Finding the cause of an outage or slow-down, therefore, can be frustrating and time-consuming. Additionally, it can lead to all sorts of internal “politicking” (to put it nicely).
But even once there’s a coherent “map”, correlating what the business wants with what the IT stack can provide is another headache, and one that’s not easily achieved. Unless, of course, you happen to use AppDynamics.
In this talk with Jim Cavanaugh from the company, we look at how organizations are joining up business aims and objectives with the technical nitty-gritty of networks, clouds, connections and all the whirring boxes in data centers.
Joe Green (host): Welcome to the Tech Means Business podcast. Now, each episode, I like to talk to interesting individuals from companies and organizations who I might feel have got something to say and to contribute to this space, where we can talk about this thing called technology that’s at the heart of every business.
Today, we’re talking about customer experience — I know it’s something of a buzz phrase — and how that customer experience relates to and is affected by issues like network infrastructure; the influence customer experience takes from choice of cloud provider, or even at the end of the day, even the choice of ethernet cables in the company. Now it sounds odd!
But here to explain all this, and let us know how we can effectively look at the effects in a business that technology can have, I’m delighted to be joined by Jim Cavanaugh, from AppDynamics. Jim is in charge of the APAC and Japan for AppDynamics. So Jim, please tell us a little bit about yourself, what you do and what AppDynamics does.
Jim Cavanaugh (guest): Thanks, Joe, for having me, I really appreciate it. I’ve had the pleasure of living in Singapore for almost five years now with my wife, and I have two young girls at home.
On the work side, I have the responsibility and pleasure of running Asia Pacific and Japan for AppDynamics. Really, the magic of where AppDynamics plays is providing the correlation between end-user behavior and business outcomes. So if you think about it in our personal lives, whether you’re using a mobile device to get online and leverage a banking service, order food delivery, maybe order a car, order any type of good or service, well think in terms of consumers have, what’s that experience like?
Now for the company providing that service, and for the IT group providing that service, that’s a pretty complicated set of things that happened behind that. So while we may push two, three, or four clicks on a mobile app, suddenly we have food at our doorstep, or we have a car waiting for us, so we don’t have to get in the rain, or we’ve moved money from one side of the world to the other. And we expect that to happen flawlessly! There’s a whole bunch of things that happen behind that application, that allow for those services to work in real-time. And AppDynamics helps companies to fix those and optimize those mobile apps in real-time.
Joe Green (host): So as you say, behind those services, behind the application that might be on your smartphone or behind the website, you know, there’s a whole stack of stuff isn’t there? There’s a database and the website server and, I don’t know, a kind of credit-checker. And all these different components, if you like, traditionally, have lived in a company’s data center. But as far as I understand it, that’s changed, hasn’t it? I mean, the topology of the network, the topology of systems are spread out over the cloud, you know, via API’s: I’ve got that right?
Jim Cavanaugh (guest): You nailed it! And the challenge is that many of the IT tools and systems were built to be able to monitor monolithic applications that were built pre-microservices, pre-cloud services. So all the things that you just articulated are things that many of the systems that IT people are forced to use, really weren’t built to go off and monitor and tackle.
And the second difficult thing for an IT organization is that most organizations have dozens and dozens of tools. So even if they find a tool that will work in one component of their environment, they then have to correlate across multiple tools to draw some analysis. And as we know, the world is so dynamic today, you may take out your phone, and you go to book that dinner or lunch. And if the application is at all slow, then you’ll just select another service to book that dinner or lunch.
And same thing in the ride service business. If you’re waiting, and you’re trying to beat the rain, most people have two, three, four different car services that they can potentially leverage. So the pressure on the IT group to be able to resolve problems in real-time is immense. Because from our personal experience, you and I and others are thinking literally in microseconds of: will that “application/service/ whatever I’m using” be slow? So I’m gonna go use something else.
Joe Green (host): Yeah, I think that’s a powerful analogy. And I think there’s more to it as well, I think that when you start using an app, or a website or a service, there’s a significant amount of buy-in, you know: you need to give people your email address, you need to create a password, and then you have to wait for the email to come back and confirm your account, and then you’re up and away on your new app or your new service. Now, if what you’ve bought into, if you like, through that process, this thing that you’ve committed to, if it doesn’t work, or if it’s slow? I mean, that’s incredibly frustrating. And isn’t that really what enterprises at the end of the day need to avoid?
