Loading summary
Transcript274 lines
- [00:06]
A
Welcome to the Analytics Power Hour. Analytics topics covered conversationally and sometimes with explicit language.
- [00:15]
B
Hey, everybody, welcome. It's the Analytics Power Hour. This is episode 284. You know, in the data and analytics world, we love the old George Box quote. All models are wrong, but some are useful. And to extend the thought a little bit, sometimes that usefulness degrades over time, making that way of understanding the world something we have to leave behind. Whatever the situation, growth comes from updating your priors, as we say in the biz. So that's what we're going to talk about in this episode. A retrospective on some of our retired beliefs. So let me introduce my co hosts, Val Kroll.
- [00:51]
A
Welcome.
- [00:52]
C
Hi, Michael. Good to see you.
- [00:55]
B
It's good to see you too. Tim Wilson. I don't know how you're going to do this episode, Tim. I don't think you have a single belief you've had to change because you're always right.
- [01:04]
A
Oh, that hurts.
- [01:06]
C
Not even two minutes in?
- [01:08]
B
No, I just.
- [01:09]
A
Here they come.
- [01:11]
B
Okay, I'm sorry. We should let ourselves get started first. And mo kiss. How you going?
- [01:17]
D
I'm going wonderfully, thank you.
- [01:19]
B
Awesome. And I'm Michael Helbling. As we get started in this episode, Val, the inspiration for this comes from an interview question. You got one time. And so I wanted to start with you and say what made that interview question stand out and what kind of made it pop out to you?
- [01:36]
C
Yeah, it was an interview question that I received and it also ended up turning into an interview question that I leveraged when I was at Search Discovery. Now further. And the question was asking what is a deeply held belief that you have had that you've changed your mind on over the past year? And they, they did have a, a time bound element to it in the interview question. Not that we'll necessarily be using that today, but I thought that was a very thoughtful question. I think that there's a lot of things that you can intuit by someone's response to that, just even kind of understanding what their mindset is or what could cause them to change their mind. And how are they investing in kind of like that continuous learning who they're kind of learning from to potentially cause that change of mind experiences. Are they gathering? But also, you know, if you have someone that says like, well, I've never changed my mind about anything, that can be a good. Checking the box on does not fit culture here at whatever company.
- [02:39]
A
Do you remember your answer to that question?
- [02:41]
C
Oh, no, I completely blacked out. But that's Pretty. Pretty common for me during an interview. I definitely have things that I'll be able to share for the purpose of today's episode, but I have definitely no idea what I said. Were you asked that question Michael or Tim when you joined? Since three of the four of us were. Went through that process there?
- [03:02]
A
Well, Michael was. I mean, did they even have One of the OGs?
- [03:06]
C
Yeah.
- [03:08]
A
I didn't have. I was. Well before or. I did not go through any sort of standardized interview process. I was soaking wet sitting in a hotel lobby in God, New York.
- [03:21]
D
I'm glad you said hotel lobby. I was a bit. Bit worri. That was going hotel.
- [03:30]
B
Yeah. I mean, I don't think, Tim, your interview process was as much of an interview as much as it was sort of a recruitment. And I think the interview question that stood out to me when I joined Search Discovery was I was asked by the CEO the difference between an S prop and an evar. So that was.
- [03:46]
C
Right.
- [03:47]
B
That was one of the questions. It was a different time.
- [03:51]
A
Webtrins. Webtrins doesn't use.
- [03:55]
B
Well, if you want to say that the. The wt dot.
- [03:59]
C
I knew you were going to say that.
- [04:00]
B
Underscore id. No. So, yeah, we go way back. What are the best campaign parameters? That's one that we all use UTMs now. But I bet back in the day we did think there were better ones. You know, like what are the core metrics? The CM Underscore. Mmc. There you go. So, okay, let's not. Let's not go into that. But that's an ex.
- [04:26]
C
Are these things you've changed your minds on?
- [04:28]
B
Exactly.
- [04:28]
A
I used to believe that shit mattered and was going to give me better measurement.
- [04:33]
B
Yeah, exactly. That's. That's right. All right, all right, so let's.
- [04:39]
D
Well, to be fair, the thing that that did trigger in me was one of my assumptions. Perfect segue I still wrestle with, which is this idea of like the single view of the customer or a unified view. Or if you collect all these events, these variables, we will have a complete and perfect picture and be able to understand what our users want. And I especially think that was true in the world of attribution.
- [05:07]
A
A thousand. I literally sat in a meeting today with people I didn't really know well, and there wound up being a discussion around if we could just get our utms kind of in better shape and standardized. And as Val noted, I'm drinking during this episode.
- [05:23]
C
I was just going to say are you going to.
- [05:26]
A
And it's a 9.9 ABV beer it's a stout beer.
- [05:31]
C
Brace yourselves.
- [05:32]
A
I mean, there are plenty of people who still believe that, and I feel like I was certainly there. Like, I just got to get the more perfect data, and that's going to get me to the answer. And it was like, so easy to obsess about. Let's build processes to make that data, data capture better.
- [05:54]
C
So for you guys, was it like a specific event or a moment in time where it was like the before times and the after times on how you thought about this, or was it like a slow eroding of your confidence and that that was like the right path? I'm curious.
