#100: Listener Questions Answered

To think, it was barely-considered subtle humor that we used two trailing zeros in episode #001. But, despite our best efforts to destroy our reputations or our livers long before we centupled that first show, we failed on both fronts, and we now need that third significant digit! For this special episode, we invited listener questions, and our listeners responded. Some of them blew right past the time limit on their questions, but that’s okay: we blew (slightly) past the one hour mark for the show.

Listeners Whose Questions Were Used

References Made in the Show

Episode Transcript


0:00:04 Announcer: 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.


0:00:27 Michael Helbling: Hi everyone, welcome to the Digital Analytics Power Hour. This is episode 100.


0:00:46 MH: Well, we made it. A hundred episodes and many miles and a ton of whiskey, but it feels great to have the number one explicit analytics podcast on fucking iTunes. Oh yeah.


0:01:01 MH: It all started with a disagreement and that disagreement was with Tim Wilson and Jim Cain. And Jim Cain that was at that time, our other co-host double CEO, but now I think actually that he’s a triple CEO. He’s had a lot of success up there in Canada. Hi Jim. Couldn’t have done it without you.

0:01:19 Tim Wilson: Hi Jim.

0:01:20 MH: And then for a little while it was just Tim and me, until we conducted a worldwide talent search. Came up with our new co-host and a perfect foil, Moe Kiss. Who is not Michelle Kiss, but that may have helped our numbers stateside for a while. And then it is me Michael Helbling, your steady friend and moderator and all around nice analytics guy. So Tim and Moe, welcome.

0:01:48 Moe Kiss: Hey, how’s it going?

0:01:50 TW: Hey it’s… I can’t believe we’re at a 100.

0:01:53 MH: I know, can you believe it? Are you guys ready to start our 100th episode?

0:01:58 MK: Absolutely.

0:01:58 TW: No, I think I’m just gonna quit right now.

0:02:00 MH: Yeah, quit while you’re ahead.


0:02:01 MH: And still undefeated. No, it’s amazing, it’s a huge milestone. And actually if you’re not a regular listener, we did something a little special with this one, the guests and the questions are coming from our audience. So we’ve had a competition, we had a lot of people send in submissions and so what we’re gonna do is we’re gonna answer questions from the audience. And so we’ll play the first clip and we’ll dive right into it. So to kick us off, let’s go to the tape.

0:02:31 Robert Petković: Hi, my name is Robert Petković and I’m a web analyst, from Grey, Croatia. My first question is, what do you think… How significant was the impact of Tim Wilson’s hair to web analytics in year 2018? Now, a serious question for Moe Kiss, what was your most memorable “why” question you ask yourself during analyses and why? Thanks.

0:03:06 TW: So I… Okay, on the hair.


0:03:09 TW: Yeah, I think the hair’s been a huge component of our success, Robert. That’s a really great question. And you know Robert has a background in psychology, so he understands how these little things can have such a huge impact on outcomes.

0:03:24 MH: Well, I…

0:03:25 MK: It’s about the cult following.

0:03:26 MH: Yeah.

0:03:26 TW: Well, one…

0:03:28 MK: Of your hair.

0:03:28 TW: The hair literally came about from a… 22-23 years into marriage, a miscommunication with my wife, and though I could spin it as saying that Eric Feinberg, who used to be kind of the most renowned hair in the industry, has not been as visible of late. And so he has fantastic hair and somebody had to step in to fill the void. But maybe we move on to Robert’s actual question.

0:03:53 MH: Actual question? Okay.

0:03:57 TW: Which is for Moe.

0:03:58 MK: So for me, the most memorable “why” question, it’s funny it actually came up again last night, was when someone in the business asked why we couldn’t move more mobile site users to the app. And legitimately someone asked me this again, last night, and I was like, “Oh God, we’re back at square one.” I think the reason that for me, it’s the most memorable is because it started me down my much talked about path of talking about cross device users and really understanding why were people going to our mobile site. So, for example, we know that a lot more new customers or new users go there. And finally, I guess cracking the, you can’t just build a better app, and then expect that everyone will magically transform over there. So for me, it started basically the last, it was kind of the question that kicked off the last couple of years of work for me, which I’m still super passionate about and still apparently answering, not very well, which is why I just got asked it yesterday.

0:04:57 TW: How much are you having to answer that as an analyst who’s looked at data and how much are you entering that as an analyst who’s been thinking about the customer journey and customer needs, especially since this is coming from Robert and his psychology background?

0:05:12 MK: I think it’s both though, right… The thinking about a user’s needs is the same as saying, “Okay, well why would we have more new users on this platform?” Like they’re kind of intertwined, you could do one without the other, but you wouldn’t be answering the question very well.

0:05:30 TW: Yeah, I mean I think that that to me that’s a great example because it’s… The answer isn’t a 100%, just in the data.

0:05:39 MK: Well, you do have to be careful though. And my favorite catch quote at the moment is “You are not your user”. So you also have to be so careful that when you’re trying to figure out that “Hang on, wait why would there be more new users on the mobile site?” that you don’t start thinking about it as you being the user and then assign all of these reasons for these behaviours that just… You’ve made this huge leap or assumption that isn’t there.

0:06:04 TW: Which is, that’s definitely also a consultancy or agency side where you’re working with a client that has no… You’re not the target audience, then I think I have to be doubly, triply… You’re at least kind of the iconic target demographic, right? So you’ve got… You can be a user. When I was working with Victoria’s Secret, I had to be pretty careful that…


0:06:26 TW: It was not me.

0:06:27 MH: It’s not about you.


0:06:33 TW: So should we throw in memorable ones for ourselves, Michael? Even though it was directed at Moe, or should we just keep chugging?

0:06:40 MH: Sure. Go for it.

0:06:41 TW: Well, [chuckle] I’ll say one that was probably if you literally just went off the most memorable, and I feel like not a year goes by that I don’t bring it up, was probably back in 2001, 2002. SEO was kind of kicking in. And I was in-house at the time. And a product marketing manager who owned the website, and the website we were still learning, figuring out a bunch of stuff. But he had gone in and made some tweaks for this search engine optimization stuff. And he basically had turned an acronym into being a fully written out word.

0:07:15 TW: So a good, legitimate adjustment, but not massive overhaul. And we were still running NetGenesis. So he waited till the monthly report came out. And he came racing into my cubicle and said, “Look, I made those tweaks on this date, and traffic went way up to that landing page. I’ve told everybody. I’ve told the VP of Marketing.” And I was like, “Oh crap. Let me hop into the little micro strategy, Doug.” And it basically turned out that Gomez was pitching us. And so they had asked for what were like some important pages on our site. And then they had basically turned their little load monitoring engine towards it. And this was before that sort of stuff was getting filtered out. It was log file analysis. So basically it was completely junk traffic, which like, I use that as my example any time that it’s been a, if you see something that’s like amazing and surprising, then chances are you just haven’t dug deep enough. We don’t find those sorts of things all that often.

0:08:15 MH: Oh man. That was before correlation wasn’t even causation, right?


0:08:22 MH: Way back in the day.

0:08:24 TW: Put you on the spot? Do you have a memorable one?

0:08:27 MH: I was thinking about this. And it’s really weird. I went through sort of this job transition, where I was dealing with the data that drives our business, that’s Search Discovery. And it was strange because I’ve got data, but I’ve always had a problem with things like timesheets, and tracking your hours, and those kinds of things, just because everyone who ever works in consulting typically has a bad attitude about it. Just a little secret.

0:08:58 TW: I love it!

