 from the MGM Grand Hotel in Las Vegas. Extracting the signal from the noise. It's theCUBE, covering splunk.com, 2015. Brought to you by Splunk. Now, here are your hosts, John Furrier and George Gilbert. Okay, welcome back everyone. We are here live in Las Vegas. This is Silicon Angles theCUBE. It's our flagship program. We go out to the events and extract the signal from the noise. I'm John Furrier, founder of Silicon Angles. I'm John McFos for this segment. George Gilbert filling in for Dave Vellante, my usually co-host. But this time, George is the big data analyst at Wikibon. Our next guest is Alex Giddelson, system admin at American Public Media, which is hard, not NPR, but they're like brother, sister, siblings in the awesomeness of content and syndicating that on a huge level. Welcome to theCUBE, appreciate it. Thank you. Great to have you. Obviously, we're big fans of what you guys do. A lot of volunteers, it's a non-profit, a lot of great content, both the fine jewelry of getting news and editorial. So really, really great stuff. So, how are you guys using the data in Splunk because the media business is struggling, right? Everyone's dying. The democratization is not happening. Oh, we think it is happening, but not for banner ads. But quality content still is a premium. But the old page view-based content's dead, dying. Quality's back. So you guys are running a robust business, such as the non-profit. How are you guys using the data? Can you share some insight? Sure. So, yeah, so American Public Media, we have a lot of podcasts, a lot of things that people love to listen to. And what we do is, we'll put in, we call it underwriting, really an ad. And we'll use that to try and get something back from it to pay for it, because it's expensive. It's expensive to host this stuff, it's expensive to produce it, and we want to produce the best content that we can. What Splunk helps us with is allowing us to know who's getting what podcast, where are they getting it from, how are they getting it, what app are they using, you know, are they going to our website, and then we can make decisions based on which shows should we do, and then which ads do people like, and who is listening, you know, what state, what region, so we can try and target that, because our software allows us to target the ad, so that, you know, if they're in one region, we can try and target that, because that's what it seems like everybody's doing to get their podcast to, you know, to pay for themselves, so that they can keep doing it. And you guys do great podcasts, so we're very familiar with all of them, so it's great stuff. The diversity around what makes it work is kind of complicated, it sounds like easy, just put up a website and say download this podcast, but the CDN's involved, take us through some of the under the hood challenges on the data that and how you're getting it, how's the CDN's involved, you have a custom player, are you measuring, instrumenting, time of listens, all that good stuff, can you share some insight into that? Sure, so the big thing is, I mean, the, what's the word I'm looking for? What the industry is trying to do right now is, everybody wants to listen to what they want to listen to, and they want to listen to where they want to listen to. They don't want to go just to one place, unless it's really, really good, but they want to get it everywhere, and we want to be able to provide that. So we have a player on our website, you go to our website, he'll start playing when you hit go, and he'll stop when you leave, you know, all that kind of stuff. But then we partner with different apps, like Stitcher and Tunecast and a few others, so they can get audio, and then we can try and get some statistics back. We, you know, we still, we have our CDN, so we'll upload the files to the CDN, and then they'll, you, we have software called AdsWiz, which does all the special sauce for making sure the ad shows up correctly, and then we can target it, and that our employees have, you know, a nice console that they can go into, and be able to manage, you know, what goes with what, and how that works out. And then Splunk gives us all that data back, so we can look at it and know, you know, this podcast was downloaded this many times by these people, this app downloaded half the podcast for some reason, this app downloaded the entire podcast, because every app acts a little differently. One of the things I didn't know, which Splunk made it very clear, is that when you use the Stitcher app, you download the full podcast, right? You get it ready on your phone. But then Stitcher will occasionally check back, download another 4K of the file to see did the file change. And when you're trying to do analytics, and you're trying to know how many people downloaded your file, how many people listen to the underwriting, because Stitcher keeps downloading this 4K chunk over and over and over again. You have to be careful, because if you're just looking at downloads, and you're not looking at amounts and sizes, you will count that as another download. And then your numbers will be over-inflicted. Yeah, yeah. I mean, Apple had this problem early on. They would download the files, and you had no idea when someone listened. You count as a download, but now you got to measure some sort of CPM or... Yeah, how do you do it? So how do you guys do it? Well, what we try and do is, we try and isolate and say, if somebody downloaded the file, we're going to count that as a download. And if somebody downloaded the file twice, then