bit about that distinction. Fantastic. So, you know, there was a question here if you wanted to touch on it. And Elena, feel free to jump in at any time for any of these. You know, you didn't really get dive into machine learning and deep learning and supervised learning, which again you've talked about before in past webinars, which I can get a little bit into those. But what about those technologies in the context of prescriptive analytics? You want to take that, Elena, from a looker perspective? I think you have the better looker. So we strongly believe in choosing the best tools for the job. And that's the philosophy behind this unified surface for all of your data. Choosing the best tool to, you know, choosing the best database to run all those queries, but also choosing the best products out there for helping you gain machine learning. And not really trying to do all of that in one place, more giving you trusted and trustworthy and governed data sets that you can then push to those and then push back into that data warehouse to surface to the people who need it. Yeah, I mean, I see looker as a tool that has a variety of uses in not only for everything else, but also for machine learning and deep learning within an organization across multiple environments. So it has quite a few capabilities there. But the question may not have been about looker, but about machine learning and deep learning and supervised learning, et cetera, and how that takes analytics forward. As Shannon mentioned, there's whole webinars that I've done on these topics, one on AI, one on machine learning. So anyway, it can just do so much of a better job and a faster job than we can do. It is time to really think about the capabilities that these algorithm sets have for you. For example, patterning, making, I'll say predictive analytics. The data readouts are at such a high volume that it would be probably impossible to really think about looking at this data from an analyst perspective. And this is where machine learning can come into place and actually take over. And the higher the level of requests you can make of machine learning, the more mature that it is. So if you can say, optimize my company. There's a thought that a lot of your big decisions will be eventually made as a company, will be eventually made by machine learning. But I mean, if you can just say, optimize my supply chain, you are doing great. Find the corollaries to the great days that we've had in the supply chain versus the non-great days. And if you label your data as such, it can do that. And so there's just so much more that you can do with machine learning and deep learning than you can from an analyst perspective. And really, they work hand in hand. We have to assess the results of the algorithms. But I'd encourage you to start out looking at some of the regression algorithms so that you can predict the future to some degree and move into deep learning over time and find ways to optimize the customer experience and do some of the things that I mentioned here. Yeah, all of these things can be highly supported by machine learning. And I think I would add one thing is that learning and artificial intelligence and fantastic for getting into that predictive. But there's also a lot that can be done with just access to clean data that's triggering alerts. So we have customers who run up-to-the-minute businesses on Looker. There are manufacturers that are sending their IoT data into their Lightning Fast data warehouses and servicing that through Looker. And if thresholds are passed in that, it's able to not only alert the person who should probably go check on that machine, but even send a request back to that machine to turn it off. And that's just simple thresholds. And using the knowledge that we have as the people who are experts in things to be able to make those prescriptions there. And I suppose, and I expect, we haven't done this in Passover, but I expect that we could do a full webinar on this next question. What are your thoughts on bringing data to analytics versus analytics to data? Well, a lot of companies are finding that their data is now being centered more in the cloud. And the center of gravity of their data is now in a cloud. And cloud to cloud is going to be more optimized. It's all about data flow. I think. And performance degradations that you may have with a subpar architecture now that includes a hybrid cloud environment. So obviously, we don't want to be moving data around just to get it in position to have analytics. We're going to move the analytics to where the data is and take advantage of that performance and take advantage of being able to be right there and to come up with the correct next best business action and have it acted upon without going through a complex pipeline environment. So we're trying to simplify the pipeline within our organization for data at the same time trying to make it elegant. So this is not an all or nothing kind of proposition here. It's a real art form to come up with the data architecture today. And you are balancing things like this. So to me, bringing the analytics to the data is a best practice that we want to adhere to, but there's going to be exceptions in order to make the great data architecture. But I think it's a great best practice to think about as you form up your own data architecture. Couldn't agree more. I think there are a lot of analytics products out there that want you to bring the data to them and like bring it into their in memory or whatever it is that they're offering. And having an option to just choose what's best for you and choose the architecture and the infrastructure that works best for your company and then have an analytics tool fit into that is so much more powerful and gives you so much more flexibility as you change things and adjust your business inevitably. Well, there's lots of good questions coming in, but I'm afraid that brings us to the top of the hour and that's all the time we have for today. William, thank you so much as always for another great presentation. And Alaina, thank you so much for joining in and to look at for sponsoring today's webinar. And just a reminder to everybody, I will send a follow-up email by end of day Monday with links to the slides, links to the recording, and yes, I will include links to past webinars for William Sari's secret, those additional insights as well. And thanks all of our attendees for being so engaged in everything we do and all the great questions and really appreciate it. And again, if you'd like to continue the networking and follow William, you can go to community.dativersity.net. Thanks, everybody. I hope you all have a great day. Thanks, Alaina. Thanks, William. Bye-bye.