Advanced Analytics is the key non-technical expertise needed for executing successful IoT solutions and strategies. Without advanced analytics, IoT data has much less value, and certainly less impact on operational efficiency and developing new profit centers.
In episode 16 of the Let’s Connect! Podcast, Rick Hall, CEO at Aginity Corporation, joins us to define advanced analytics for IoT, talk about Artificial Intelligence (AI), edge computing, digital twin, and metadata analysis.
Rick and Ken explore the inside track on how companies can and should be using advanced analytics to make the most of their data collection efforts and drive real business outcomes. We also dig into consumer IoT, digital twin as a tool for analytics, and how and when you should be thinking about metadata.
Rick Hall is CEO of Aginity Corporation, and a software entrepreneur focused on the analytics market. He has led the development of over a dozen software products and taken several companies from the early stage to an eventual sale. Rick led a group in the purchase of Aginity in March of 2020 and has taken over the CEO role as a result. Connect with Rick on LinkedIn.
Aginity was an early innovator in Analytics Management and launched Aginity Pro early this year. The product is off to a fast start, attracting 13,000 subscribers in the first nine months and is set to grow by more than 250 percent through 2021.
Key Question and Topics from this Episode:
(0:00) Welcome to the Let’s Connect! Podcast
(1:17) Introductions to Aginity and CEO Rick Hall
(2:57) How Data Analysis Works in Consumer and Retail
(6:39) How Do You Measure Activity?
(11:13) Digital Twin as Analytical Tool
(13:51) MetaData Analytics and How They Work
(16:09) Let’s Get into AI for Analytics
(20:30) Final Thoughts on Advance Analytics
Transcript:
- This is the IoT For All Media Network. Hello friends in IoT. Welcome to let's connect the newest podcast in the IoT for all Media Network. I am Ken Briodagh Editorial Director for IoT For All. And your host. If you enjoy this episode, please remember to like subscribe, rate, review, and comment on all your favorite podcasting platforms. And to keep up with all the IoT insights you need, visit IoT for all.com. Before we get into our episode, the IoT market will surpass $1 trillion in the next few years. Is your business ready to capitalize on this new and growing trend? Use leverage is powerful IoT solutions development platform to efficiently create turnkey IoT products that you can white label and resell under your own brand, help your customers increase operational efficiency, improve customer experience or even unlock new revenue streams with IoT. To learn more go to IoT changes everything.com, that's IoT changes everything.com. Now let's connect My guest today is Rick Hall CEO of Aginity. And we're going to talk today about data! Data management, analytics, and how they all intersect with basically everything that underlies the IoT. Rick, welcome to the show.
- [Rick] Yeah, thanks for thanks for having me really appreciate being here Ken
- [Ken] Pleasure is entirely mine and as they're about to find out, my listeners. Can you tell us a little bit about you and your background, Rick, and sort of how Aginity fits into IoT and the whole spectrum?
- [Rick] Sure. Yeah. So I've been working in the analytics space for a long time, like 30 years and came into it kind of from a business perspective of actually a nonprofit worried about how often their members renewed which became analytics and went from there to start up working in the analytics space and then started my own company. It was called G for analytics, which was focused on using analytics and retail pricing and store activity sold that to Nielsen in 2012. Was there for five years. Ran a global practice around analytics for retail and then moved to be the CTO of a retail services company and watched three CEOs go through it a year and decided it was time to go back and do another startup and had this really kind of interest around how business users were able to get analytics done and kind of moving it out of the world it was, you know, kind of the, the core of the priesthood of the IT teams. And that got me to Aginity, which is a, it was a company in the space and I bought that company a year ago. And here I am.
- [Ken] The idea of how data analysis works in the consumer space is really fascinating to me. And retail is sort of the most the leading edge of that right now, I think it's fair to say, because there I don't think most people realize that there's a ton of data points to look at and all of it taken sort of in aggregate becomes really rich and powerful. Like the science around, sort of, shelf placement in grocery and big box is by itself a whole crazy discipline with like gamification elements and all this other stuff. There's a great documentary about it. Now the name escapes me, but I am I am fascinated with that whole element because I think that it can turn into not just really powerful and interesting for the retailer, but also for the customer, Right? I think that there's really a lot of potential for customers to be getting better service, lower prices even though a lot of, a lot of good benefit out of just this kind of activity tracking. And I'm wondering sort of where do we sit with that now? Because it's easy to imagine the scifi version of it.
- Sure
- [Ken] Where does that Sit?
