#109: RAD Podcast Analytics with Stacey Goers from NPR

[music]

00:04 Speaker 1: Welcome to the Digital Analytics Power Hour. Tim, Michael, Moe and the occasional guest discussing digital analytics issues of the day. Find them on Facebook at Facebook.com/analyticshour and their website analyticshour.io. And now the Digital Analytics Power Hour.

[music]

00:28 Michael Helbling: Hi everyone, welcome to the Digital Analytics Power Hour. This is Episode 109. Did you know that you’re pretty cool? And no, I don’t mean that in that generic all of us are special kind of way, I mean, you listen to podcasts and right now, podcasts are cool. It turns out that this podcast is part of a wave of podcasting started many years ago and now millions of people consume information in this way. But you know what they say, well, they don’t really say but I say to paraphrase something in the New Testament, wherever two or three desirable demographics are gathered together, paying attention, there will the advertisers be in their midst. And advertisers always would like proof that their dollars are not spent in vain. Not only that, but organizations with a need to identify how podcasts are being used so they can make them better also need information and data. Hey Moe, do you feel cool today?

01:32 Moe Kiss: I feel pretty cool from a coolness perspective, but pretty warm from it being summer in Australia perspective.

01:38 MH: Oh, summer.

01:39 Tim Wilson: It’s a brutal record heat wave in Australia this summer right.

01:43 MH: Oh, wow. Well hopefully, this podcast and its glow of coolness can help bring the heat down for you a little bit. [chuckle] What about you, Tim? Everybody knows you’re cool but do you feel cool?

01:57 TW: Fine. If I open a window, since it’s winter in the northern hemisphere, I feel cool.

02:03 MH: Hi, I’m Michael and I had to turn on the air conditioning today in Georgia so I’m feeling pretty cool. Alright, but we are talking about today podcast measurement and some new exciting stuff that’s happening, just in case you weren’t aware from that intro what was going on. Well, we needed a guest and so we brought in the expert. So, Stacey Goers is the Product Manager for podcasts at NPR and she and her team have developed RAD, Remote Audio Data, it’s a whole new measurement methodology. Prior to NPR, she’s also held multiple roles as CQ Roll Call, but today she’s our guest. Welcome to the show, Stacey.

02:44 Stacey Goers: Hi, great to be here.

02:46 MH: Awesome. Well, hey, to kick us off, why don’t you just give us a background on RAD and how you got there and then we’ll take it from there.

02:54 TW: And if you get an insert an NPR personality reference, drop in a Sam Sanders anecdote or… Every third sentence, I’d appreciate it.

[laughter]

03:05 SG: It’s power at the Goers. Power of the Goers.

03:07 TW: Emma Keith likes green tea or something like that, that would be awesome too.

03:11 SG: No problem, I’ll keep sprinkling them in.

03:14 MH: Tim will pay you in cryptocurrencies for any autographs you could gather for him.

[laughter]

03:21 SG: I love how lucrative public radio is.

[laughter]

03:28 TW: That’s right. That’s a tweet worthy statement, right there.

03:30 MH: Okay, come on let’s get back into it.

03:33 SG: Yeah, so yes, it’s great to be here and talk about Remote Audio Data. It is the methodology that NPR has been working on developing with others in the podcast space for close to two years now, if not a little longer. But the whole start of it really came with the currency of podcasting is downloads. And while downloads are wonderful and great, they certainly do not tell us the entirety of the story. And podcast technology, very much has room for growth and development and there’s many eager smart minds in the space, including some here and when we started to talk with others about how to get something beyond the download or something that was scalable, to actually measure listening and to be be able to understand when someone started to listen to something and then dropped off. When somebody listened to an ad break, when somebody listened entirely through to every episode of every podcast. And that’s kinda the genesis of why we started working on Remote Audio Data as an idea in the analytic space.

04:40 TW: So how does it work?

04:41 SG: Yes, so podcasting is obviously an RSS feed and their media files infinite and your media file has certain components within the ID3 tag of the MP3 or MP4. For most, it’s your title, your artwork, those very basic metadata. But what RAD does is within the comment section of the ID3 tag, a publisher marks certain points in that audio file, so those points can be anything from every five seconds, to interview Q points, to chapters, to advertising, sponsorship breaks whatever they want to have an understanding of the progression of the listening of the audio file.