we're trying to look in like, is it a specific app that we can somehow whitelist? And Splung makes that very easy to find and then do something about putting a filter, something like that. Give us some insight and day-to-day what you guys do. You're in Minnesota. What's going on across the company? How many people are working on the technology piece? You obviously have reporters and whatnot in there, around the fields, freelancers, full-timers. What's the sausage factory of American media? I mean, because this is a great operation. Just share some color of what's going on behind the curtain. Let's see, the day-to-day, I mean, it's like any other company is someone, at least in the IT side. We have the projects coming down saying, we want to do this, we want to do that. We have a news floor. We have one of the largest news floors when it comes to public, I think, public radio. And they have people out and about all the time. We have reporters out in the field, interviewing people and finding out stories. We have editors in-house that are making stories ready. We have a website for people to go to and get news. We also do a classical music station. So we have people figuring out what classical songs are going to be put in, what's the order of listening. We have classical station that's syndicated out to the entire country. We have the BBC going through us all the time. And so we have, as part of our news program, we'll have stories from the BBC interweaved with stories that we're doing so that when you listen to our content, you get all the things you want. You get the regional stuff, you get the national stuff. It all comes together in, I think, a very seamless, very easy listening. You hear a lot of velocity. Yeah. The volume and velocity. Yeah. Good flow. Yeah. So a question back in the sort of infrastructure or the plumbing that makes it all work. One of the things we've seen with customers, whether it's Splunker or Hadoop, well, primarily Hadoop, there's so many moving pieces that it takes a lot of skill to get going. It was, you know, you could use it as a platform to do some of the same things you're doing with Splunk. How long did it take you from sort of getting it out of the box to up and running with your first production kind of app? Well, so with Splunk, you know, the initial installation was pretty easy. Getting install was pretty easy. Getting data in there was pretty easy. The thing with, I mean, the way that I work is I like to, I create my report, I look at my results and then I'll be like, oh, you know what, this is a little different. So tweak the search, I'll do it again. And I just keep iterating through it. And that iteration hasn't really stopped. We just keep building on our previous work. But that's then exactly like what happens with Hadoop, which is an iterative exploration. So it sounds like the front end collapses in time because it's so easy with Splunk. And then you get to the iteration and the exploration much quicker. Now, once you get to that stage, are you thinking about what happens next? Because with Hadoop, it's a general purpose platform and you put other apps. Have you started thinking that way with Splunk? Yeah, we have. I mean, right now we're focusing, our focus initially was on podcast data. Then we're moving into streaming data and then interweaving all of that together. So our reports will give you the overview and give you the separate things. The big goal has been, we have a woman in our organization who takes all of the content numbers and puts them into one nice PDF that she sends to all of our executives. And the thing I said for Splunk was, Splunk will do that. And so it's building every single little component of a report and it's still an ongoing process. And if we can finally finish all that, I think then the next step will be just getting Splunk in front of the users of the organization. We had already one of our content producers came down and asked, you know, you've got all this reporting, how much of the listenership per episode is in the first day, the first week, the first hour? Like, you know what, we can guess, but was it really? And so one of my coworkers went, created a quick report, and in minutes showed that sure enough, the first day, 95% of the listeners. And if you don't have it ready, if you have something wrong, you've just lost all that traffic. Yeah, and that's why with mobile apps and media, your launch is critical. You've got to have your best content out there from day one. And also the cadence of the programming matters too. Frequency, right, with certain schedules, do you find that to be something? I think with the prevalence of mobile apps, especially in iTunes for podcasting especially, with how many people just subscribe to their podcast and a lot of people care, but I think a lot of people don't. I think it's just, they want it to be there and as long as it auto downloads when it's available, if it was a little bit late, I mean, some people will obviously care. But I know me, when I listen to podcasts, I don't care when it came out. I just care that when I go to listen to my phone, I have a podcast ready and waiting for me. Yeah, yeah, exactly. Versus like two weeks without missing one, like at least having a deliverable of new content. That's it. Do you like Stitcher app by the way? No, I mean, I've used it a little bit. I mainly like a different one, it's called I Can't Remember. You just put stuff on SoundCloud at all? We do a little bit. We just start. I mean, as the organization, we definitely like