Speaker 1: Hey everyone, this is Nick from Tenacity.ai with my co-host Jason. Continuing the FinOps series, we are going to talk today about how automation and data analytics are transforming financial operations, transforming our organizations in and of themselves. And this is not a new topic, of course, it's been going on for a long time, but I think it's important to understand the context of automation being able to do things faster, data analytics, information, data-driven decision making, especially the downsides as well of noise and doing bad fast. And so I think we want to dive in today to the pros and cons of automation and analytics inside of your organization. Jason, why don't we, to dive in first, what are the sorts of impacts that automation has had on our organizations, not just financial operations, but just on our organizations in general over the past, let's say, 10 to 15 years? We've lived this change, we've seen this significant change in the industry.

Speaker 2: So automation and analytics as it pertains to financial operations has dramatically changed over the last 10 to 15 years. One is we have access to more data, right? I mean, like, automation is supposed to help us, it's supposed to streamline processes, it's supposed to help us make decisions in a better way, more effective decisions using data and improve overall efficiency. But if you just look at what we've been able to get from a costing, just from a cost perspective and how we're able to understand what cloud, or not cloud, but what portion of our IT spend down to the individual resource, right? Has given us the power to be able to make these decisions. And we didn't have any of that 10 to 15, can you imagine 10 to 15 years ago, being able to say, that one server that sits, that one virtual server that sits in that farm over there costs us this amount per month, you just couldn't do it before. And so I think it's dramatically changed because it's given us more information that we can use to help improve overall efficiency and as well drive data driven decision making.

Speaker 1: Well, I think automation itself has transformed, I'll touch on analytics in a little bit, but automation itself has transformed because early in our career, I'll call it the pre-cloud era, I mean, cloud did exist, but it wasn't yet pervasive throughout the corporate world. Pre-cloud. In the pre-cloud era, you were trying to automate to make everything similar, right? To create consistency. And so we would write these batch scripts or in the Windows world, write these VB scripts or whatever to try and set up things the same way so that you had this consistency in how you were deploying your infrastructure. And that was a world of overbuying and way underutilized resources and all of that. Automation over time, it went away from trying to necessarily create consistency, it still does that. It creates it maybe at a macro level, but at a micro level made it so we could deploy things atomically really, really, really fast for a very specific need. That's where I think that automation has morphed. And both of those, by the way, there's many benefits, but both of those also create many problems. And I think that's important for practitioners to understand. Automation has all these benefits we want to get there. We want to be building everything in Terraform or CloudFormation or whatever you use to automate your infrastructure so you can wipe out your infrastructure and redeploy it and so on. But it can also deploy bad really, really fast. We've experienced this where someone goes in and changes the template and suddenly API deployments don't work anymore or one of the Lambda functions is messed up or whatever. Something goes terribly wrong.

Speaker 2: Yeah, I think the other thing I would say technology has changed this as well, right? Because 15 years ago, there wasn't a word container, right, 15 years ago, right? There wasn't a Lambda function. And so what did you do with a Lambda function 15 years ago? Well, you just spun up a server for it, right? That's what you did. And that server served that purpose or maybe a collection of virtual servers. So I think also the technology, the further and further we get away from actually managing operating systems, right, and the more difficult it is to do it without code, I mean, how could you deploy as many Lambda functions as we use all by hand? It would be impossible. I mean, you could do it. I wouldn't, but you could do it. So also, I think the technology lends itself to it as well. It just makes it, I mean, you're right. It's changed from something few did to basically make everything the same to now something that's required. I mean, it's almost, you can't function without the automation anymore. You can't deploy, you can't leverage the benefits of the cloud or any of the things that companies are striving to do today without automating this stuff right now. So it's become a niche, nice to have, better way to manage my already poorly managed infrastructure internally in my data center to something that's required just to be able to do your job.

Speaker 1: I agree. And so this can have a significant impact on the financials, on the cloud spending, because as we think in terms of automation, there's a practice that happens all the time. It's the Control A, Control C, Control V in your code where we've seen this mistake where you copy out components and replicate them in order to test something, get it up and running and then it turns out that you've over allocated it or you have something else going to be underutilized. That never happens. Yeah, you're right. You're right. We never made that mistake.

Speaker 2: The replication of poor automation never happens, right? Because we only automate things once we fully clearly understand the problem, right?