05:22 TW: Are they designating them by timestamp, even if they describe it in some other way, but they’re having… The markers have to be at this point in the audio file?

05:31 SG: Right, exactly.

05:31 TW: Okay.

05:32 SG: Yes, very good point. So RAD is a three-legged stool, you have your file and then you have the podcasting app that you use that consumers use to listen. If the podcasting applications are configured to read and understand these RAD tags, then they send this information to an analytic standpoint, just year round that the publisher designates to be able to collect all this information. And really what you’re understanding is not who is listening to the audio file. Not that Josh Moore was listening to it, but rather that a person, a thing has consumed this content at a certain point. And the mobile app is configured to understand within the progression of listening of the file if a listener has hit these cue points, have they hit these mile markers that the publisher has designated? And then the mobile app wraps that information up and then sends it off to the analytic server, to say that this file, this MP3 hit these six RAD tags that you published or marked in your file. And the end goal for a publisher then is to have all this information in one consolidated space. To not be looking in different spots, different methodologies to try to understand the listening of the podcast and really to kind of own the narrative and own the progression of folks interacting and engaging with their contents.

06:54 TW: So just to make sure I’m understanding in kind of a web analytics parallel, you’ve got the W3C standards that browsers comply with and so they’re gonna execute JavaScript. So it seems like mobile apps are kind of analogous to a web browser for web analytics. The tag is a very, very simplified version, it’s not JavaScript it’s tagging the file. That analytics endpoint, there is not a defined analytic server so that’s basically making a call out to a server that’s building a log file and then for now, who knows if somebody at some point will provide something like that, but then you’re having to mine what is essentially a log file that says, “We got all these requests and we’re parsing out the data in those requests to do our reporting as a publisher, is that right?

07:47 SG: Yeah, yeah that’s really it. And at each one of those elements, there are those certain complexities but the standpoints, the whole process itself is pretty simple and straightforward if you’re already working in podcasting or already working with download data. And the struggle right now for publishers is I go into Apple analytics and they give me listening data within their dashboard. It’s recently exportable which is wonderful and great, but that’s just within their system. And then I go ahead and I log into Spotify and they give very comparable streaming and that kind of information and then I log into this other system, and I look into this other place and we’re starting to get to a point where we’re comparing apples to oranges or you have very smart individuals who are pulling this information together on behalf of publishers. Whereas if we can all have everything that is truly what me as creative of a podcast wants to measure and be able to understand and have each spot be able to tell me what has happened, it’s like that’s the holy grail, and that would give us so much more information on a very anonymized and lack of a better a word safe way.

08:58 MK: And so I’m prolific skipper when I listen to podcasts, and I often will fast-forward bits. How exactly are you tackling that?

09:07 SG: Yeah, that’s a great question. So what we did, NPR released mobile SDKs for the mobile app developers to be able to look and understand and implement their end of this RAD cycle. And it’s the same as what we’re using in NPR One. When we were building it we realized that it would be the most advantageous to open source it as well, which is really exciting. But within those SDKs that tells the mobile app developer what to do and what not to do and their configurations like batch sending and offline listening and all those things that they care about. But when you get to something like skipping or fast-forwarding or some of those other actions, I mean the mobile app is really just looking to see if you’ve hit those points. And it’s almost like a yes, no. And if you go all the way down the end of the pipeline, and you listen to the end of the podcast, then you just hit those spots and not everything in the middle. And yeah, it can be set up to be able to understand those kinds of nuances and listener behavior.

10:03 TW: Which again, it sounds very similar to when you have a video, if you’re tracking a video player in web analytics, and you’re kinda stuck ’cause you wind up saying capture these milestones, it’s just a tough thing to capture.

10:17 SG: Yeah, yeah.

10:18 MK: And so I did RAD come about because I guess the leadership and the business felt that the download metric wasn’t sufficient or was it something that was driven by the content producers? Where did the I guess, the drive come from from a business perspective?