0:09:00 MH: There are some people who really love it. And I appreciate those people so much more deeply now. But what I realized, my why was sort of as I struggle with that data, I tried to really serve… Use that data to serve the folks at Search Discovery as best as I could. The why was actually around, “Why do this well? Why really get good at the operational aspects of running a consulting practice?” And so that was kind of a big why for me in the last few years. And what was cool about it is, it was in embracing that, that I was actually able to then bring my analyst skill set to bear and actually be able to see in the data the nuances of our business and how to tweak, and make changes, and drive things in the way that I wanted them to go. And so that was a pretty cool learning experience for me, is just the application of the data, and actually embracing it as sort of as an analyst in a whole different area to try to leverage it, to actually drive the business in the right direction. So that was kind of my “why” moment in the most recent. It’s not digital, but it matters.

0:10:11 TW: I like it.

0:10:11 MH: Matters to the lives of analysts.

0:10:14 TW: Certainly a more profound “why” than mine. So…

0:10:17 MH: Yeah. Exactly. Well, you’ll find that a lot Tim, is I’m really… Really…


0:10:23 MH: Alright. Let’s get to the next question.

0:10:25 Justin Goodman: This is Justin Goodman with Analytics Pros, Director of Analytics. We all know that whiskey is the quintessential drink of the quintessential analyst. What I wanted to know is if there’s a specific brand that as Analytics Hour hosts and podcasters in general, you’ve found to really inspire creativity and drive your quintessential podcasting along with that quintessential analytics skills.

0:10:50 MH: Okay. So what folks listening don’t know, is that as we voted on these, it wasn’t always unanimous. But I made sure this one got through because I think it’s the quintessential question.


0:11:03 MH: Justin that is…


0:11:05 TW: ‘Cause it is more quintessential…


0:11:08 MH: The quintessential-ness of this question cannot be under-quintessenced. So first off, I think it’s a good thing to talk about whiskeys, because they do kind of vary from time to time. And so I would say for myself, the best whiskey for the show comes from a lot of testing. Both what gives the voice the most resonance, provides the most smoothness on the palate. No, I’m just kidding. It’s whatever I’ve got in the house at the time.


0:11:37 MH: So… At that though, I would love to say that our first sponsor ever has now joined the show. That’s right. We’re so glad to welcome Buffalo Trace. No, I’m just kidding…


0:11:47 MH: It’s sort of wishful thinking. But I do…

0:11:53 TW: I actually know a guy who works at the agency that does all their marketing.

0:11:57 MH: Hello. This is a huge opportunity.

0:12:00 TW: I was actually just at his wedding, and they had Buffalo Trace, right there.

0:12:04 MH: Yeah, some excellent brands in that, but I do love American whiskeys and ryes, and I would say I have my boss, Mike Gustafson, to thank mostly for that. Prior to working with Mike, and being part of Whiskey Wednesday at Search Discovery, I tended to drink single-malt scotches, that were like on the fruitier side. So Glenlivet, Glenfiddich, those are my go to’s and I still have some nice bottles of those, but I gotta say, I love Blenheim’s winter wheat or wheat whiskey. It’s like the longest answer on the show is gonna be what whiskeys we like…


0:12:42 TW: Hey, that’s all we have time for folks.

0:12:44 MH: Blanton’s really good. I don’t know, lots of different kinds.

0:12:48 TW: That’s a Buffalo Trace.

0:12:50 MH: Yeah, that’s a Buffalo Trace, so that’s why I like Buffalo Trace. So they’ve couple I…

0:12:52 TW: Moe, you had some fancy Japanese whiskey at one point.

0:12:56 MK: Mmm, that’s still my favorite. Yeah. The Japanese whiskeys is what actually got me into whiskey, and then I’ve been able to go backwards and retreat into some more classic stuff. But I’m still… I always offend people when I say that I hate peaty whiskey. I think that’s why I like the Japanese, ’cause it’s so like delicate, and light and packs a bloody punch.

0:13:19 TW: That’s so funny ’cause I would have said Lagavulin or Laphroaig. I’m a, I like the… I like peaty whiskey.

0:13:24 MK: Yeah, I hate Laphroaig, ugh. So the whiskey that I love is called Nikka By The Barrel, which is a Japanese whiskey but I’ll pretty much drink anything Japanese. And I’ll drink some non-Japanese stuff, but I’m not… Not Laphroaig, ugh.

0:13:36 TW: Well, I don’t always drink whiskey.


0:13:40 TW: As a matter of fact, I’m drinking Guinness, right now… I’m spilling Guinness all over my desk right now. I’m kind of whatever someone hands me. I would like to think I like variety, which is maybe how I like my analysis as well. I mean, I will go with the smooth stuff, I will go with something that has a good story. I actually, anyone who has a good story about a whiskey where they had it or a rum, or any sort of alcohol, I might have a problem actually. If you hand it to me, I’ll drink it.

0:14:17 MH: We’re learning that to be the quintessential analyst means to be a pretty indiscriminate drinker.

0:14:26 TW: Yes.


0:14:27 MH: Yeah.

0:14:28 TW: I think we have a… Yeah, done this one into the ground. Thanks, Justin.

0:14:33 MH: Yeah, Justin, that was awesome. All right, let’s get to the next question.

0:14:38 Matt Gershoff: Hi, this is Matt Gershoff. This question is inspired by a chat I had with Moe on the Measure Slack. I gave it to her as a little follow up homework to explain a trick to get a rough estimate for confidence intervals. But it doesn’t look like she’s gonna get back to me. So as a nod to Car Talk, I thought it might be fun as a sort of puzzler. So, here’s the puzzler. Tim and Michael are about to present results of a marketing campaign to their boss Moe. Moe is a real stickler about always including uncertainty measures with any results. Right before walking into the meeting, Tim asks Michael, “You know how angry Moe gets when we don’t include confidence intervals, right? Please tell me you calculated the confidence intervals along with our conversion rate estimates.” Michael replies, “No sorry, I totally procrastinated last night. I wound up binge watching House Hunters.” Tim replies, “Again? Uh, okay, well, she’s waiting for us now. We don’t have the time to calculate it exactly. Just use this trick. Divide one by the square root of the sample size. You can just use that to calculate the upper and lower bounds of the confidence intervals. That will give her good conservative estimates for 95% confidence intervals.” To which Michael replies, “It will? Why?” Please explain to Michael why this simple heuristic is a reasonable upper bound on the 95% confidence interval for proportions and conversion rates. Thanks.

0:16:00 TW: Fuck, I need a drink.

0:16:00 MH: Okay, so first off, Matt has made a couple of grave errors. Because in prepping for this meeting, I would have understated the personality and strengths of my boss Moe, and would have kind of understood from my presentation what I needed to do with or without those confidence intervals, and she would have full confidence in me, not because my data was correct but because I was emotionally intelligent and was actively listening to what her concerns were as a business person.

0:16:27 MK: Ah nice.


0:16:30 MH: Yeah.

0:16:31 TW: Given that…

0:16:31 MH: So, now divide that by the square root of one.


0:16:36 MH: Okay, what’s the real answer to…

0:16:40 TW: Well, I think I can give the answer which will then, Michael just by the way, I need like half a day off ’cause I’m sure Matt will be hitting me up and will explain to me how I rounded a few too many corners on this, but I did not know this off the top of my head. We do have a Google Sheet that Matt actually put together ’cause he’s Matt Gershoff. So he not only delivered a question, he delivered a companion workbook. I will try to clean that up to make it a little easier to follow, but as I dug into it… So this has to do with the standard deviation of proportions which is what we’re dealing with in this problem.