- [Rick] Yeah, it's it's it is, you know, it is interesting space, right? So when I was at Nielsen I ran, what we call the sales effectiveness practice, which was all of the analytics around, how do you price a product? How's it promoted? Where is it on the shelf, in the placement? What stores are you in? What other products do you put around it? It's a, it's a bunch of stuff as you say and all of those things are, are, are now modeled right? So, and if you actually think about it every item on every shelf in every store at every price point is a chance every day to succeed or fail with execution, right? So if you're a retailer, that's your problem space is like it's not just how many stores I have. It's not even how many products I have. It's how many products and how many stores and how many combinations and every day and even multiple points of the day that you can succeed or fail at. Right. And you're trying to do that and get the maximum sales 'course How do you store if you're a retailer with the minimum amount of inventory right? So, you know, you can make sure you have all the right products in the store. If you put, you know, 10 days of supply in every store but that would cost you a lot of money, right? So you're trying to kind of keep your inventory as thin as you possibly can while you're executing all these things, very interactively Right? And so all of it's modeled. So you start with price, right? You know, kind of price elasticity is if I increase the price I might sell lot more, I mean a lot less if I lower the price, how much more? And so you're trying to find an optimal price trying to find a promoted price. So, you know, if it's normally a dollar but I have 50 cents off, what does that mean? And then how does that impact other products on the shelf? So all of that has been built into models. That was kind of what I was doing at Nielsen and what the company that I had built G4 did that we sold to Nielsen. Now
- [Ken] How does, how does that, how do you measure that? Is it just in sort of the Delta of sales between changes and what's the data you're looking for there?
- [Rick] Yeah. So it says very astute question. So you start with, well, let's call it a baseline which is this figmentary, you know kind of concept is all things being equal. How many of these things will I sell in the store? Right? So you start with an assumption about what you've got already. Right? Right. And then you kind of create a baseline which is some historical average. And in that baseline, you're trying to take out all of the the variables that you're modeling for. Right? And then, so say promotion, right? You know, kind of you don't want to consider the promotions in your baseline. And once you have a baseline then everything is an increment against baseline. Right. So if I decrease the price, how many more do I sell? Right? And or if I increase the price do I sell fewer? how many more? And it's all a Delta, you know kind of different incrementality, right? Every one of these variables, price, promotion assortment placement are all measured as incrementality.
- [Ken] That's really, really cool. I could go down the sort of retail nuts and bolts all day but we gotta keep it moving. So I want to start us moving a little bit higher up and talk about executions against data. Because I think that that in the IoT anyway data collection is pretty well understood. We, you know, we do a lot of environmental sensors. We do a lot of activity tracking. We do, you know, there's, there's a lot of mechanism for data collection, but, but I don't think folks always understand the execution against that data. And what they're looking for in analysis, one example that I like that's related to retail is a couple of years ago there was a big move to try, like augmented reality shelf things for coupons. And, and, and to me, it always seemed hokey but I sort of looked that up to the technology not being quite right, but now the technology is actually really good for AR no, one's doing it in retail as far as I can tell. So to me, that seems like a poor execution against a lot good data. And I'm curious about in the IoT world, how should folks be looking to execute against the data they're collecting? What are some of the key factors that they need to be paying attention to?
- [Rick] So, so I think in in the use of data in general, you know I think the model that we like to think about is is almost a science model. And by that, I mean, I have a hypothesis and I want to use data to validate that hypothesis test that hypothesis. Right? And so, you know, kind of I think a lot of early analytics is like "I'm going to come up with the answer." And I think actually the answer is a lot less important now than the question, right? The, the question being the hypothesis. And once you start thinking that it's the hypothesis that matters, then that immediately puts you into the mindset of how can I test that hypothesis as quickly as possible and then figure out, do I have the hypothesis, right? Is it, you know, or do I need to change it? and test, you know, kind of other ideas, right? So I think just analytics think use of data, think hypothesis, think validation. That's the first part of it for me.
- [Ken] That that makes sense. I think that if you, you know, the the survey data collection folks have known this forever if you don't ask the right questions you're not going to get the right data. And in IoT, I think we maybe sometimes forget that because so much of the data collection process is automated or it's environmental, or it's it's sort of a constant data, you know, and and we're only looking for change in in a state or whatever, but if you're not looking for the right change, if you're, if you're not if you aren't sure what the failure state looks like yet those questions, those questions really really matter. And this is another sort of execution against data question but how do you feel about the, the value of sort of virtual environment for, for testing, for measurement for experimentation against hypothesis, that kind of thing the digital twin idea.
- Yeah.