10:33 SG: Yeah, absolutely. So the currency of the industry right now is through the download and download standards are great and wonderful, and we’re still operating on that. RAD is early. We’ve only started really working on this externally in a very public forum within the last handful of months. So we’re all operating and doing our business on the basis of downloads right now, but NPR has our flagship app, NPR One, and when we released that we realized within the own and operated system, that we have this wealth of information as to user behavior; when people were dropping off, how long they were listening, what was really engaging. And it actually has really driven how we pilot and run some of our shows, but then also just the structure of them. A very tangible example is the NPR Politics Podcast. They used to when they were starting to ramp up the development was they would open it up with every single host giving his or her name and what they were doing and kind of having this long ramp up to get to the juice of the episode and listeners weren’t as excited for that.

11:37 SG: Whereas if you started right away with a core of content and then started to introduce everybody as you went along, that had much higher engagement at the beginning. And we learned very fast in NPR One that the first, the beginning of a podcast is paramount. That’s when you hook someone, that’s where you keep them and this really has helped us perfect so many of our shows which is exciting and great. And then secondly, as well, from a parallel way, as it gets from the business side, you have advertisers who are very used to other worlds of the advertising and sponsorship where they can directly attribute how users are interacting or reacting to their content and their spots. And we knew that we wanted to start to get to that place in that space, but really be smart and really be strategic and really be thoughtful about how we get there. So these things kinda came in parallel and then once we were so excited with what we could get with NPR One and then Apple started to ramp up their analytics offerings pretty close to the same time, it was like, “Oh, well, what if we had this everywhere? What would that mean? How could that work?” And that’s really where those things started to get rolling.

12:47 TW: So if I hooked up my phone through Charles Proxy or something and was listening to Tamara Keith and Domenico Montanaro and Ron Elving, [laughter] I can actually watch presumably and actually see as I’m listening see calls get sent to the NPR’s analytics service end point.

13:08 SG: Eventually, yeah, you could see. When I was… We have a test application for how we are developing RAD and that’s precisely how I was testing it. The mobile apps will likely batch these up and send them in the calls that are more friendly for their battery life. But yeah, essentially, that’s what it is. It’s not though necessarily purely live. I’m not staring at a dashboard and getting live data in that way. It is truly going to something at a log basis.

13:35 TW: We’d like to get real-time reporting. Good lord. How many…

13:40 MK: Kill me.

[chuckle]

13:42 TW: That’s an inside joke, yeah.

13:44 MH: So it’s in NPR One now, is it in any other listening apps yet or…

13:51 SG: It’s too early. We just started. We released these mobile SDKs and we had a little press announcement in December and right now, beginning of February, we have a group of podcasting colleagues, industry stakeholders who have said that, “Yes, we love this. We’ll implement this at some point in 2019.” Which is really exciting. If you use… Actually, I take that back. If you use Hindenburg as a service for your audio, they have the way to write RAD text which is exciting. But we are working with others and talking to others in the space and seeing where it makes sense to fit on their workflows. Podcasting is a small great community. We’ve been able to talk with so many exciting people about this along the way. NPR very much sees ourselves as the facilitator in this. Even though we put these SDKs out there and we put a spec and there’s a basic website that has all this information at rad.npr.org. We very much want this out there in the podcasting on-demand audio ecosystem. We aren’t selling this, we’re not packaging this up, we’re not making a profit off of it, rather we’re really looking to be able to see how can we get the industry and forward in a way that makes sense for all of us.

15:07 MH: So what about… Obviously there’s a lot of concerns these days about user privacy and those kinds of things, how does RAD kind of interact with sort of privacy and sort of user information?

15:21 SG: Yeah, that’s a great question. So right now, when you listen to a podcast and you have the idea of a download, when that is sent, you have the basic internet information, your IP address, your user agent if the mobile was configured for it, some of that basic things and all we know that IP address are sometimes a little rough with your driving. You’re pinging different towers if you’re using VPN. But it’s more or less to be able to understand what is true traffic versus what is a botted traffic. And within podcasting, the Interactive Advertising Bureau has a certain degree of standards. They really tightened those back up in this fall which is actually great to redefine the idea of what is a download. I mean, that something is within 24 hours, that it has a certain byte request of the file and really to make sure we’re operating on a solid foundation, But then what RAD is doing, RAD is not looking to see of a specific user. It’s not gathering extra information, it’s just depending upon that kind of file transfer request and it can perhaps, more importantly, mobile apps need to adhere to GDPR, analytics providers need to adhere to GDPR and other privacy rules and regulations so it’s very much at looking to them to make sure that they’re handling all that in the way that’s appropriate.