0:17:15 TW: So the standard deviation of a proportion, like the proportion of visitors who convert, is the square root of the proportion who converted times the square root of the proportion who did not convert, which is just one minus the proportion that converted. So that maxes out at 0.5 because if you have 50% conversion that’s 0.5 times 0.5 which is 0.25 and the square root of 0.25 is 0.5, so it falls off kind of slowly so it only drops to like 0.4 if the conversion rate goes down to 20% or up to 80%. So we simply, and this is where I get to use simply in the Matt Gershoff perspective in this, some people have already gone to sleep and are, like, lost. We simply divide that value by the square root of the sample size to get the standard error. To get a 95% confidence interval, we need to go plus or minus two standard errors from the mean. So if we have just 0.5 times two for our numerator which is one, and we’re again dividing by the square root of the sample size, we get the confidence interval for a 50% conversion rate, and any other conversion rate will definitionally have a lower confidence interval. So for a really, really low or a really, really high conversion rate, this short cuts returning a result that’s much higher than the actual confidence interval. But that will at least buy Michael some time to get his ass back to his desk and do a proper analysis.

0:18:37 MH: Yes. We did forget at the tip of this question to tell everyone listening to either speed really fast or slow way down.


0:18:48 MH: ‘Cause that was a lot. Oh my gosh, that was so much. Moe, do you agree or disagree?

0:18:57 MK: So…


0:19:00 MK: I’m trying to listen to Tim and follow the notes that Matt gave me.

0:19:05 TW: Oh, wait, Matt? You had notes from Matt and you didn’t share those with me?

0:19:08 MK: Yeah.

0:19:09 TW: Did I follow it… Did I… Was I even remotely close?

0:19:12 MK: When he gives you homework, he gives you some notes to go with it but…

0:19:19 TW: Son of a bitch.

0:19:19 MK: I feel like this is gonna be like a life time of now just letting Matt Gershoff down because I just keep rereading it, and then going, “Uh-huh?” But this is gonna help me answer another question later. Sorry, you said divided. Oh yeah, okay, no I stand corrected.

0:19:34 TW: Either way you’re dividing by the square root of the sample size. So that kinda goes away.

0:19:38 MK: Yup, yup, yup. Okay. I agree, you are correct. Well, correct-ish. Yeah.

0:19:41 TW: Well, no, don’t jump into that boat with me, ’cause then you’ll be… You’ll be back with Matt again. It’ll be a group chat.

0:19:47 MK: No, but when Matt decides to ping you about this, can he just keep me on CC, so that I can just listen to you guys talk about it, and then maybe that will solidify my understanding.


0:20:00 TW: Oh sure. We love you, Matt.

0:20:04 MH: Have we… Or our audience been #MattGershoffed at this point or is it all of us?

0:20:10 TW: Maybe they were, they were #MattGershoffed by proxy.

0:20:14 MH: By proxy. That’s like…

0:20:14 MK: Yeah, I feel like it’s by proxy for sure.

0:20:16 MH: Okay, alright well let’s get things back on track with our next question.

0:20:22 Jamarius Taylor: This is Jamarius Taylor. So everybody on this panel here is a seasoned analyst. As a young analyst, I just kinda wanna know a few things that you kind of wish you would have known when you first got into the game. So if you could answer that, that’d be great. And I’d also like to know when Tim Wilson’s gonna cut his hair. Alright, thanks.

0:20:40 TW: What’s with all these twofer questions?

0:20:42 MH: I think people just had 30 seconds, so they…

0:20:44 MK: They learned from you and your last poll…


0:20:46 TW: Oh snap! Oh snap! .

0:20:51 MH: Alright, awesome questions, Jamarius, and I’ll answer them in reverse order. I don’t think Tim Wilson was ever gonna cut his hair, I believe it’s a secret of his analytical strengths. So that’s the only reason why he’s great at Data Science, because growing his hair out and his Hour knowledge happened all at the same time. So he’s sort of like…

0:21:10 TW: Samson?

0:21:11 MH: The Samson of R. So he would lose all of his R skills if he ever cut his hair. It’s in the Bible, check it out.

0:21:19 TW: The truth is, this is a one time experiment, so I have to get it to the point that I’m like, “Yep, I’m done.”

0:21:24 MH: But it’s been a long… You’ve been running this experiment for… At what point will you reach statistical signi…

0:21:30 MK: Most of the time that I’ve known you I think…

0:21:32 MH: At what point would you read statistical significance?

0:21:35 TW: I track things down. It’s been two or three years. It’s not been that long. So…

0:21:40 MK: No, I feel like when I met you, you had just started this pilgrimage.

0:21:43 TW: I can track down photographic evidence, and we can… ‘Cause I was at the… That was at the…

0:21:51 MK: Loves Data Conference. Yeah.

0:21:52 TW: Loves Data Conference in Sydney in 2015?

0:21:56 MK: ’16? ’15-’16?

0:21:56 TW: ’15.

0:21:57 MK: ’15.

0:22:00 TW: ’15.

0:22:01 MH: So…

0:22:02 TW: But actually the… Let’s go to the real question.

0:22:04 MH: Yeah, the real question and actually…

0:22:05 TW: There are no more questions about my hair.

0:22:08 MH: This actually tees up something really great because Moe, I remember something you’ve written about this, so I’ll let you take it from here.

0:22:16 TW: Moe.

0:22:16 MK: You need to give me a direction of…

0:22:18 TW: You don’t even have a hint is what she’s…

0:22:20 MK: I thought you were like…

0:22:21 MH: Didn’t you write a blog post about this?

0:22:23 MK: Oh yes!

0:22:24 MH: Yeah.

0:22:24 MK: Oh God, yes.

0:22:24 TW: That was like one of your first blogposts.

0:22:27 MK: It was my first blog post but I’ve changed my mind.

0:22:29 TW: Oh that’s so rare.

0:22:30 MK: Which… Actually, Matt’s question is a really good segue into this. So…

0:22:36 TW: Jesus Christ, where is this going?

0:22:40 MK: I do wish before my first ever digital analytics job that I knew what SEM meant, but hey, that’s… I learnt that on the fly. So, that’s okay. I wish, and I was actually talking about this on the weekend, I wish that I had gone back, or I kept my math skills up, like I wish I’d gone and done like a stats course or even just like some basic math. So I was talking about, I did double maths and physics the whole way through school and basically didn’t use that skill at all for probably eight or 10 years. And then when you’re trying to pick it back up, I just wish I’d maintained some base level of usage.

0:23:17 TW: I think I’ve said that, told that anecdote maybe on the show, that when Matt tells people that they should go really take a good linear algebra course and then he points them to a video of a professor teaching his Linear Algebra course. And that was the Linear Algebra course I took in 1993, and completely… It was not hard, at the time it was kind of cool, but yeah, go figure, fast forward a decade and that wasn’t really sticking with me. And it’s only coming back very, very slow.

0:23:48 MH: All right, so your first one was you wish you had kept up with your math and physics skills, okay. That’s really good.

0:23:56 TW: What do you have?

0:23:57 MH: Well, so I… As I recall, I was so excited to be in the analytics industry, and learning all these things. And I think there were two things that would have made a huge difference to me in the first few years of my career. I think the first one is actually working much harder as a listener. I feel like that has served me so well, later in my career, if I had been using that early on, and just really listening for what the real problem is, especially in digital analytics, and it’s probably this way in other analytical pursuits, but there’s often this inability for people to express what they need or desire in a way that actually maps to what it is that we do, or the metrics and the associated. And so, listening for the underlying problem became sort of this Holy Grail. And I feel like if I had worked harder at it earlier, I would have saved myself some bumps and bruises.

0:24:53 MH: I think the other thing… And this is actually a more byproduct of my own learning and education, but I actually think it’s still kind of prevalent today, even though there’s a lot more resources available is sort of always be synthesizing the knowledge you gained, and putting it into context. So there’s so much to learn, and there’s so much data, and details and arcane knowledge about how tools work, how tag management systems work, all that kind of stuff, that you always need to be kind of putting it in its proper place. So what does this mean? How does that fit? What does that mean to the business? How do I make that part of the bigger picture? And I think that’s probably the thing I wish I would have known to do early on as well. So those are my two answers. Thanks, Jamarius. That’s a really great question.