Speaker 1: Well, it's converting a great piece of code into a poor piece of code. That's how you do it right there. Spoken like a true engineer. I've done it. I've probably done it in the past few days. So I think that's a good topic. I think we covered off on that. I would love to hear people's questions on this, by the way, or their comments and feedback, because automation is something I'm passionate about and have been my entire career, as have you. And it's amazing to me to watch this transition over time of how we use it and the sorts of impacts that it has. The greater the benefit, the greater the failings as well. So let's talk about analytics. How has analytics changed? Taking financial operations out of it for a moment, analytics has matured so much in the past 15 years. What has that done to how we operate our businesses and our environments?

Speaker 2: Well, one of the byproducts is a huge data problem. That's obvious. There's a huge data problem, right? It's almost the first thing that comes to my mind when I think about data analytics is that, you know, now even how I pay for things has too much data for me to consume. And that just wasn't the case 15 years ago, right, or 10 or 15 years ago. So now if I'm an organization that's not even spending that much, let's just say $50,000 a month in the cloud, that's not that much, right? You know, we're talking about dozens of millions of lines of rows of data on a monthly basis just to understand your costs. That's a huge problem, right? I mean, that's just a huge problem to try to solve. And this is just the cloud bill, right? I mean, let's forget about cloud for a second. Just, you know, think about this from my applications perspective, right? Now I've got 16 different tools I need to plug into my SaaS application just so I can understand how a user is thinking about when they move their cursor from this place to this place on the application. That's now a data point that I need to review and make sure. So it's great because I can now start to track how humans interact. In this example, humans interact with the software, but can you imagine the amount of data it's going to take for every time I move and click and do this and do that? It's just a huge data problem. So DNA analytics have changed quite a bit.

Speaker 1: I think we have always, in data-heavy organizations, and I was in the early days of healthcare data analytics.

Speaker 2: How big was that database of data back then, 15 years ago, just in general, size-wise?

Speaker 1: This was before there were data warehouses. This was pre-cloud. So we were actually building these systems. We were building high-performance compute clusters. We were building data warehouses before there was a data warehouse in the box.

Speaker 2: Before that was even a topic, right? So this was very large for 10, 15 years ago. How big was that?

Speaker 1: Yeah, so this would be 2006. So we were, and it was actually before that, but I'm going to use that as kind of the anchor point, is we were collecting data towards, not that this was our goal, but it became this, is we had the second-largest medical dataset in the United States. And so there were some 1,800, I think, federated data marks, because you have to keep data separate. And if you were to aggregate that data, you had six petabytes, I think, of storage backing that up. But there were so many copies of that data that were in there, because you didn't throw out anything that might be relevant. I mean, you're sending a person in the field to actually go and collect data out of a medical record. You don't want to throw that data that might be relevant in a study later.

Speaker 2: So it could be really useful. Just like a context, that might be 500 petabytes today? I mean, I'm just, like, it's going to be vastly different, right?

Speaker 1: Yeah, well, you know, there's efficiencies now, too. So it is, you know, the world has changed in how we store data. So I don't know what that would look like. But part of what we experienced is, A, we, in that time, so I'm going to use the analog here. In that time, we were bringing data-driven decision making to health care. So things like for chronic care, for comorbidities, say, folks with diabetes, they're going to end up with a comorbidity of heart disease. And this is a path towards, ultimately, they're going to die from that. And we were bringing data to the forefront that would show that there was a way to mitigate risk in this diabetes community by spending money to proactively care for that person's needs, making sure that they had orthotics, making sure that they were getting their insulin, making sure they were taking it, making sure that they were getting to their doctor, talking to a nurse once a week. Maybe it's even an in-home visit, right? Maybe it's sending a car to get them to the doctor's office just for their testing. When you do that, it dramatically reduced the cost to care for that patient over the short duration of them tumbling down the path towards, it's a horrible end with diabetes, but basically towards the 24 months it would take for the disease to run its course and ultimately they would pass away. It's this crazy amount of interventions, and it's really, really expensive, and it's high risk, and it's horrible for the patient. You can actually take care of them for like 25 years if you do it proactively and aggressively and well, and it costs way less. It's way less risk. Way less. And so that was how we were applying analytics in the healthcare industry. Now I'm going to take that in the analog into technology, right? So we would way overbuy stuff in the past, right? Because you're trying to predict out on a spreadsheet using magic product economics to try and guess as to how your users were going to use something. And now we've gone into the cloud world where you have real costs and real utilization. If you manage it, if you have the data, the real time feedback to make good decisions around your infrastructure and how it's operating, and then roll those up into data that can drive your decisions inside of the organization. So then this is a superpower. Now you're able to take and say, well, instead of having to make these huge capital purchases and amortizing them, we can actually leverage the benefits of the cloud, try new technology, see if they're going to drive our costs down, see if they drive our unit economics down and actually gain big benefit from them. That is completely new. The downside to all this is noise in the system. Are you actually looking at the right data? The number one problem I run into in folks that are actually leveraging the data, because most aren't doing a great job at it, but those that are, is it's just an overwhelming amount of data to try and wrap their arms around. And so they're spending all of their time kind of in the noise of the data in trying to massage it into something they want, but they don't really know what they want. So that's the downside. That's the danger side.