16:41 MK: So how does that work if you’re user… So let’s say you start listening to a podcast in the morning and then you pick it up again in the afternoon. Are you guys, I guess, stitching that together or how are you handling that?

16:56 SG: So in the current spec for RAD, we have a idea of a 24-hour session, so if you start listening at 8:00 AM and you finish at 8:00 PM. Morning commute, drive home whatever, that is considered one session and we’ll take that as one.

17:10 TW: But how is that getting stitched together? Is it on the app to say is the app passing that this is the same person who’s listening?

17:17 SG: More or less, yes. So look to see, this is how long we hold a session for within the configuration. But that’s something that can be adjusted, that’s something that some others have thought about, well, perhaps if that’s longer should that be persisted perhaps not within 24 hours. That’s where we put something out there as the first but I would love to have analytics minds.

[laughter]

17:39 TW: We’re gonna have to know what’s the definition of a session, all over again ’cause 10 years of that and web analytics wasn’t enough okay.

17:47 MH: That’s right, is that a weapon or a unique listener?

[laughter]

17:53 SG: So yeah, and what I’m really… And what’s one of the reasons I was really excited to talk to you all was one of the things is to really have those who are so savvy with analytics, like dive into some of this and give us feedback. We have great colleagues here who have been extraordinarily influential and involved with it, but I see a space to have so much more discussion and that’s something we’re really excited about.

18:15 MH: No, that’s really interesting. So you probably saw on the news recently that Spotify has purchased a couple of podcast-specific companies like Gimlet Media and another one.

18:25 TW: Anchor.

18:26 MH: Anchor, thank you. I knew you would know Tim. [chuckle] And that, it’s really interesting, ’cause if I look at it from sort of like a user perspective, Gimlet media on their own probably didn’t have any idea who I was. But if I have a Spotify account and I’m listening to Gimlet media podcasts now, that’s gonna be tied to my Spotify account. And so, if they for instance leverage that, playing there’s a big user connection that these companies are gonna be able to do, and I wonder at sort of you guys are at the very base level of that, but I do think it puts a responsibility on the people who are creating the apps to kinda think through what they’re gonna do with user data. And we’ve all been burned before by Facebook and others, right? So it’s definitely an issue.

19:15 SG: Yeah, the landscape is definitely changing fast and in podcasting, and that’s one of the reasons why we were really happy just to be able to see how many we could talk with this, get everybody’s feedback and see how we can get something that fit as many needs, and then very consciously put it out there as open and saying, that this is something that we would love to see built out and give your feedback go ahead and modify this in a way that makes sense for you. That’s for us was really what we love to see with RAD.

19:48 MK: In terms of the actual, I guess, almost like tagging, how does that happen episode to episode? Because surely there’s different markers for each episode or different points that the content creators want their listeners to hear.

20:03 SG: Yeah, so we’re working on this internally right now, so it’s definitely something that we’re trying to figure out what makes the most sense for our workflow ourselves. A lot of this were actually… And I think will be the case for about the medium to large size podcasters will actually be relying on the ad server technology themselves. A lot of folks use dynamic pre-roll or dynamic mid-roll or post-roll or something of that and these companies have certainly been talking to us and are well savvy and aware of how this can be incorporated in some of their systems which is exciting. But then at a small point of view we released an open source tool radeditor.io, where you can go ahead, upload an mp3, add these ID three tags and download it, it was necessary for our testing so that I was manually editing JASON and sending it off to audio engineers and that was just not a sustainable workflow.

[laughter]

20:55 SG: So this really serve like that immediate need, but it’s a really nice way to start to play around with it. And I think every publisher right now, the podcasting landscape in terms of creating your audio files is really, there’s such a plethora of options and I think that there is something that could be worked for every show, whether you’re a big company or someone’s small, who just uploads something and downloads it and it works fine. So we’re still exploring that but it’s definitely, there’s definitely ways to do it pretty seamlessly.