0:25:39 TW: Mine will be a simple that skipping QA of your analysis is not an option.

0:25:44 MH: Oh that’s… Yes.

0:25:46 MK: That’s a good one.

0:25:47 TW: ‘Cause I feel like I learned it pretty quickly to start triangulating and vetting and validating, even though it was painful and time consuming, but I wish that more who were starting out kind of realized that was a… That’s a big deal, ’cause we’re so excited when we pull the data. And it’s there. And we’re like, “We’re done.” And then, boy, like one time out of 10, you’ll find that you really… Just doing the QA, more ideas will occur to you even if you don’t find a problem in what you did.

0:26:19 MH: Yeah. That’s…

0:26:19 TW: Cool.

0:26:20 MH: Nice. Alright, let’s get to what is our next question?

0:26:25 Yehoshua Coren: Hello and welcome to The Digital Analytics Power Hour. I am not your host, Yehoshua Coren, and I have a question for the fantabulous, tremendous, supercalifragilisticexpialidocious 100th episode of the Digital Analytics Power Hour with your host the quintessential analyst, professor Dr. Timothy Bimothy Jimothy Wilson and his friends, the other hosts. My question is about dealing with dates within web analytics data. Since web analytics data is innately time-series data set, how would you measure how many users took action A and then took action B within 30 days without explicitly implementing event tracking or something similar for it? For example, what percentage of users came to the site via PPC ad and bounced, but came back within 30 days? No matter what I set my date range to, any users with a bounce session from PPC within 29 days of the date range that I set in my reports, will not have had the full 30 days to meet the condition. So I pass it to you, analytics superstars, rolling date ranges, cohort analysis, Tim Wilson’s hair. However you want to answer, that’s fine with me. Good luck and mazel tov on the 100th episode.


0:27:45 MH: So a lot of our listeners might notice that some entries exceeded the 30-second maximum, but we just felt we had to include them anyways. And Yehoshua being such a great friend of the show, we…

0:28:02 TW: Also known as Yehoshua.

0:28:02 MH: Yehoshua.

0:28:03 TW: But that’s okay.

0:28:04 MH: Yeah, I said it right, I think, I don’t know.

0:28:06 TW: Well we’ll find out.

0:28:07 MH: Yehoshua.

0:28:08 TW: We have it on tape.

0:28:08 MH: Yehoshua. Anyway, my answer is actually a little bit out of the box, Yehoshua. What I do is, first I build an industry leading analytics practice and then recruit talent that knows how to answer that question. Okay, take it away, Tim.


0:28:27 TW: Actually, this is… I love the question, because it’s actually the sort of thing that’s really hard to explain to a business user, right? Even if you pull the data, the fact is, as Yehoshua pointed out, you basically have to tell them that “I’m not gonna use any initial interaction data before 30 days ago. I am throwing out some proportion of the data because I need that to play out.” That extends… I found myself a couple of weeks ago trying to explain that that’s one of the… It’s awesome if you can use last click, if you’ve vetted for attribution that last click, and first click aren’t that much different, then you’ve actually bought yourself not having to deal with a similar sort of thing.

0:29:16 TW: So I think the best option which ties into this whole journey into the world of data science, is to be working with the raw data, because I think if you query the data in its raw form, you can query it to address that. I do think my experiences with the cohort analysis in both Adobe and Google Analytics, is that I keep thinking cohort analysis is gonna give me what I want, and then I immediately hit a limitation or a wall and it doesn’t really work out. So I guess I’d say that’s like a awesome question, and it’s like, you really have to dig in and figure what’s the least imperfect way that you can do it, and then Moe can just say, “No, you just do this simple thing.”

0:29:58 MK: Well, I do use event tracking, that’s the issue. So when I have this problem, I do rely on existing event tracking which… And then I basically run into my code like here is the first thing they did. Did the second thing that I want them to do happen within 30 days? And if it didn’t, then get rid of it. But if it did, then count it.

0:30:17 TW: But you’re querying… You’re doing that in BigQuery though.

0:30:19 MK: Yeah, yeah, I would do that in BigQuery with the raw data.

0:30:22 TW: I think it’s event tracking with more saying, “Don’t do something crazy, like pull the… Put your… That, you know, timestamp of first PPC interaction.”

0:30:30 MK: Yeah, yeah.

0:30:32 TW: So I think that’s legit. But you are going to sing, you have to get to the user level data to do that within the aggregated Google Analytics or Adobe Analytics interface. That’s the sort of thing that seems like a question that it would easily answer, and it’s not…

0:30:48 MH: Yeah, it’s tricky.

0:30:50 MK: Yeah, I did… I did…

0:30:51 TW: I’m not sure. Moe, you don’t even really know how to navigate the Google Analytics web interface at this point do you?


0:30:58 MK: Well, actually, someone…

0:30:58 TW: You’re like, “Look, if I’m not writing code, get the fuck out of here.” [chuckle]

0:31:00 MK: Someone in my team was laughing at me ’cause when I used the Google Analytics UI, I always go straight to the search box. I never go down and navigate to which report I want, I just put it in the search box. And she’s like, “Oh my God, I never do that. You really don’t know your way around.” So I’m like, “Yes, I do. I just… It’s a shortcut.”

0:31:15 TW: I did a little bit of prep and could not find the cohort analysis, so it was in the search box. It’s under audience as well.

0:31:21 MH: Yeah.

0:31:21 TW: It’s under audience as of the time of this recording. No telling where it’ll be before that.

0:31:24 MH: Actually, what’s interesting is probably about half the people who use it go one way and half the people go another in terms of navigation via search and navigation at least in user usability studies we used to do on our website. That was actually pretty true.

0:31:35 MK: That’s really interesting.

0:31:36 RP: Yeah, so tell your friend that maybe she should stop thinking of herself as the only user. [chuckle] Maybe there’s other kinds of users out there. No, maybe not. [chuckle] But in a nice way. Alright.

0:31:48 MK: Okay.

0:31:48 MH: Before I dig a [chuckle] deeper hole, let’s get to our next question. [chuckle]

0:31:54 Helen Cripps: Hi, this is Helen Cripps coming to you live from Helsinki Airport as I’m about to fly back to Australia after presenting at an IoT conference. For Tim Wilson, the quintessentialist analyst, what do you think the impact of IoT and connected devices is going to be on marketing analytics in the future? Thank you.

0:32:16 TW: Oh, great question. And I think we need to start a new rule. Anytime we hear the word quintessential, we gotta drink. Yeah, everybody? [chuckle] That includes you out there too, listening.

0:32:30 MK: Well, it’s 9:30 AM in the morning here.

0:32:30 TW: Oh, yeah.

0:32:31 MK: So I’ll drink my water.

0:32:34 TW: You don’t have like a Kombucha or something like that?

0:32:38 MH: Alright. Well, Tim, [chuckle] that question was for you.

0:32:40 TW: It was. I love many aspects of the question. One, and I got to meet Helen when… Actually, when I was at the Loves Data Conference in Melbourne a few years ago.

0:32:48 MH: Wow.

0:32:49 MK: I love that you said data and not data.

0:32:53 TW: Yeah, I’m trying ’cause I was speaking… I’m speaking…

0:32:57 MH: And you said Melbourne.

0:32:58 MK: Australian.

0:32:58 MH: Yeah.

0:33:00 TW: Melbourne?

0:33:00 MK: Yeah.

0:33:01 TW: Melbourne.

0:33:01 MK: Speaking Australian.

0:33:02 MH: Yeah.

0:33:02 TW: I’m working on it. I’m prepping up for a trip I have in February. I wanna kinda fit right in with the locals.


0:33:06 MH: Vegemite.

0:33:07 TW: Yeah. [chuckle] Oh, Vegemite’s coming back for everyone.

0:33:09 MH: Yeah.