- [06:10]
D
For me, it was definitely a series of debates at Measure Camp about why attribution sucks. And they would always get very, very heated and dramatic. And, like, I would definitely say, for me, it was a slow. And I would still say that fundamentally, I think you can learn things from attribution or collecting different attributes or events about users. That's not to say that I fundamentally don't believe in it entirely. I just think it was the overconfidence we had in the answers it was giving us.
- [06:42]
B
Yeah. With attribution, I think my sort of skeptical nature always sort of was like, what is really going on here? And then a big moment for me was I was looking at a tool that this company was using, and they were kind of showing me like, here's how we do it. And they got to this one point where they're like, and here's where we just put in the weights we want for each channel. And I was like, so that's not attribution. That's just adding things up to 100 based on whatever you feel like.
- [07:12]
A
I was like, that's what I was like.
- [07:15]
B
That's. That's in your tool. That was in the tool. And so that was when I was like, okay, I'm not for this anymore. Like, whatever's going on, I now know I'm against it.
- [07:25]
A
I sort of had two cases. I had built so many spreadsheets or repurposed spreadsheets or looked at other people's spreadsheets, and then I looked at platforms that would manage. So this is very digital analytics specific around managing campaign tracking parameters in the digital context. And then working with one client that had built this thing that was. They were paying for a tool. They had like 50 different values. They were keying off of this thing. And the guy who owned that tool was just like, we just got to get all these disparate agencies to always use this thing. And Then I was like, and then what? These values nobody cares about and it's not going to get you anything. So I was like, that was my. This is getting so in love with the data collection and totally losing sight. And there were fields that were being mapped that nobody in the business even thought about the business in that way. It had been dreamed up somewhere years before. And at the same time, which maybe blends to a second related topic is I fundamentally did not understand that again in a digital analytics marketing world that there was no model that was actually going to demonstrate incrementality, that there was so much around. If you get the right algorithm, the right data driven thing, the right Markov chains, it's going to tell you what value each channel is contributing. And literally that not addressing incrementality. So I think it was actually probably talking with Joe Sutherland. That was when that light bulb went on. And I feel like I've watched it go on pretty broadly across the industry and it kind of stops at marketing all too often.
- [09:18]
D
It's funny Tim, one of the. Because I said like for me it was a slower burn and Sam Redfern like berating me with the word incrementality over many, many months. But I think one of the standout moments for me also was your chip analogy at the grocery store where like someone puts a pack of chips in their trolley. And it was that analogy that really clicked of like, sorry, why don't you explain the analogy? Because you have a very great visual that I semi use.
- [09:49]
A
Well, I feel like that visual came from what I would put. I think I saw Rand Fishkin write about it first. That was the pizza shop analogy.
- [09:58]
D
No, the pizza shop analogy is the one I use because I.
- [10:01]
A
But I feel like that would have been around and then you, I remember you struggling with the visual for it but just that that idea of, of I don't know where I think the potato chips came up with. What could I find an image for? And then I retrofitted a story to it. But just that idea of somebody is buying potato chips, they're going to go buy potato chips and they're loaded up their basket with potato chips and they're rolling towards the checkout aisle and some advertisement person jumps in front of them and says you should buy these potato chips that are already in their cart. Again in the digital analytics it seems so laughable among anyone who's at all deep in statistics or causality or causal inference and that for all those tools they'd say yep, you're going to get Some credit because you were the last touch, and then it spirals off into a well. Yeah, but if you get the right algorithmic last touch, shouldn't get all the. It's such an easy story to tell, and it's infuriating. So it had been a slow burn for me for a while. The fact that I was standing on stage just talking about that, to me indicates I certainly felt like a lot of people were still operating under that confusion. And, I mean, I was 15 years into my career where I think I was operating under that lack of understanding. Everything else, though, no changes. I had a nail that's all nailed down.
- [11:28]
C
Now you can just sit back and have a drink for the rest of the episode.
- [11:33]
D
What about you, Val? Like, what's one deeply held belief you had that you've changed your perspective on?
- [11:39]
C
So one, as I was thinking about this, I think one of the ones that is bigger because it has a lot of parts to it, it's kind of systemic in some ways, and I would say a change maybe in the past three years is that I used to think that there was this optimal model of operating inside of an organization and that a center of excellence meant that you were the most mature and that every organization should try to strive to build out a center of excellence as the ideal model for delivery of analytics and experimentation inside of the organization. And I think part of that was just because of the organizations that I had worked inside of that when there was a significant investment made that there was enough to create a guild, like team of people to kind of support the whole organization. So it was based on my own individual experiences. And also it felt like when I started consulting and I saw kind of pockets of analysts across the organization, it felt very scattered or they weren't using same definitions. And people had very different experiences engaging with those teams. And it felt like, well, that can't be the most mature. So we should all just try to strive for this Org model. But I think the more. The more time I spent in consulting, I started to realize that there was a lot of very mature organizations that worked off of a more decentralized model, or like the Hub and spoke or fractional roles, things like that. And that the. The way that the org is designed isn't necessarily the signal of maturity necessarily, and that those two things should be decoupled, but there was a lot of assessments. I don't know if you guys, like, remember at the time, like, you know, assess your maturity in, you know, whatever practice it was, and it did kind of always feel like it gave a lot of credit to orgs that had dedicated governance teams within the coe or, you know, people that supported the different pillars of the business. So I don't know, I don't know if that's one you guys have come across too, but that was definitely one for me that was pretty recent.