21:23 TW: Yeah, it’s really, it will be fascinating to see to what extent things like natural language processing and things like that, could sort of, integrate with automation of some of that tagging and things like that.

21:35 MH: You’re thinking it’s gonna be like this automated that you’re gonna get tags every time I start talking to see how much the fall off, the fall off that was?

21:43 TW: Tim talks and then it disappears.

21:44 MH: It would be like if Tim talks for more than 30 seconds. What’s the drop off after that?

[laughter]

21:50 TW: How many times have I talked for a less than 30 seconds?

21:54 MH: Well then, so we’d have a lot tags is what I’m saying.

[laughter]
[chuckle]

22:00 TW: Oh man, well, I was sort of thinking about NPR specifically has a unique place, and this is kinda neat where you guys fit into the overall industry, you’re kind of the only ones that could really pull this off in a certain sense. And so, look, talk about that a little bit just in terms of your perspective on the overall industry and where you guys fit into it.

22:27 S1: Yeah, thank you for that. We felt very excited to dive into this. This was something that from my department at NPR, we just kinda geeked out about which was fun and great. And I felt that that was a really fun opportunity for us to be able to be a leader in the audio space in that way, to spend the time to try to solve this problem. We obviously have a strong vested interest in the success of podcasting as do our member stations as to figuring out the best way to give them the best information as well. And when we started to kind of see if there was a there, there in this, it was something where when we could come at the table, we could kind of throw some things out there and then see how others could run with it. We held a series of internal working groups amongst those in podcasting tech were interested, just very large Google Hangouts, but they worked to talk about some of these really interesting and thorny details. We hosted a number of folks at NPR headquarters about a year ago now, to white board, what this would mean, we brought down our mobile devs, kind of just talk through. So it’s been really fun to be able to just kind of charge head with it, and really figure out what’s the next generation for these kind of analytics and audio.

23:45 MK: So what do you think is… ’cause in my mind, my dream state is that we can include podcast ads in say like our attribution model or something like that, which would be really cool, but what do you think is the direction that podcast analytics is gonna go in or what’s next for RAD?

24:03 SG: In terms of RAD very specifically, we’re just head down focused on a mobilization right now. Because there’s so many parties involved with it, it works perfectly when more and more adopt and work it out. So what our goal right now is just to talk to as many folks as we possibly can and make sure that everybody understands this technology. It’s not hard for a lot of folks to implement or to bring into their work. But a lot of podcasting and a lot of tech companies are not large, so we totally respect and understand that, and it’s an ask. But overall, we’re just excited to be able to get smarts, just complete information. A lot of times we’re chasing a user agent or stringing back information that… Like how it performed on this platform versus that platform. And I think right now we’re just very looking… Very much looking to be able to just understand the whole picture and be able to just move forward with that. We’re excited, there’s a lot, there’s a lot of improvement. Podcasting tech, the glorious RSS feed is great, but there’s a lot more than… In this space that we can do and we’re excited to kind of see what would be next.

25:15 TW: A little minor rant, RSS completely… It’s just like web analytics is hacking the Internet and RSS is being… I don’t know if that’s our… To our benefit or our detriment that we wind up being so hungry for data that we come up with clever ways to hack technology that was never really designed for that, but we had your colleague, Steve Mulder, on the last episode talking about smart speakers and you can listen to podcast through smart speaker. So when it comes to podcast players, do Alexa and Google Home and those kind of fall in as potential target apps as well? Would they be working the same way or is that technology work in a very different way that RAD wouldn’t work as well?

26:04 SG: Yeah, it’s a good question. Some do, most that do not re-host audio, like this is a really nice way to be able to work it. A lot of times if you’re creating a skill or you’re working even within the Google world, it is RSS feed base, which is exciting. So it’s exactly the same kind of formula for that. So yeah, so that technology is certainly not far removed.