0:33:10 TW: Everyone in the office.

0:33:12 MH: I’m a fan.

0:33:12 TW: So I like that somebody would be at a conference and think, “Hey, this actually sparks a question to ask of the Digital Analytics Power Hour.” But to me, IoT has been sort of replaced as a hot topic by a lot of the voice stuff, which I don’t know if you consider Alexa and Google Voice or those things that are like Internet of Things, those are like this. I think when were first talking about IoT, we were thinking about the smart refrigerator, the smart microwaves, the smart garage door openers and there’s a lot of potential there, but then voice came in and that seems to be like a… Felt to be like a bigger opportunity for marketing and analytics, marketing analytics and with privacy concerns along the way.

0:33:50 TW: But I think it’s gonna come down to the degree to which marketers really start fulfilling what they… What we learned in school, which was you own the customer all the way through to retention, because so often, marketing is acquisition and maybe loyalty, which I guess is retention, but if… For cases where there’s a product being sold and that product can be internet-enabled, the potential to use that information to tie it back to the CRM and effectively market to existing customers. But that I think requires a collaboration between R&D, engineering or product or whatever it is, the physical product and the marketer. And I think that’s gonna be a heavy, heavy lift for a lot of companies. What do you guys think?

0:35:04 MH: Well, humbly submitted, because I’m not a quintessential analyst like Tim.


0:35:10 TW: Really?

0:35:12 MH: Yup, and drink, but…


0:35:16 MH: But like…

0:35:17 MK: Hey this could work out great for me.

0:35:19 MH: Like most of my opinions, Gary Angel spoke about this at an exchange conference a few years back and, of course, I’ve tried to hold on to his many teachings. But he compared behavioral data and sensor data, which is kind of what a lot of the data we get from IoT is. And I thought that was a very interesting juxtaposition. I think behavioral data has a shelf life, sensor data has a shelf life in terms of its actionability and they’re not the same, and they don’t mean the same things. And so there’s sort of a interesting discovery hurdle, a sort of applying a sense of meaning to both data sets and the combination of those data sets appropriately. Sensor data from six months ago, if you’re trying to grow grapes on a hillside is completely worthless.

0:36:11 MH: Whereas, what customer behaviors were six months ago, buying wine from your vineyard, may actually have some weight still. So, I like that juxtaposition. I don’t know that it really answers the question of the impact of IoT in connected devices. But it’s interesting. I also, think it’s interesting that we’ve already seen a chip maker in the IoT space, ARM, buy a customer data platform company, Treasure Data, just this year. And so, obviously, there are people who are seeing a connection between needing to incorporate IoT, or sensor data into things like customer data platforms, to really understand and explain data at the enterprise level. And so, I think that’s pretty fascinating. So, I probably have more questions than I’ve got answers at this point. But I think those are the two things that were kind of bouncing around in my head when I heard that question.

0:37:04 TW: As you bring up the… Because there is the time nature of the data that when there’s… It’s again, it’s unlike a lot of customer data of gender or location. Not that there’s not time components of how fresh is my customer data. But there’s a reason that people have taken Raspberry Pis and Arduinos, and hooked them up, and said, “I’m gonna pump that data through the measurement protocol into Google Analytics,” because to your point, web behavior has a very… Has a time-series component. And IoT data also, has a time-series component. And both of them can be married with a customer component, of who is the customer. Who is actually owning this? So there are some parallels. It’s interesting.

0:37:46 MK: So warning. About to step on to my soapbox.

0:37:50 MH: Oh, here we go.

0:37:52 MK: It’s gonna come down to actually connecting these things, right, from an analyst’s perspective. If you ask Alexa to order you more laundry detergent, do you actually see that customer then go and buy the laundry detergent? Or see it arrive? Or if they ask the weather, then how does that impact what they choose to wear for the day? To me, it’s about how are you gonna connect these things? They’re watching Netflix on their fridge. And now they’ve stopped, and they turned it on in the lounge room. It’s about connecting those things so that you can actually understand how your user is engaging with your product. Or what would the reasons be? Like for example, I often watch Netflix on my laptop in the kitchen while I’m cooking. And then I swap to the TV when I’ve made my dinner. Those are all questions that we wanna understand, in developing our products and our… Now I’m back in marketing marketing strategies.

0:38:41 MH: But how important to all of that is the temperature second-by-second of your refrigerator as you’re doing those things? So that’s…

0:38:50 MK: Oh, I don’t care about the temperature of my refrigerator. But I do care about the temperature outside.

0:38:53 MH: But that’s the IoT data that’s coming… We’re gathering that kind of data. So, it’s a lot. Right? I’m not disagreeing with your point. I’m not. I just wanna highlight sort of, like… That’s what I mean by sensor data. Is just like, “It’s a lot of data.”

0:39:11 MK: We had a really incredible presentation at Web Analytics Wednesday, which was actually talking about specifically, that point. So they were using the opening and closing of the fridge to help determine whether people in retirement villages were actually cooking dinner. Or testing the humidity in the bathroom to see whether or not they’d had a shower. And so, there are lots of really incredible use cases for that technology, and how we as analysts can help determine really complex… Answer really complex questions based on that. So yeah. I do think there’s a place… Yeah, the fridge might not be for everything.

0:39:47 TW: But I would say you don’t always have to connect it to a user, right? If you take, and say, the assisted living, or the retirement community, and you’re just looking in aggregate, what proportion of our… Or what time in aggregate are people opening the refrigerator door in the first time in the morning? ‘Cause that’ll give us an indication of when they’re waking up. You know what, maybe we should shift this program by 15 minutes or 30 minutes one direction or the other based on when they’re naturally waking up. I agree, it’s gonna be the same thing. There’s a lot more value when you’ve got it connected it to a user. But I think, just learning about device usage, or device deterioration, or if you’ve got garage door openers that are reporting data, and you can take that, and say, “This is when it’s gonna start to kick crater.”

0:40:34 MH: I’m gonna make an executive decision, ’cause what I think actually happened is Helen just stumbled on a really good topic for a whole show. But we don’t have that kind of time.


0:40:43 MH: So, we’re gonna move on.

0:40:44 MK: But I do just wanna shout out to Donna Bradford, who did that presentation. It was really incredible.

0:40:50 MH: Okay. Fine. Fine. You can give a little shoutout to somebody. That’s okay. But we gotta go on. Alright.


0:40:57 MH: Actually, our next question is from somebody who’s done a little playing with some IoT data. But let’s listen in.

0:41:04 Erik Driessen: Hi, Moe, Michael, Tim. I’m Erik Driessen, data technology lead at Greenhouse Group in the Netherlands. My question is about the reason we do our jobs. We actually love it. My question is also about Christmas Eves. Do you remember that as a small kid, you got this lovely feeling of anticipation on the day before Christmas? I was wondering if you ever experienced this feeling doing your work. When you figure out this theoretical solution to the problem, set up the required systems, filled yourself up with expectations. You then give the system a spin. And then that feeling of, “Yes, it actually works.”

0:41:38 MK: Okay. Eric, this is literally, why I do my job.


0:41:43 MK: And I think that all analysts… Well, I would hope all analysts have that feeling, ’cause otherwise, I don’t really understand why they get out of bed. But for me, the moments that I most have that feeling, are the times that I have spent so far too long cleaning data. So you feel this real sense of achievement because you’ve waded through something pretty messy. And the other time is when you’ve had to iterate multiple times because as you’ve run and rerun and rerun, each time you’re like, “Hang on. If I tweak this little thing, it will be more accurate. If I fix this little thing it’ll be better.” And both of those times are the ones that I probably have that biggest like, “Yeah, nailed it!” But basically that’s pretty much why I love my job, is that feeling.