Jim Cavanaugh (guest): Yeah, you’re right. And, you know, a couple of statistics to validate what you just said. The APAC tension index survey, 54% of consumers said they’d actually pay more for a good app or service that had a better digital experience. So in spite of all the work we go through to download that app and put in all of our information, and again, most people have multiple applications that would allow them to do whatever they’re trying to do. And consumers are acknowledging that they’ll pay more for that experience. Another interesting stat: 85% of consumers said that over the next three years, the digital experience will actually drive the selection of the brands that they choose.
Joe Green (host): So I guess that means that it’s the back end, as well as the GUI, the graphical user interface of apps and websites, that will drive decisions. And it’s those elements really, that are more effective in terms of customer experience and brand loyalty than say, you know, enormous ad campaigns. Because, we’ve all done it, we’ve switch to a new product or a new platform, because of a big an ad campaign. But if the experience that we get is, for want of a better word, fairly crappy, then we can certainly get turned off the whole brand. You know, it reminds me really of talking to my non-technical, if you like, family members, who will show me an app or show me a website and say, you know, this is rubbish, this is garbage, because I don’t know — I tapped here, and nothing seems to be happening. And actually, it’s just one element, you know, in what’s often a very, very complicated and complex back end, that taints an entire brand.
Jim Cavanaugh (guest): That’s absolutely correct. And the difficult thing for again, for that IT organization is being able to find whatever that one thing is: is it a third party service? Is it a service provider? Is it that line of code that was added in? Is it a security component, and there are hundreds and hundreds and many times thousands of different things that could go wrong in that stack. Meanwhile, the consumer literally in seconds wants that problem resolved, so back to, you know, what do we do? And how do we help customers, by leveraging our technology, which is inclusive of AI and ML. We can actually allow our customers to anticipate a problem before it happens. So they can go resolve the problem before you and I or other consumers actually have that experience.
So think about the concept of ordering car in the rain. And you’ve got massive load as everybody decides that they want to order a car as opposed to take another mode of transportation. Our technology in real-time can: number one, identify that there’s massive load, and that there’s going to be a potential performance issue. And then prior to that issue, actually being experienced by the end-user, [the AppDynamics customer] can leverage our technology to move a workload — as an example. So now we’ve remediated the problem prior to the consumer even experiencing that degradation, maybe even more impactfully, we can actually quantify for our customers, again, in real-time, what the financial impact of that was.
So it’ll actually it probably makes sense: if the mobile app for that car ride service is performing poorly in the rain, they’re probably going to lose some revenue. But wouldn’t it be nice if they knew in real-time exactly how many people experienced a performance degradation and then what those users did, and what the potential impact of that revenue was by those users. Either they’re abandoning the site, or choosing to do something else on that site that then book car.
Joe Green (host): That’s really got me thinking, actually about the rain issue. And it’s I know it’s a it’s a kind of silly analogy, really. But it’s almost like, you know, we’re quantifying the cost of rain, literally, to the business, we’re actually putting in empirical figures. I wonder if you’re the person to ask really, because it strikes me just while you were talking about that I was thinking about a weather app, which I was running on Android, and is now no longer available called Dark Sky. And it kind of taps into local weather radar. And I thought, wow, it’d be really powerful, wouldn’t it to be able to, you know, pull in that data that minute by minute view, from rainfall radar, and then actually be able to predict the financial impact of rain on a business, you know, talking about the ride hailing business there? Because essentially, we’ve got everything there at our fingertips, haven’t we, we know it’s going to rain. And we know our systems aren’t really capable, say, or are capable of coping with any peaks in demand. And, and we know, therefore, what will the next downpour literally cost us in terms of dollars. And it’s that type of intelligent big data, I think that has some really exciting possibilities.
It’s gathering this information together in a business sense, which is usually where things like artificial intelligence can help because, you know, computers are good at munching through big amounts of data, as you know.
So in AppDynamics’ case, you guys call it the Central Nervous System, tell us a bit more about that particular implementation, about how that works in a business.
Jim Cavanaugh (guest): Sure, some of it goes back to the real-time pressure that IT organizations face. As customers continue to want more services, they want them faster, they want a seamless performance in the delivery of those. And the back end is more complicated. As we talked about, IT organizations now have to deal with the challenges of cloud and microservices and all the other complex things that are in an organization that today deliver a service. In addition to that, customers are trying to make real-time business decisions.
So in your analogy of ordering a car, and the impact of rain and or other things on that, if we leverage human interaction, if we wait until humans actually have the time to go in and analyze all that data, then very likely the opportunity to leverage that is already passed.
So technologies like AppDynamics do provide customers with the ability to, in real-time, leverage AI, ML, to go off, and without that human interaction, go off and make tweaks or changes to that IT infrastructure, so that the customer doesn’t feel any degradation of performance.