26:26 MH: So when it comes to the other… To the podcast to the stitches and the Spotify’s and the Apples or smaller ones, what is… You sort of said it’s a small community and there’s just kind of a recognition that better data is better, but it feels like that’s this piece where the app has to implement and then commit to maintaining a part of their application which you even alluded to it, it’s not real… They gotta consider their battery life and kinda just like with mobile analytics, batch it up if somebody’s off-line, what’s kind of the motivations for for those players, to say This is gonna be, there’s an opportunity cost if we invest in adding this to our app, what are kind of the big drivers for them to implement?

27:17 SG: Yeah, first off it’s a real light weight-like framework, it’s extraordinarily light weight, which is exciting. I know work is work, but it’s a pretty quick implementation. And that was one of our main goals as well. Just to not have this be a huge thing, just to have this be quick and easy and lack of better words, but really podcasting applications is very much understands that there is a community, a group of content creators who rely on this type of information to make decisions who have sponsorship who have dynamic sponsorship who need this kind of information back in a smart way. We’ve seen a lot of great movement, even from your large companies and making sure that publisher tools and publisher interfaces and the way to do things within those who submit RSS feeds is smart and makes sense then that’s something that we love, and we wanna continue to see and we kind of see RAD falling into that spectrum, as something that publishers are looking for this information, that this is something that sustained those content creators and those sponsorship and those companies that want to get into this space keeps and it buoys that ecosystem.

28:31 MH: So how is RAD licensed? So people can start to take it and put it into their application. So I assume you have sort of some sort of licensing structure.

28:40 SG: Yeah, I think we’re just an apache kind of pretty open with that, so that, yeah, it’s all within RADs at npr.org and there’s links to just a text back that is still being modified and worked on as others give feedback to that which is exciting. And then, the tag editor is at that location as well, and the SDKs or just the framework for iOS and Android and then an email address to reach out if you have questions or wanting to get involved, if anything, I’m the recipient of those emails so send them forward.

[laughter]

29:20 TW: So if you wanna get in touch with Stacey after the show now you know.

29:27 TW: All that is there, and we’re really hope that everyone can kind of interact and understand this and become educated with it.

29:33 MK: Okay, just one last question on the business. I feel like I’m always like dragging people back to this, [chuckle] but as you started to roll out RAD, what’s been the response? Are people just so hungry for like give me more, give me more. Or just the downloads was there for so long that people are so thankful they’ve got something a little bit more like how are content creators reacting to… Hey, this content sucks, or this content has made it. How are people reacting?

30:03 SG: No, it’s a good question, so because we don’t have RAD at scale yet. Some of the good examples or something even of our NPR One or Apple analytics. Before Apple analytics was downloadable or you could export all the information, we had to share a dashboard and we were manually giving things to our producers and that was certainly something that was not fun nor sustainable. My colleague, Steve Mulder who’s on this knows probably his department knows that well.

30:29 TW: I’ve screen-captured our Apple Analytics podcast for Moe and Michael because that was the last time I looked. It wasn’t downloadable.

30:40 SG: Yeah, exactly, so for NPR One we have these individual Dashboard that we give our content producers and they are so excited to be able to see this kind of information, and even within what we give out to our teams on a weekly basis like they scour those reports. Very much trying to understand where this audience is, what are the triggers that can influence when we publish or how long our episodes are, or what are those things that really can make it better to give this community what they’re looking for?

31:17 MH: Yeah, I mean if you think about it, Moe, the extensive user research. We did to find the right British accent for the intro of our show. Now you’d be able to complete with tagging.

[laughter]

31:28 SG: Yeah great.

[laughter]

31:29 MH: I cracked myself up. Alright, yeah. I have a lot of thoughts but I feel like we’re gonna have to start heading towards a wrap. One of the things we love to do on this show is go around and do what’s called a last call. And it is just sort of anything you found interesting recently that you find notable. Stacey, you are our guest. Do you have a last call?

31:57 SG: Well, I have to mention podcast. I feel like that is doing my due diligence but it’s an exciting time because NPR just released a new history podcast through line. I just listen to the first episode, actually, today, and it is absolutely wonderful. So I highly recommend that you download and listen to that. I’m also listening to a member station WBEZ out of Chicago has a new podcast, Public Official A, all about Rod Blagojevich which is also really great. So that’s been something I’ve dived into.

32:29 TW: The podcast is great Rod Blagojevich is a mockery.