0:42:26 MH: I can’t agree more. You know, it goes back to me. The very first episode we ever did on this podcast was about what it means to be an analyst. And I feel like Erik, in a certain sense, kind of actually put a pin in one of the major points of what that definition is, is that feeling of excitement about seeing something you’ve analyzed or built or created actually come to life. And what’s crazy is I remember an analysis I did that saved the company, I wanna say, approximately $30,000. It was not a lot of money, it felt just as good as other analyses that were worth like millions of dollars. And that’s the cool thing is, I’ve got something. I’ve stumbled on something that is the truth and I’ve been able to prove that this is something that actually means something and it will be impactful. And those are the highlights of any analyst’s career is to get to those moments.

0:43:24 TW: So I had one that on the one hand is really just a data cleanup, proactive data cleanup thing, which on the one hand it’s like, “Oh I don’t wanna be excited about something that just makes my data cleaner.” Except it was really affected metrics that went into a weekly report and that weekly report was actually one that a lot of high level people looked at and that was… And I was also getting into R, and so I had gotten this script that basically would do a whole bunch of brute force processing of data from Adobe Analytics, basically doing anomaly detection looking at device type, GO, browser, OS, things that were often kinda cues for bots, then it would combine them together. So, it would crank through, it would run for a few hours and then it would spit out this nice little report that would say, “These are probable bots.”

0:44:13 TW: It was using MAD, median absolute deviation, but it basically meant that instead of regularly publishing a report that actually had bad data in it, I was able to say, “Oh, I can go and segment that traffic out.” Didn’t get it totally fully automated to where it ran and it would add itself to a segment, ’cause it still took some human judgment, but that was cool. And then, there was another cool one I did where I combined Twitter, Twitter API data with Google Maps API data, and was actually able to kind of visualize where a bunch of a user’s followers were on a map, and that just wound up being really useful, ’cause I was like in theory, these are data from two different sources. I’ll rely on the magic of Google to resolve whatever somebody’s put in their Twitter location profile, and then it kind of spit it out, and it was actually pretty meaningful because this was… The main reason I was doing it was a person who had an international profile had switched to a very US-centric job, and we were trying to get a read on how much of his… Just his followers were actually in the US or not, but it was pretty, pretty cool. I was very excited with that one as well, when it actually worked.

0:45:22 MH: Yeah, that’s… Yeah, thanks, Erik. All right, let’s go to the next question.

0:45:28 Simo Ahava: Hello, esteem hosts of DAPH and Tim. And congratulations on the amazing milestone. It’s Simo Ahava here, web analytics developer from 8-bit-sheep. My question is simple. This analytics and data stuff. Do they really matter? Why do companies that don’t care about data still survive and even thrive? Are we kidding ourselves? Are we all just hot air inside one big bubble floating through the noisy and smelly undergrowth of corporate business? Help me.

0:45:57 TW: Oh, Simo. Who wants to take a swing at that one?

0:46:04 MK: This is something that I’ve been thinking a lot about, because… Yeah, so I’ve been reading lots of books on leadership as well, and the culture that you have at a company and you read all of this stuff about like, “Do you have a good culture, or to be data-driven, you need to do X, Y, and Z.” And I literally have been thinking about this question. Well then, why are there still so many examples of companies that have shitty cultures or don’t make data-driven decisions or whatever it might be, that still seem to thrive? I don’t know the answer to that, but what I do think is that we have a pretty incredible responsibility as analysts to help direct businesses, but I actually think that responsibility is growing. When we start thinking about concepts like machine learning, and whether you can perpetuate a bias through the model that you build, our values can impact some pretty incredible technology that we need to spend more time thinking about how we direct businesses, because I don’t think we can just say, it’s up to the company to decide whether or not data matters or how it’s used.

0:47:15 TW: Come on, it’s a straw man. There’s no… You cannot find a company that does not care about data, they may not be using it effectively, but there is not a company out there that is not at least pushing reams of data out to people, right? I think that’s a, I think that…

0:47:33 MK: Yeah, but that’s not the same thing. When you say “I wanna do this, now go find me some numbers to support that.” To me, that is not caring about data. When you go, “Here is the answer. Find me something to support that,” that is not caring about data.

0:47:50 TW: Literally, that doesn’t happen either. Like right, we’re standing up that, that…

0:47:54 MH: But wait, the question is why do those companies still survive and thrive?

0:48:00 TW: I argue there is a tiny number of companies that aggressively don’t care about data. I would say 100% of companies tell you that they are being data-driven, whether they are or not, they certainly don’t care about… I think we have to be really careful about dismissing these…

0:48:19 MH: It feels really awkward to me to come down as more pessimistic than Tim Wilson on any topic.


0:48:31 MH: Because I think on the surface, everyone is definitely concerned and cares about using data, but in practice, I don’t believe a lot of companies have really embraced what it means to use data.

0:48:46 TW: But I think that’s a different thing that is they’re not… They don’t have effective strategies and they don’t actually know how to…

0:48:51 MK: Misuse to me is the same as not caring.

0:48:54 MH: But they’re comfortable without those effective strategies, Tim. And that’s the thing that has to change is the relative comfort. The thing…

0:49:02 TW: And I can also tell you why those companies still, still thrive, but…

0:49:08 MH: Well, no, I have an opinion on that too.

0:49:09 TW: Okay. Well, you go first.

0:49:11 MK: If you’re making, if you’re making… To clarify, I very strongly believe misuse of data is the same as not caring.

0:49:19 TW: No.

0:49:20 MK: How are they successful if they’re making shitty decisions based off bad data or bad practices with their data?

0:49:26 MH: Here’s the thing, businesses are a system, right? And that system incorporates a number of data points that have been proven over time. And companies still operate with those. So there’s an understanding of the business, and how it works, and how that system should optimize, and how it should work better. Digital data, and all the stuff that we do with analytics and those kinds of things, are sometimes net new to that equation. And it’s not enough that we suddenly show up with all this data and potentially information, we have to earn our ability to influence those executives and leaders in how they assimilate and understand the systems they are responsible to measure and optimize. So in other words, a business executive has to raise revenue. And so, he knows or she knows: “I need to do this, this, or this, irrespective of any data I see to do that, because I just know it. There’s business cases about it, I got my MBA from a great school. I’ve learned how to do these different things.: That’s the thing is people are using data up to a point. It’s the incorporation of new data or digital data that is actually sort of the outlier, and we’ve gotta earn our place in that, not just be granted it. That’s my opinion.

0:50:56 MK: I agree with the last bit of that, not the rest.

0:50:58 MH: No, the rest, no.

0:50:58 TW: I think when you regularly undervalue, as analysts, we regularly undervalue the inertia of a well-established brand. If you are a well-established brand that is a hundred years old, you have and you dominate the market, then guess what? You probably don’t need to be quite as nimble, and agile, and effective with data. So that means if you’re a new company, yeah, there’s more of an imperative. I think we undervalue the value of just creative thought, and great product, and great ideas. And I think there are tons of things. It’s like you’re not just gonna do… You have to have some core fundamental ideas, offerings, products, services, and then, yeah, the data’s gonna help you be more effective. And over the hundred year, yes, you’re gonna have to use the data more. But I think we also need to get over ourselves a little bit and recognize that there are brands that we’re gonna be playing at the margins with what we’re able to do with data.

0:51:57 TW: But I also completely dis… Willful misuse of data is not caring about the data. Ignorant misuse of data, being naive, because you haven’t had the right people inside to kind of point that out, I don’t really look at that as not caring about data. I’m not quite that… I’m actually way kinder, I guess, on the intentions and the motivations than…

0:52:20 MH: Yeah. Moe’s the hard line in here.

0:52:25 TW: Again, weird for me to say, I am less cynical on anything, but I think I am.