Speaker 2: Yeah. That's a good. Yes. Good example. So you were able to, using all of the data you have, figure one of the stories you've always told me is figure out that a big part of it is just getting the person to the doctor, which is like, and it's not about like forcing them to go. It's about actually just giving them a ride. So just paying for a voucher for a cab or for Uber on a monthly basis, the health, and we're talking about a healthcare organization, like a insurance company drives down insurance costs by like thousands of dollars, just committing to that. And so you can, how many examples can you think of in the cloud space is like, there are so many, we just need to make sure that person gets a ride in your cloud environment. That's it. Like there's so much of that in there that you just go, Oh, wait a minute, this, this, this, and this, that can go away. Or all I need to do is change this over here, which could just be an instance type, like, right. And I can then procure this over here and save double. It's like, it's, it just, there's so many of those little teeny. We just need to get you a ride to the doctor to make sure to avoid these thousands of dollars and in month in, in a healthcare costs or so many of those in your cloud environment.

Speaker 1: Let's use a very simple, real example. And I think we're close to time here, but a very simple, real example is when you actually implement allocation, you now have accountability. You now have your FinOps organization, whoever's responsible inside the organization, going to the line of business and saying, Hey, we're storing all of this data for you because this is, this is, you know, a part of the requirement of the product. And it costs this much to then have the business go, wait a minute, I don't need 50 copies. I don't need the dailies of that data. I just need what's relevant. Maybe I need, you know, the summation of the past month or whatever this, this is, we've really seen this, right? Where it just, it starts a conversation that drives a different behavior because the cost was made real for an activity that previously didn't have context. That's you know, that's how analytics has really changed the way we think about our technology, which in the past we so over allocated everything that it didn't matter that, that

Speaker 2: was a different, there was a different driver. You, you could never go down was, was the mantra back then, right? I mean, that it was just a different world because going down was the biggest problem. It wasn't cost. It was going down. And so, or at least we did, that's what we thought anyway. And so, so what, what, what do you, what did we do? Okay, well, we, you want to never go down. We need to make sure we have enough, enough resources. Well now going down, isn't a problem because I have, how many data centers can we deploy in an AWS or Azure GCP? There's probably 50 different data centers and regions. So going down, isn't a problem anymore. It's just a different problem set.

Speaker 1: Yeah. I mean, compared to 20 years ago, it's actually kind of hard to build an application that can fail horribly.

Speaker 2: So people do it, you know, they do it, they do it, but this way to build an application that fails is put out all of it in U.S. East one, in AWS, but that's not a problem of AWS's. It's just way too, you know, way too many people rely on one region. Now that's, again, to your point, you have to code that into your application for it to not use the rest of the data centers, right?

Speaker 1: So I, I think, I think we, you know, I think we wrap it there. I think, you know, automation and analytics have had a huge impact on not just financial operations are, you know, all of our organizations, but yes, when it comes to impacting financial operations, there's both good and bad. I mean, you know, the efficiencies that you can gain, the sort of data driven, data driven decision-making that you can, you can now do it's, it's really important. It's fundamentally changing how we think about technology and how we think about competitive advantage. There's also downsides, right? You know, sometimes the day, sometimes, you know, we're miscorrelating data or we're looking at the wrong data or there's just too much data or we're, you know, we're, we're trying to save store, keep, you know, data that's not really necessary to the goals of your organization or what you're trying to do, even inside of the FinOps part of it, as well as, you know, automating bad can be a bad way to go and not really thinking through the problem. So I think that kind of addresses it. I would love questions and feedback on this topic because I love the real world experiences that people go through when it comes to advancing their technology, analytics, automation, et cetera. Jason, any parting words? No, I think we got it, man. Awesome. All right, folks. Head out to reddit slash r slash cloud cost optimization. You can find us there. Ask your questions, drop your comments. We're happy to answer them on the show. You can also head out to tenacity.ai, download the cloud cost optimization handbook or, you know, hit us up for a demo or just to have a conversation around what tenacity can do for your organization and helping you in your FinOps implementation or getting your arms around cloud cost optimization. Thanks for listening, everyone.