32:33 MH: Rolberg hey second only to Jim Traphagen in terms of flare. How many episodes of that podcast are dedicated to just his hair? Tim would like to know.

32:45 SG: Not enough. Not enough.

[laughter]

32:47 MH: Not enough.

32:49 SG: And then…

32:49 MH: That’s awesome.

32:50 SG: I know. And then the one out there that I wanted to give a shout out to which I think some of you listeners might be interested in is… It’s Harvard Business Review’s Women at Work. It is a wonderful podcast. They don’t have too many things that are new, but the back catalogue is absolutely fantastic.

33:06 SG: Cool. Oh, fantastic. Oh, I really need a new podcast. So this is great.

33:10 TW: This is so funny, I feel like with Steve last episode, we kinda teased a little bit that you were coming on this episode and with that last call, I’ll say that you’re inadvertently teasing our next episode so well done. It’s like we just have an editorial calendar.

33:23 MH: Mysterious. It’s almost like we planned this stuff out in advance.

[chuckle]

33:31 MH: Speaking of planning things out in advance, Moe, what’s your last call?

33:36 MK: Okay, so I’m working at a company that uses Jango which I’m not very familiar with so if I fab something up. But basically, it’s like a web application written in Python and it creates a whole bunch of reports automatically using SQL and I’m just trying to get my head around it and Mike Robins from Snowflake was completely amazing and pointed me in the direction of a Jango debug toolbar that you can install on Chrome. So when you’re in a page, it actually shows you what are the Jango calls that are going on in the background. So I still need to muster my way through installing it ’cause it’s not super easy, but I’m pretty excited to tackle that.

34:17 MH: Well, I will astutely avoid any kind of Quentin Tarantino movie joke and all that.

[laughter]

34:22 MH: Alright.

34:22 TW: So what’s your last call?

34:27 MH: I’m gonna do mine ’cause I might be stealing yours that I don’t know… ‘Cause it’s something you were talking about in Slack but one of our SEO team threw this article up and I really found it fascinating because I’d really never considered it which is thinking about how search engines think about entities when they actually index and promote listings to you. And so when you think about all the different devices you might conduct a search on, they’re all gonna maybe be doing results just a little differently because they exist to serve different purposes and the search engines are already doing it. Google’s already doing this, probably Amazon and all the others. So I just thought that was really neat because in my head, I just think, “There’s one Google and the results are always sort of the same if not slightly personalized to me.” But actually very different. Anyways, fascinating read about entity-first indexing. So that’s my last call.

35:21 TW: Nice.

35:21 MH: Tim, what about you?

35:23 TW: Well, now I feel like I’m gonna have to do a quick twofer because the genesis of that whole discussion which I had also read that article when it got posted by our colleague was very interesting, but where it came from, what started it was that when I was at Super Week back at the end of January, Fred Pike from Northwoods Web Solutions, loyal listener, actually presented something that some people on his team had built called the digital marketer’s BFF and it’s an entity analyzer. There are two aspects of it. But it’s basically, you can throw in a web page or two webpages for the comparison and it goes through with some machine learning, natural language processing and it spits out the entities that is detected and some of the kind of ratings and assessments of that so…

36:09 TW: And he was saying this is useful for SEO and I’m like, “What? I thought you just had to have a good title tag?” But no. I know it was more than that. But I think that kind of sparked this whole thing. But it’s this is cool little free tool that’s worth checking out from an SEO perspective. But if Moe’s gonna talk about Python, my actual last call is gonna be something that Jim Sterne tweeted which is a post on Medium, called how the BBC visual and data journalism team works with graphics in R. And it’s funny, BBC like NPR, they just published their playbook and their approach and they went through using ggplot and said, “We think we can make a package in a playbook and produce BBC styled visuals using R when that is most expedient to do.” And it kinda talks through their process and how they do it and it was about R so I had to include it as a last call.

37:08 MH: Perfectly acceptable to have R as a last call, Tim, and pretty normal too.

37:15 MK: I’m still loyal to the cause. I’m still loyal.

[chuckle]

37:18 MH: Well, I told you that my son just took his first Python course at the library so he’s getting…

37:25 MK: That’s amazing. I love it.