Joe Green (host): Yes, given an infinite budget and 10,000 staff, of course, you can tweak your infrastructure, you know, based on data, because you’ve got 10,000 people sifting through endless Excel spreadsheets. But of course, who’s got 10,000 people at hand? And I think the point about AI is that it can do just that: it can munch through data at a much lower cost than an equivalent number of human beings. And of course, doesn’t need the bathroom and works all around the world, 24/7!
Jim Cavanaugh (guest): So if you think about it, in “the old days,’ people would have technology — monitoring technology — that would give them green, yellow, red, and they would have some level of performance for their application or service. And if things were green, then that was good. The interesting thing is that doesn’t tell them things are green. But if you increase the performance, if you actually gave your consumers a better experience, would you actually bring in more revenue?
So what AppDynamics can show is in real-time, where consumers are having a better experience than maybe was identified as green, or optimal in the legacy system. We can show how more revenue is coming in from those customers that are specifically achieving better performance.
So as an example, if the application is now giving [or] returning the experience to the user at 10%, then last time, we might all think all’s well. Then maybe more users will use the service, maybe they’ll click a few more times and the customer doing that can monetize it. What we can do in real-time is show the customer.
Think about a bank. We can show a bank in real-time, your users that have an experience that’s 10% better than your “green”; than your normal. Actually go through the online application to fill out for a mortgage, a motorcycle, a car, a boat, or in the food example, food delivery example. While there’s a baseline that says the performance of the app is fine, the customers that have a 12% better experience than the average normal, are actually ordering 17% more items.
So what the IT organization can then do is go to the business and say, instead of being a cost center, I could potentially be a revenue center. The data shows that we can increase by 17%, for this set of customers if we provide the application 10% faster. And then the CIO can go to the CFO, the board, etc. And, specifically say, here’s the investment that I need, so that I can provide that, because certainly there might be some investment associated with providing that faster service.
But in the past, that was guesswork: I’m gonna go build a battleship, my customers are going to come to my application, and I’m going to monetize it. So now we can actually think about it as a dial that you could “turn revenue up and down” based on the performance of your application, with the ability to quantify that in real-time,
Joe Green (host): Yes, gone are the days, if they ever existed, of course, when the IT function could go to the boss and say, “Look, there’s this new box that we want, and it’s got flashing lights on the front that light up and amusing colors. And we want it because it might work or, you know, it might help or at the end of the day, we just want it because it’s coooool!
But these days, what you really need to do is to have to go to the boss — to move up to the C suite, if you like — and say: purchase this system, this service or framework that we’re going to buy, it’s going to create an uplift in sales, or it’s going to cut costs here, and it’s going to cut costs here. And the point is, I guess, that the more figures, the more empirical data you can take to your boss, the more weight you hold, if you like, or the more weight your argument holds.
And in that way, you’re beginning to move it away from being a cost drain, and an endless succession of costs leaving the departments and actually turn IT into a more of a strategic function.
Jim Cavanaugh (guest): Absolutely. I had a CIO of a large retailer, say (his words), by leveraging your technology, you’ve changed my relationship with my CFO! So what does that mean? He said, I used to have discussions with the CFO that started with “I think”: “I think if you invest in this project, I can do that,” “I think if you give me this amount of money, we can deliver that.” He said, Now I go in, and the words that I use are “I know based on this data that I’ve leveraged from AppDynamics, I know that when our customers experience X, they spend Y. And that’s drastically changed that relationship, because he and the CFO are dealing with real data, as opposed to guessing based on extrapolation.
Joe Green (host): Yes, and you can, of course, dive deeper into customer experiences here, and begin to model customer behaviors and potential customer behaviors. And therefore, the quality of customer experience.
If for instance, we have a peak of 10 times in demand, say, at the end of Ramadan, or the beginning of a holiday season, or Couples Day or rain, as we talked about earlier on, or the train shut down because of a fault on the track somewhere up the way, on a particular day: then, we can extrapolate empirical data. (I love that word empirical, by the way!) You know, how that data affects infrastructure and how it affects resources, and therefore how we might get new infrastructure or changes in resources going forward. And that leads, of course, into this correlation between gathered data, or business activity and the physical infrastructure, and the physical facilities. How do we draw those strands together? What’s the best way of going about that?
Jim Cavanaugh (guest): Sure, there’s a couple things that orient on. The first one is this concept of a end-user journey, or a user journey. So as opposed to thinking about the IT organization from the IT perspective, thinking about the silos of physical gear, software connectivity, different data centers, third party services, we think about the world from the end-user’s perspective, even when approaching our customers.