0:52:29 MK: Yeah, I do think that’s the nature though. Maybe when you’re in an agency and you’re talking to clients, clients want to understand something, which is why you get their business. When you’re in a company, you might have stakeholders who are like, “Well, this is what I think, and I don’t really give a shit what the data says,” or, “I wanna use the data to support whatever it is I already think.”

0:52:49 TW: I was at a creative agency for three years that had plenty of that.

0:52:52 MK: I’m taking in recent experience.

0:52:55 TW: Okay. [laughter] At the same time, they were like, “Give me some insights.” And I was kind of like, “Well, that’s not gonna happen,” but the fact is I was like, “Well, you’re a household name that’s just gonna make all sorts of decisions, whether the data’s there or not, and you’ll probably be okay, ’cause you’re a… The three laundry detergents you manufacture are 80% of the market. You’ve got a place of market dominance. You’re not gonna be up-ended in the near… “

0:53:21 MH: But I don’t think that’s the only way, Tim. Anyway, obviously it was a good question.

0:53:25 TW: Another new, Tim [0:53:26] ____.


0:53:28 MH: It flipped a switch for all of us. I would say on that question, Kevin Hillstrom is a guy on Twitter who I follow, and I like how he talks about this, he is @minethatdata. Okay, next. Next question.

0:53:42 Jack Golding: Hi, this is Jack. I’m a web analytics lead at a large Australian online publisher. And I want to ask, are analysts, especially junior analysts, too hard on themselves? And should these analysts focus more on managing their stakeholders than providing a perfect or even an analytical solution for types of problems they face and they’re asked everyday. Thanks.

0:54:07 TW: Yes.

0:54:08 MH: Well…

0:54:09 TW: You guys have thoughts on that? [chuckle]

0:54:11 MH: Maybe.

0:54:12 MK: Yeah, okay.

0:54:13 MH: So I think that’s interesting but Moe, you were about to say something. So why don’t you get us started?

0:54:19 MK: So I think… All of these questions are so relevant to my life right now, it’s really interesting. And as an analytics team that sits within tech, one of the things that I struggle is that we create a culture where it’s completely acceptable for our engineers to make mistakes and in fact, we sometimes even celebrate those mistakes. I think that as an analyst, when you make mistakes and it does happen, I have made plenty of mistakes, the difference is that when you make a mistake as an analyst, you often lose credibility because of it. Well, that’s in my experience. So I don’t think we can afford to have the same culture as our engineers, but I do think that man we’re… We are our own worst enemy sometimes. And maybe we need to find the right balance between how do we get good QA so that we feel confident enough, but also don’t kind of beat ourselves up, because it doesn’t matter what job you do, you’re gonna make mistakes.

0:55:21 MH: It’s interesting. So I feel like I know why you said yes, Tim, but I’m not gonna try to explain you. I’m gonna explain [chuckle] the opposite because what I have observed is actually, it’s maybe. There are actually two different, and it’s more than two different kinds, but there’s sort of like, there are people who approach analysis with a supreme confidence that they will arrive at the right answer. I really gotta be honest, I don’t understand what makes those people tick.

0:55:49 TW: That’s a good point.

0:55:49 MH: But I have observed them in the wild and they’re way different than me. And I know they’re different than Tim, too, ’cause he says so, but that… And that’s the thing where I think sometimes yes, analysts are way hard on them… Way too hard on themselves and they need to give themselves some relief. At the same time, Moe, you’re right, the stakes are extremely high for analysts and that’s why you benefit from having other analysts vetting, looking, helping, driving in a healthy collaborative way. It’s a great thing, but there are also those ones who have to balance their own confidence against their ability to actually analyze. And so there are people out there who are not hard enough on themselves.

0:56:33 TW: I’d like to… I’d like to withdraw my yes answer and change that to a maybe.

0:56:39 MH: Okay. So, but…


0:56:41 MK: Jeez. Bit of fence sitting for you today then.

0:56:43 TW: No, it’s a great point. Like the overconfident, yeah.

0:56:46 MH: Well, it’s part of what makes him quintessential.

0:56:49 MK: Bottoms up.

0:56:50 TW: That’s a fair point ’cause I do think that’s, that I was thinking about when there are… When a junior analyst is being said, “You need to go and answer this question,” and they beat themselves up because it might not be an answerable question. They’re getting asked…

0:57:03 MH: Yeah, that’s a good point.

0:57:05 TW: Asked something that’s not realistic.

0:57:06 MH: Yeah or the data doesn’t support it, right?

0:57:10 TW: Yeah.

0:57:10 MH: Having the confidence to say, “I’ve analyzed the data. The data has no influence on the decision you’re trying to make.” And instead we torture it into trying to answer the question and that’s where we get things like time on site and stuff like that.


0:57:29 TW: Perfect.

0:57:30 MH: That’s really good. Wow, man, these questions are… It’s almost like we hand-picked them.


0:57:36 TW: Shh.

0:57:37 MH: All right. Now, we’re winding down the stretch, turning the corner and headed for home. And so here comes our last question.

0:57:46 Jim Sterne: Hello, Power Hour gurus. This is Jim Sterne, and my question is, how might the Digital Analytics Association knock it out of the park over the next few years on behalf of the quintessential analyst?

0:58:00 MK: I believe it’s time to drink, gentlemen.


0:58:07 MH: Alright. Well, Jim, thank you so much. How could you have known that we’ve turned this into a drinking game at this point?


0:58:15 MH: Although in a certain sense, Jim knows this industry and us pretty well. So I can take a stab at this.


0:58:27 TW: It’s such a… Who else, everybody raise your hand if you’re on the board of the DAA currently?

0:58:31 MH: So this year, I was elected to the board of the Digital Analytics Association. It’s something that surprised me and that A, I wanted to do it and B, that it actually happened. But three things emerged as I started to try to think about what could the DAA be to its members and its nonmembers. And so to knock it out of the park on behalf of the analysts, I think one of the things the DAA already does really well, is it highlights educational opportunities. I would love to see the DAA expand on that and highlight those opportunities no matter where they are in the world.

0:59:11 MH: I think it serves our members and it serves all people in analytics when we highlight what’s great about what’s happening in analytics, whether it’s a DAA sponsored activity or not. And so I think that’s one thing. I think the second thing is you’ve got to make sure that you as an organization, and we’re… It’s a struggle, it’s… As I’ve become more aware of how the organization operates. It’s hard, it’s really hard to do all the things that are possible to do in the prioritization necessary. And I owe so much to others who have served on the board and are serving on the board. People like Amy Sample and Valerie Kroll and so many have come before. So I’m not ever gonna mention enough people, but I just… These are for the folks who have been kind of helping bring to a focus what has to happen. And I think in our world, we need to make sure that we’re giving people what they need to advance their careers in the way they wanna do it. So right now, I think areas like data science and things like that, that are top of mind for analysts, needs to be something the DAA is associated with. I’ll stop right there.

1:00:22 TW: Moe can you just… Now can you explain why we’re just doomed?

1:00:25 MK: I have a really self-interested perspective on this. So as someone that sits outside of the US, I do struggle a little bit sometimes.

1:00:35 TW: Specifically in APAC, right? You’re in a time zone that gets woefully underserved.

1:00:40 MH: Yeah.

1:00:40 MK: Every time there’s something really interesting on, it’s like 2:00 AM in the morning, which I get that you can’t cater for every time zone in the world. But the recommendation I would have would be around, how do we cater to a global audience and that might be, for example, a different price point for international members, because noting that we can’t go to a lot of the DAA sponsored events, but the other would be that…

1:01:03 TW: Wait, that’s a great point. I knew at least those two points were gonna come up, but it had not occurred to me that you’re literally not… Not practical to avail yourself of many of the live events. I mean in theory, yes, there could be a DAA chapter in Sydney, but that still would be only one little slice of live events which you have access to. Sorry, go ahead.