37:28 MH: Yeah, he’s…

37:28 MK: How old is he?

37:29 MH: He’s 12 so he’s pretty excited. He was able to create a little program that wrote the word potato 20 times.

[laughter]

37:37 MH: And he was really excited to show that to me. I don’t know why the word potato is so special to him, but that must be the Irish part of our family’s history.

[laughter]

37:45 MH: Anyway, this is actually, sorry for that little side tour, but anyways, this has been a really great show and I’m sure as you’ve been listening, you probably got thoughts and ideas and… Stacey, just based on the way you’re talking about it, it sounds like you and your team are very much in the middle of really rolling this out to the world so what a great time for people to interact with both the SDKs that you’ve promoted out on GitHub and all those kinds of things. So if you’re listening and you have interest this, I’d say definitely show some interest and get involved. We’re available on the Measure Slack and also on Twitter. Are you on Twitter, Stacey? Wanna throw out your…

38:28 SG: I am. Stacey Goers, G-O-E-R-S.

38:31 MH: @StaceyGoers, G-O-E-R-S.

38:32 SG: Yep.

38:34 MH: Awesome, thank you so much for coming on the show.

38:37 SG: Thank you.

38:37 MH: It was great having you. It was really exciting to kind of explore what is sort of a brand new capability, methodology, and analytics that’s pretty cool for us and our audience and I’m sure it’s tweaked a lot of interest for our listeners and I’m sure I can join Tim and Moe, my two co-hosts, in suggesting to all of you to get that new data from podcast and use it to keep analyzing.

[music]

39:07 S1: Thanks for listening and don’t forget to join the conversation on Facebook, Twitter, or Measure Slack group. We welcome your comments and questions. Visit us on the web at analyticshour.io, facebook.com/analyticshour, or @AnalyticsHour on Twitter.

39:26 Speaker 6: So smart guys who want to fit in so they’ve made up a term called analytics. Analytics don’t work.

39:33 Speaker 7: Analytics. Oh, my god, what the fuck does that even mean?

39:43 S8: Wow.

39:44 MH: So that’s our last quality check and now, we’re really gonna get going.

39:49 MK: Oh.

[laughter]

39:49 TW: Oh. Jesus Christ.

39:51 MK: That’d be amazing.

39:52 TW: That would have been…

39:54 MH: Well, that’s my value add for the day.

39:54 TW: Holy shit.

39:58 MK: [39:58] __ that was a fairly important value add for the day. Well, played.

40:04 TW: Good lord.

40:06 MH: Alright, bring it together, don’t worry, Tim. And it’s worse I think because it’s NPR because secretly I think Tim would feel like, “Hey, yeah, hire me.”

40:18 MK: I don’t feel like you’ve asked so many questions. Cool your jets, man.

40:23 TW: I’m sensitive.

40:25 MK: No, you’re stupid. I did kinda think that ’cause I talked a few times and everyone just kinda kept talking and I’m like, “I don’t think they can hear me but I’m just gonna wait and see.”

40:36 TW: Just the typical disregard that we normally have for you, right?

40:41 MK: We got this. We got this.

40:43 TW: You’re still very good at what you do.

40:44 MK: Of course, I see is like, “What is this shit show that you guys… “

40:49 SG: No, are you kidding? You’ll be fine. You’re fine.

40:53 MH: I feel like we’re done apologizing to Stacey and she knows who we are now for real. We don’t have to apologize anymore. It makes me feel bad when you’re passive aggressive response to Christmas stuff is way better than anything I turned out ever.

41:09 MK: Are we measuring it correctly though?

41:15 MH: I guess we’ll have to record this episode. Find out, Moe. That’s how important it was to my childhood.

[laughter]

41:24 MH: These two people were at their school dance and they danced together on bicycles. It’s amazing.

41:31 TW: Did you have to look up that it was 1986?

41:33 MH: I did. I IMDB-ed it before the show just so I’d have my facts straight.

41:38 TW: That’s good. That’s good to have. That’s that sort of in-depth show prep we like to see.

41:41 MH: Yeah. Oh, you know, we’re hot messing around here.

41:45 TW: Rock, flag, and grad.

Podcast: Download | Embed

Subscribe: RSS