Stitcher. Stitcher is one of our top apps. Personally, I use a few others, but Stitcher is definitely very popular. How about iTunes? Oh yeah, huge. Huge numbers on iTunes. Huge numbers. I mean, it's like, you have to be in the top list of iTunes. If you're in the top 10 list of iTunes, you get a huge amount of traffic. And that's going to be great. When you were talking about streaming data, what type of data was that that you're getting? Well, so we're a radio company, we've got our radio stations, and then we also stream all those radio stations live. And we also have HD radio stations that you may or may not get depending on your car or your radio. So we stream those as well. And so you can go to a couple of our different sites, either in PR News or The Current or your classical, and you can just start listening. And we have people who, maybe they left Minnesota or California and they still want to hear the content. So they'll just listen over the web. And we want to know who's listening, how are they getting it, you know, the same kind of stuff. Yeah, downloadable media is very tricky. It's so hard, live on on demand. And something a lot of people struggle with. So best practices, stepping a step back. So people out there that are doing a lot of podcasts, that might be interested in saying, hey, cause podcasts is one of those things where, when it works is you have very good loyalty, very good stickiness, because you got to commit to listening to it. So it's hard to really measure that and also justify the investment. Cause manager wants to know, wait a minute, you didn't have a zillion viewers. But you can have a really niche audience and have really high impact for an advertiser or an underwriter. So what best practices can you share with folks out? There's a ton of people out there that I know watching, certainly watching our program that are either doing stuff on medium for blog posts, putting podcasts out on SoundCloud or who want to collaborate and put some good content out there. What's the best practice for them, a company or a small, growing blog or whatnot? I mean, for anybody, I'd say some of the best practices are you have to get your RSS feed right. Like on the technical side, your RSS feed has to be perfect because if you're not doing it right, what'll end up happening is you make a mistake, you update that file and people will redownload all the podcasts without even noticing. And that'll cause- Then they'll be pissed off. And they'll be pissed off because they just- Just say it. Unhappy customers, they'll be pissed off. You want it to be right. You want your logo to be correct. You want your title to be correct. You want it to be working right from the get go. But at the end of the day, if your content is good enough, if somebody wants to listen to your content, they will get it in any way you have it. But our main goal has been to make sure that it's available everywhere that people are listening. So putting content everywhere in the hands of users is great. Any device, any time, anywhere, real time. What new things are out there that you guys like that you're looking at from a distribution standpoint? Social networks, Twitter, Facebook, LinkedIn, Splunk, Sprinkler, this, that. It is a zillion networks out there. What are you guys finding? Or is it some areas just shoot it out there and see what happens? Well, I mean, Splunk has definitely been a really big help. Getting all the log information in there and then getting the reports, especially we have a couple different pieces of software working at the same time and trying to get all of that together, Splunk has made it possible. We weren't able to do it before. Before it was all, you know, export these files, put an Excel, modify it and then get a number. And Splunk has just made that so much easier. Yeah, you got your weekends free now. Yeah. Basically, right? Yeah. I mean, the old days you have to kind of, a lot of manual work, a lot of brute force. Yeah, and it's the thing that, I mean, my adage has always been, automate the things that you can automate. Pretty much automate everything. So you can take the time with the things that, you know, you can think about things. Yeah. You can decide things that you can't automate, necessarily. Yeah, the creative process. Yeah. Creating good content. Yeah. There's an art behind it. Yeah. Creating fine jewelry. All right, well, we really appreciate your insights on the inner workings of your opportunity there. And again, any final thoughts on this event? Share the folks that are watching that aren't here. What's the vibe? What's going on at Splunk event? What's it like here? I mean, it feels really busy. There's a lot of stuff to do. There are more sessions to see than I'm able to see. I definitely am going to have to watch some videos. The session I've liked the most was the first session of best practices and better practices. They filled the room out and then some. And that was a really useful session. There's a lot of really key features that, or a lot of key tips on how to make Splunk searches more efficient and just work better and provide what you want. Are you on Twitter at all? Not really. No, I really can't say. Can't plug the Twitter. Alex Giddelson with the American public media doing great stuff. Obviously making sure the numbers can be reflected so they can get more podcasts out, get the justification, which is always key. Thanks for sharing your insights and the data here on theCUBE. This is theCUBE. We'll be right back with more after this short break.