- [Ken] I've seen a lot of interesting executions in small scale. I haven't seen a lot of sort of systematic builds of this space, but I don't think
- Yeah you know, it's, it's interesting. I'm not sure to be honest with you of how important that's going to play as a role. I think that digital twin is, is really useful in a case where you can't test in the real world. Right? So, you know, if the reason you're, if you start your your approach to analytics with the hypothesis the question and your follow up question to that is how do I test that as quickly as possible and the environment you're in makes it such the testing in the real world is too difficult, too time consuming. Now you've got a real case for the digital twin, right? And so, whereas sometimes like, you know concepts like the digital twin which is a really cool, interesting idea like people get so interested in the idea of the twin that they didn't necessarily think about why they were putting the twin there in the first place. Right? It's like, oh, the 20th, the idea. No, no, no, no. The idea is the problem you're trying to solve. And the twin becomes a mechanism to do that. So I think if it's put in that context I think digital twins can be really useful. Like I, we did a project with a customer who's in the shipping industry and, you know kind of testing shift conditions. Well, that's really hard. The ships have to sail off to sea, et cetera, et cetera.
- Right.
- [Rick] So digital twin became a really useful way for them to accelerate their ability to, to test conditions.
- [Ken] Yeah. I think, I, I mean, I think we can think of a hundred examples of, of how useful that would be like I mean, here's this crazy hypothetical. What if a ship turns sideways in a major canal? You know? What would you do then?
- [Rick] Sue the pants off them apparently is the answer, right?
- [Ken] Well, I think we all knew that was going to happen most likely outcome. The minute that situation began, somebody's getting sued but, uh let's all right. So we're, we're, we're talking about a lot of sort of execution pieces. Let's dig into some of the the real nerdery here about, about analytics. One of the things that I find really fascinating is sort of that next step up above the direct analytics into the which and I may have been using this word term were wrong all my life, the metadata where you look sort of the, the outcomes of sort of ostensibly unrelated data sets, you try to look for correlations and try to find unexpected outcomes. That's sort of the part of it that I find really fascinating.
- Yeah.
- [Ken] So please tell me that that's not super useful if that's true and, and that I should stop worrying about that kind of weirdness.
- No, no.
- I think interesting
- [Rick] I mean, first of all, metadata in general is, Is a huge problem and topic, right? So, you know, what, what kind of take us a Securitas route back to your question, but, you know, one of the things that we work on is we work on this idea that, that people in the business can be empowered to use data and do analytics themselves. Right? So used to be, you had to wait for some central team. Now you want people to be able to do it themselves, right? Well, when people start doing themselves, you really need a real clear way to govern that and have some metadata around it. So you end up with like a million points of chaos, right? And variation. So metadata is useful in that You know, circumstance, right? A related place, which I think gets more close to your question is, you know, kind of in and I'll say modeling, I know you were gonna, you know you have a kind of an AI, you know, bent that you have a discussion about. Right? But for me, modeling just means some set of kind of higher order math to look at a problem, right? So, you know, kind of correlation analysis, simple type of model. But, you know, increasingly what we find is people wanting to say, "I want to apply lots of models like maybe a thousand different models to see which one is better." And so that becomes a meta analysis of the results of many models that help you find an answer. And I think that's an, that's another of many places where metadata matters. Since you opened the door to talk a little bit about, about AI. I think that that's all that current AI really is is a higher order, mathematical analysis. I think there's certainly an argument to be made that AI is right now, functionally only good statistics and automated good statistics, to some extent, how much of this sort of process do you feel like should be automated or is there a reason to not have just results come out at the end with the assumption that you've asked the right questions and let it run through the algorithm, the AI, the statistical analysis and then just get a result at the end that says you should tune your your manufacturing rig, you know, three degrees to the left so that it doesn't vibrate in this way anymore kind of thing.
- [Rick] Yeah. You know, it's the, the robot answer bit, which is I think the part that, you know kind of creates all kinds of social discussion, right? Is I actually think it's in many ways less useful than you know, the analysis as a tool to feed humans. Right? So I I'm in the world and I wouldn't use AI, but let's just say advanced math is useful to empower people. But I think a lot of times it's really the people who are going to be making the decisions, not the algorithms. Right? So I think that the output where it says "just do this" is less often the case. And more often the cases I've got a set of different things that pointed direction that, you know I'm gonna, you know, make some decision about. But I think, you know, we use AI as a term to just mean higher order math. I'm with you on that. A lot of times it's, you know, regression analysis. Right? Which has been around forever. It's funny because when I was at Nielsen I inherited a team that was this was a bunch of statisticians. Right? And they were guys, you know, kind of my age so old guys, and they all came up as statistics people. Right? And somewhere in the middle of my time there you know, they got rebranded as data scientists. Right? And, you know, it's like, okay, they're they're still doing the same stuff. Right? And yeah, they were starting to use more ML on top of the statistics, but which is, which is again a different set of algorithms. It's not, you know, kind of this deep learning thing where like I'm just gonna throw a dataset at something. It is going to come back and tell me what to do. That's so rare as a really useful way to do it. Most of the time we're applying models there's multiple different types of models. Those are just calculations uh, And the calculations just tell you, you know kind of something about variation. Right?