So what our technology allows our customers to do, is look at the journey from the mobile application, or the website if someone’s coming in via a website, all the way back through that entire spaghetti web, that complicated web that the customer has.
And as I mentioned, the first thing that we provide is this correlation between what type of experiences the user have, and then what type of behavior, i.e., are they spending money or selecting a service, when things aren’t optimal, when something’s going wrong?
What our technology does in real-time is identify exactly where the issue is. So if you were trying to order ice cream, tonight, for your kids’ after dinner, and you went to the app, and it wasn’t working, we could tell that delivery service that it’s actually a specific line of code, or it is a service provider challenge. Or maybe it’s even your home situation, because you’re on your wifi and someone else is watching Netflix and someone else is working from home in the other room and the kids are studying in the other rooms. But, we can tell them in real-time exactly where the issue is.
Then in many instances I use the example of workload optimization, we can offload or we can remediate that problem in real-time. In the case where there are challenges within that IT organization that need remediation, they need to upgrade things, they need to put more storage behind something. Again, obviously, the customer would have to go off and action some things, but [it] will tell them in real-time exactly where the challenge is. So that they can go off and remediate that.
Joe Green (host): Yeah, I think that’s a good example. And let’s take it as an analogy, a contention ratio problem, if you like: too many people are on a limited connection. You used the example of being at home. And so therefore, it might be too many people on Netflix, I mean that the situation is the same.
But of course, the trick is finding out what the issue is. And its causes amongst not just a simple home, wifi network, and ADSL, but you know, amongst a wildly complex enterprise IT system.
Jim Cavanaugh (guest): The other thing is that people have, in the past, sort of guessed at what the consumer going to demand. So an app’s designed, and in theory, [the developer is] going to provide some type of SLA for that app, [and] it gets pushed out to the world. And then people love it, or they don’t. And sometimes they don’t love it, because there are some challenges with the app. Sometimes it’s the situation you just articulated, where there’s a bunch of contention at home. But sometimes, the company just might have missed what the consumer is really going to go and demand. That challenge is that consumers really wants that right away and resolved in real-time. If there’s an issue, the application might be performing as designed. But the company may be missing a big business opportunity, because people are looking at that application and saying, well, it works. But it doesn’t work at the speed that I want it to.
The other interesting thing; when we when we talk about applications and performances, three quarters of people say that if they have an application, and they don’t like the performance, and they delete that app: they’re not going to take the time to go tell anyone. So we think of applications in terms of, well, an application has a poor rating, or a poor score in an app store or online. But the reality is very few people will actually go and complain prior to deleting the application.
So the challenge, obviously for the app owner, is they create an application, they push it out, they start to get some data, but they may have a lot of people leaving or deleting their application or maybe they don’t even take the time to delete it, they just stop using it and start using something else.
Joe Green (host): Now in many cases at the lower end of the market, you know, losing 1% or even 5% of business for whatever reason, well, it’s probably okay. And part of that’s going to be that the cost of amelioration, the cost of fixing the cause of that 1% or 5% loss is going to be higher than lost sales.
Okay, so that’s, I think, that’s probably fair enough, I would say. But at scale a 1% loss, you know, due to poor customer experience, which is what we’re talking about here, that can spell millions in lost revenue.
Jim Cavanaugh (guest): Absolutely.
Joe Green (host): Now I want to touch on another issue. And that’s that of large systems, you know, bought and developed over time; that kind of [system] well rooted into an enterprise. We’re talking really about packages like NetSuite, SAP, Salesforce, and the like, you know, ERPs: enterprise resource planning software. Now, if it turns out that after our investigations that it’s those proprietary, more closed systems that are the problem, and I’m assuming there’s not much I can do about it?
Jim Cavanaugh (guest): The first challenge associated with applications, whether it’s SAP or another large application is very often the service, the app that is tied into that is leveraging more than just the core SAP application. So the first piece of, or, the first opportunity for us to help is identifying back, to if it’s an application that’s leveraging SAP. We can actually show the customer, what else is it leveraging, and believe it or not, [has the] IT organization become too complex?
One of the things that we provide is this end-user journey map, that’s documenting from the end-user all the way back through the complex systems. But very often, when we show that to an IT organization, their response will be, well, I didn’t realize that application actually calls our credit check application, I didn’t realize that application actually has a call to our external third party provider! I didn’t realize! And most of them say I didn’t realize how complicated our application was!
So the first part is just mapping the topography to allow a customer to figure out what components it is touching. And then the second piece is while there are tools (back to my comment before) there are tools that allow you to go optimize inside of SAP as an example, or optimize inside of Oracle or optimize inside of other areas. Because applications are typically going across a bunch of different components.