1:01:25 MK: Yeah. So the other thing I would recommend is like opening the door a little bit. So how do we include people that aren’t members and get them to contribute their thoughts and their ideas particularly because a lot of those people come from places outside of the, outside of America who also have a lot more technical expertise, which is the way that the industry is going and we wanna bring that expertise into the fold.

1:01:52 MH: That’s really good.

1:01:53 MK: So I don’t actually know this strategy that’s for Michael and his board to figure out. But I think tapping into that knowledge, how do we get people to contribute to the cookbook that aren’t members, or little things like that, that just like opening up the door to bring everyone into the fold so to speak on it. Yeah.

1:02:14 TW: It’s so funny ’cause I actually had a… On one of the committees that I’m on, I had something that’s very, very aligned with that specific point. The one thing I’ll throw in, ’cause I also had the US-centric knowing that there’s… It’s tough… It’s like…

1:02:29 MH: Yeah.

1:02:29 TW: Where you’ve got your critical mass is where you need to cater to to grow it farther. So I get that, but I think there’s a challenge or I wonder if there’s kind of an impossible challenge, and that the DAA from day one has said we’re going to represent industry in-house companies, we’re gonna represent agencies, consultancies, and vendors and we’re gonna have all of those represented in this association. And the reality is, is they are not inherently necessarily aligned on what their interests are. So there’s kind of a level of kind of everything is gonna get sort of boiled down to the lowest common denominator, and I just think that’s a challenge. It would be really interesting to watch that get hashed out. What if we removed one of those from the equation and said, “You know what, maybe we’re not gonna worry so much about agencies and consultancies,” if that would drive something. That’s a… That’d be a huge shift. I don’t know that it’s the right thing, I just think it’s tough for the association to take a strong position on anything.

1:03:32 MH: Yeah. We are doing some work in 2018 with the DAA in Germany, and we’ve partnered with them to create sort of a test. And I’m really hopeful that will help us understand how to make more capabilities available internationally, over time. But… Well, that your point is really, really good. It’s not easy, it’s not easy at all. Okay, wow. Man, this actually exceeded my expectations. I really loved going through these questions, and it comes down to our listeners who submitted these questions. They were amazing. Before we…

1:04:13 TW: They will be richly rewarded with…

1:04:15 MH: That’s right, with a proprietary, never before seen prize of some kind.

[overlapping conversation]

1:04:24 TW: We know what it is, but just in case plans change…

1:04:28 MH: We’re just not talking about.

1:04:29 TW: But, actually…

1:04:32 MK: No, we’d like it to be a surprise. That’s what it is.

1:04:34 TW: Oh that’s okay. You should be in marketing.

1:04:34 MH: Before we go, Moe and Tim, I’d love to hear from you what’s been your biggest take away from doing the podcast a 100 episodes in or slightly fewer than 100 Moe, in your case, but we’ll just give credit for all of Jim’s too.


1:04:54 MK: I have written about this in my blog and I don’t actually think my thoughts have changed, particularly. It’s more that I was really scared to do this, because I thought that I’d say something stupid. And let’s be honest, I say stupid shit all the time.

1:05:08 TW: Yeah. Get in line.

1:05:09 MK: Now it’s just recorded for everyone’s hilarity, but I just, I keep learning from it, I actually… The guests that we get to chat to and the conversations we get to have, I feel like every time that we have an episode, I have some takeaways that help me do my job better, so I am really thankful that I get to be a part of the show.

1:05:30 TW: I mean, I will say from doing a lot of the editing that somehow I have become believe it or not, a more concise speaker who says “uh” less. It’s a long journey, it’s a lifelong journey, but that’s actually been a benefit. I am the same way when I talk to people about the podcast, I honestly regularly come back to… I just learn so much from getting to dive in to any one of these topics for a focused amount of time. Whether it’s just with the two of you, because it’s a topic or whether we have a guest who’s on that’s great. I mean at one point we said, “Are we gonna run out of topics?” Our topic list is growing faster than we can put out shows. So the answer is no, but it really is both fun and super, super educational, to be on every one of these.

1:06:22 MH: Yeah. I agree. I think this podcast is for me just a continuing expression of that kid on Christmas Eve that Erik was talking about. It’s profound that I could walk into a room of analytics people and say, “Hi everyone.” And half of them might turn around or recognize who’s talking. That or drunk helps on the Measure Slack. So either, either way. I’ll take whatever I could get. But yeah, the guests we’ve had, the conversations, the knowledge I’ve gained from the show and the relationships with people all over the world that I’ve been exposed to through the podcast, I think, I guess it doesn’t surprise anyone that’s it about the people for me, but that’s how it always works. But yeah, I think 100 episodes in if Jim doesn’t keel over of like a stroke from all the work he does to keep this thing alive…


1:07:23 MH: We’ll keep this thing rolling. Well, I wanna say thank you to all of our listeners who submitted questions, and a big thank you to everyone who got their questions answered on the show today, but all of you are people who are near and dear to us. Thank you for being with us through all these episodes, we’re excited to see where Episode 101 onward goes from here. And I think I definitely speak for all of the folks who submitted questions, and for my two co-hosts, Moe and Tim, when I say be that kid on Christmas Eve or whatever gift-giving holiday you celebrate ’cause you know not everybody celebrates that one. And get out there and keep analyzing.


1:08:14 Announcer: 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.


1:08:35 JT: So smart guys want to fit in. They made up a term for analytics. Analytics don’t work.


1:08:43 MH: I don’t want that on my conscience.


1:08:48 MK: Where’s Tim gone?

1:08:49 MH: He went to go get another beer.


1:08:53 MH: Tim is not like…

1:08:54 MK: Trustworthy?

1:08:56 MH: I don’t know, we’re a 100 episodes in and sometimes I question his outtake choices. Now it comes out.


1:09:04 MH: You’re an elder statesmen, Tim.


1:09:06 MK: I though that there was gonna be sarcasm there, I’m impressed with you.

1:09:10 TW: Oh, did that not come across as sarcasm?


1:09:15 MH: You’re supposed to start Tim, it was directed at you.


1:09:18 TW: It’s really gonna come down to whether marketers who, if you go…

1:09:22 MH: Even your dog disagrees with you, Tim.


1:09:27 MH: Okay, then. No, I’ve got something to say but I ways kinda getting, let Moe go first.

1:09:30 MK: No, you go first ’cause I kinda zoned out a little bit.

1:09:33 MH: Okay.


1:09:35 TW: You’re not even drinking your water.

1:09:38 MK: I know but it could work great for me getting both of you to mutually…


1:09:43 TW: If you haven’t connected your mobile device to a desktop, then get the fuck out on IoT.

1:09:49 MK: No, but okay.


1:09:52 TW: You know what? If I jump in and start every time and then you’re like, “Yeah, Tim’s always fucking over talking on Moe.”

1:10:00 MK: Well, I feel like I jumped in and started a lot this time so I was trying to be like…

1:10:03 TW: No, you’ve done it like 7% of the time. I’m still gonna sound like an asshole who’s mean to you.

1:10:09 MK: You are not an asshole that’s mean to me.

1:10:12 TW: Aww.

1:10:12 MK: I’ve got lots of evidence. You said something nice to me in the last episode.

1:10:15 TW: I still think you should answer this question first.


1:10:19 TW: Rock, flag and eagle! Rock, flag and eagle! Rock, flag and eagle!



Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.

Have an Idea for an Upcoming Episode?

Recent Episodes

Three pay phones mounted on a wall next to each other

#245: Dear APH-y – An Analytics Advice Call-In Show

https://media.blubrry.com/the_digital_analytics_power/traffic.libsyn.com/analyticshour/APH_-_Episode_245_-_Dear_APH-y_-_An_Analytics_Advice_Call-In_Show.mp3Podcast: Download | EmbedSubscribe: RSSTweetShareShareEmail0 Shares