- [Ken] Right. You know, it's funny, my listeners will have already learned that the side conversation is the main conversation on the show. So welcome. You know, I always sort of laugh when you hear like you know, Elon Musk is super worried about AI or whatever. I'm like, look, I've met a lot of computer engineers, coders, data science folks. I'm pretty sure that all of them are pretty familiar with scifi. And none of them are going to create a century in AI us leaving that aside, even if they did all of them are just going to program it. So that the answer is always 42. Every time anybody asks you the question.
- There you go.
- So like, I'm not worried about this at all.
- [Rick] It's the ultimate answer of everything in the universe. Right?
- [Ken] Right. That's right. If you don't get that reference, I don't know send me a tweet and I'll tell you where to go read about it. But I, I also don't think that actual send you an AI as possible but that's a whole another philosophical discussion.
- [Rick] I don't know, I'm not smart enough to know whether it's possible. I can say that that's not the stuff that's going on now. Right?
- No, no.
- [Ken] It's what we're trying to do either as far as I understand, not really. So as we start to get near the end of our time I want to start to give you the floor, Rick, and sort of we've talked about a lot here. We've covered a lot of ground really fast. That's what we do. But I like to give you a chance to sort of leave the listeners with a final thought as they're sort of thinking about their own analysis and, and, and data management where should they be focusing their attention? What should they be thinking of? They're thinking about to get the best possible outcome.
- [Rick] Yeah. So I think there are two kinds of things that are really important to me. The first, I think I mentioned at the beginning, which is I think really critical, which is that, you know, all, all all analytics is, is using data to drive an outcome and start with the outcome and form a hypothesis about how you can drive that outcome and then use data to help you test a hypothesis to me like that's how you go about analytics. That's kind of the first thing. The other thing I would say, which is a really interesting side thing, is that used to be this was something done by the priesthood. You had some group of, you know, analytic people in your organization or data scientists, and they did it. And you waited for the answer. It's like, you know, I would say in every organization every user, it should be about empowerment. And so I think data literacy for everybody is, you know, is really important. Like I actually think we should be teaching data literacy in our schools. I think that should be a program now that's kind of a perhaps a crazy thought, you know, kind of I think we all need to use data. We can all use data. I think the basics of it is, you know, just just what we, what we said.
- [Ken] Yeah, yeah And I think that, I think that you're exactly right. And my kids are, are fairly young. They're in fourth and fifth grade. And I'm, I'm looking forward to hearing that they're doing stats here any day now, hopefully because I know it was later than I wish it was. And I was, I was a little bit ahead in math. So for me, so if they get to high school without looking at any statistics, it's going to be a shame.
- [Rick] Here's something that, you know, is just a slide to that point. Right? I was on the phone with my brother and his daughter. Right? And it she's in high school and she's not really a math and science person but she's taken a stats class. And she was telling me how cool it was. And I thought, wow. When I was in school everybody thought statistics was the most boring thing in the world, unless you're a math nerd you were not into it. But data has affected. I think that that generation so much more than the rest of us, that, you know, it was really cool to hear somebody who really is a humanities person seeing how data was relevant to them. And it like, you know, it just made me smile.
- [Ken] That's, that's really cool. It's probably also a function of the fact that the ones coming up are definitely smarter than us that came before So that's good news for all of us too. I, I really want to thank you, Rick, for joining me today. This was a really great conversation and a lot of fun. Thanks again to all of you listening out there. I hope you've enjoyed our discussion. And if you have, please make sure you like and subscribe. So you don't miss out on any of our episodes. We post every week. And I hope you'll leave us a rating review and comment on your favorite podcasting platform. If you'd like to suggest a guest, please click on the link in the description. And we also have a great sister podcast on our network called the IoT For All podcasts. So make sure you check that out.
- [Ryan] Hey, Ken, let me jump in real quick and introduce your audience to another awesome show on the IoT For All Media Network. The show that started all the IoT For All podcasts where I bring on experts from around the world to showcase successful digital transformation across industries. We talk about Applications in IoT solutions available in the market and provide an opportunity for those companies to share advice to help the world better understand and adopt IoT. So if you're out there listening and haven't checked it out be sure to go check out the IoT For All podcast available everywhere.
- [Ken] Thank you, Ryan. Now get back to your show and thank you all for joining us on this episode of let's connect. I've been Ken Briodagh Editorial Director of IoT For All, and your host. Our music is sneaking on September five, Otis McDonald and this has been a production of the IoT For All Media Network. Take care of yourselves. You are listening to the IoT For All Media Network.
Industry News