You really want to think about a tool that allows you to traverse multiple hops. In the past, what was happening is you would have people [who] would have their silo-ed tool. So there might be a database person that looked at their tool, there would be a server person or storage person that would look at it infrastructure tool, you’d have a networking person that would look at a networking tool, and they would all have a view of the world.
But back to you, you and I, and the consumer: what we really think in terms of is, well, how fast did that application work for me? Was I able to order that car in the rain in a matter of seconds? We’re not thinking about whether the network’s up, or whether there’s contention on the server, or whether the line of code inside the database, whether that’s a packaged application, or a custom written application, whether the line of code is perfect or not. So, it really goes back to having one tool that allows the IT organization to figure out in real-time, where is the issue. So that’s the first piece.
The second piece is baselining. So if you think about the concept, most businesses would have an appreciation that there are some periods of time where they have a stronger load on their environment than other periods. A dramatic example would be: we have a customer in Asia Pacific, it’s a large government organization, and they collect tax payments, and obviously give out tax refunds.
So as you can imagine, they have some acute pressure when it comes to the tax filing deadline of people using their application. So on their normal Tuesday prior to the deadline is different than their normal, regular Tuesday. What we can help them understand is the baseline: on tax day, over the past three tax days, the performance of the application has been X at this time of day with this number of people filing and this number of people hitting the application. That allows them to do some really dynamic things around anticipating, before issues come up.
The third thing that I talked about is the ability to dynamically, in real-time, kick off actions. So I gave the example of the ability to optimize workloads by moving a workload from point A to point B, so that you can remediate a challenge in real-time without an end-user being impacted.
Joe Green (host): Now, just to go back to that first stage, when you’re trying to pinpoint a problem. Now, there are going to be lots of different people involved in the IT function, and each of whom are carrying their own rattle bags of Bash scripts and tools and bits of software and methods that are designed specifically for their role: databases, or websites, or APIs, or DevOps, and so on and so forth. And of course, this brings us then to the concept of the war room, where everyone has to sit down in the war room and try and get to the heart of an issue or a problem. Let’s be honest about this, I think there’s a good deal of finger pointing that goes on. And of course, really, that’s the last thing that a grown up enterprise needs, isn’t it, this kind of blame game, this endless war rooming?
Jim Cavanaugh (guest): Yeah, you’re absolutely right. We had a customer that explained it. He said that AppDynamics was their flashlight. He gave the analogy: he said the war room was, in his words, people from different departments that were all in a dark room, and they were all positioning and talked about MTTI as “mean time to innocence” as opposed to identification! So his task in that room was to prove that it wasn’t his organization’s fault. And so if you think about that war room, and you think about this concept of a flashlight, what our customers can do very quickly is go identify the exact nature of the problem, and then go figure out what the resolution is in real-time.
What we see for customers is from an identification or MTTI, and from a resolution or MTTR, drastic reductions — very often more than 75% [in times to resolution].
And another example, we have a bank in Asia Pacific that (in their words): they’d spent more than three months, and they had a weekly war room where they had not just IT but the different components of the business; marketing, security. They would spend several hours every week trying to diagnose what they deemed a very critical business challenge. They installed our software and within 20 minutes, they were able to find the line of code. And that day, they were able to release the code. And the problem was then solved.
So, the ramification of being able to go identify the root cause very quickly, as well as the remediation, for many of these companies, is massive when it comes to increasing top line, but also decreasing the cost associated with remediating IT problems.
Joe Green (host): Now, if people want to go and have a look, and maybe get proof of concept for themselves, and see what the possibilities are, what are the next steps that they need to take?
Jim Cavanaugh (guest): People can go to the AppDynamics website, https://appdynamics.com, and they can download a free trial. We allow customers to use our technology for a couple of weeks as a proof point. In addition to that, we will work with the customer to help them identify the business value associated with remediating whatever the application is that they select.
Joe Green (host): Now, as ever on this podcast, we’re running out of time probably well before we’ve talked ourselves out. So it only really remains for me to say at this point, a big thank you to Jim Cavanaugh, from AppDynamics. It’s been a really good talk, and I think it’s going to appeal actually, to both business and IT professionals, which is of course, what Tech Means Business means — it’s all in the name, it’s what it’s all about. So Jim, thank you!
Jim Cavanaugh (guest): It was a pleasure, Joe, the pleasure was mine.
Joe Green (host): And I hope you can join me, you the listener next time we take a dive into the technology that underpins just about every business, from the one man band right up to the global multinational. Until then, take care and see you next time. Bye!