- [13:50]
A
So do you feel like maturity can be measured or just not measured easily?
- [13:54]
B
I.
- [13:54]
C
Yes, I think maturity can be measured, but not easily. I thought that, I thought that a lot of those assessments that you would take and you know, a lot of them were like lead generation tactics, that the gap analysis that would be provided back would be like the list of things that you could go do and that assumed you were striving towards a center of excellence model. So I thought that there was, there's a lot of things that, you know, not just me, that we're kind of saying like, hey, this is the right way to do things. But I don't, I don't think that those two things need to be coupled.
- [14:28]
D
Yeah, it's funny, the thing that has changed for me in this space is that I think previously I more had a deeply held belief that one way or the other was better. And this is probably like, is it, I don't know, experience or whatever age, whatever thing you want to call it. But I think over time you just start to be like, it's a series of trade offs and you either trade with this for this or they're opposed to this and cons to this. And it just depends on the specific business. It does make for kind of an apathetic answer though because you feel like you're always being like, well, it depends. But I feel like I used to have a strong opinion about the structure, especially centralized or decentralized, and I no longer do. I'm just like, we'll figure it out.
- [15:13]
A
Yeah, I wonder, is there an analogy to chasing the tool, that there's a tendency, I mean, organizations even broader than the center of excellence. We've all worked in organizations where things aren't going well. So it's a reorg and you really shuffle the reorg and it's like that if you blow everything up, you kind of just hope that stuff lands on the deck in a better, more organized way. And really it just, it gives you six months of everything being blown up and then you're kind of back, you're always going to have trade offs and like making that decision of saying, do we need to fundamentally shift our, how we do our analytics support or do we need to work within what we have and figure out how we should adjust it. That seems like the never ending question. Michael, you were going to say something though.
- [16:05]
B
Yeah. On the maturity topic, Val, you mentioned that I kind of came to a conclusion at one point. It was like, why do these never seem like they work right? And it's like, oh, because these maturity models always assume sort of like this linear function and no organization works that way. So it's like, like, oh, first you'll go from descriptive, then to prescriptive. And it's like, well, no you don't. In some areas you're already way out in front. In other areas you're barely getting started. And different functions in the business do different things. And so it never rang true whenever I would sit down and try to apply a model like that to any real world examples. And eventually it started being like, oh, pull back a little bit and realize this doesn't just. It's kind of fun for a PowerPoint slide, but it actually doesn't work in the real world this way. The real world is many points on a map or some type of quadrant chart or something. I don't know.
- [17:03]
A
Well, what do you think of the. And there's the maturity. Boy, I could not plus one that more. You're moving up some curve to prescriptive. Yes, but what about the maturity models that are giving you like six different lenses through which to evaluate governance and tooling and.
- [17:23]
B
Well, since I was pushing one of those at one point in time at Search Discovery, I love those.
- [17:31]
C
I think everyone, everyone's pushed those.
- [17:35]
D
I don't think I ever have.
- [17:36]
B
Yeah, but no, I definitely wish. Yeah, exactly. You're on one end of pushing somebody tried to sell. I mean, in a certain sense it does. Like, Tim, to your point, it does try to sort of like evaluate it on more axes to try to get a clearer picture. But even that is sort of a little difficult because it's really like. I don't know the right way to say it, but I've been running into this thing in the AI world lately where people are like, yes, if you just get all your data out there, you know, and you have the right quality of data, you could just apply AI to it and go to the moon. And it's sort of like, at what point do you know you have the right quality of data to be able to do that? Business people don't know that. So you're telling people they should be shooting for this crazy AI goal of using their data without having any kind of understanding about how they've ever achieved or arrived at that number. And so the same thing with maturity models. A little bit like a lot of times you're kind of throwing something out there that sort of doesn't necessarily have like measurables against it that are really going to work for them. So I think you can apply it, but you have to go through the weeds more. So no one model is going to work exactly the same for everybody, I guess.
- [18:49]
C
And I think a lot of like not every industry is going to be like analytics nirvana. Like scoring a 10 out of a 10 doesn't look the same for every industry or every company. And I think like that's back to most point the, that it depends. But I even think that like if I, you know, 10 years ago Val wrote the questions for a maturity survey would look very different than they are today. Right. Like today it's would be things about like, how are you ensuring that analytics moves at the speed of the business, whereas 10 years ago it might have been like, you know, how, what's, what's the governance for those UTMs? Like, how are you making sure that.
- [19:24]
B
Every marketer taxonomy are you using now.
- [19:27]
A
You would be asking like, are you using just multiple shades of pink in all of your data visualizations?
- [19:33]
C
This is true.
- [19:34]
A
That hasn't changed. That's still, that's a constant.
- [19:37]
B
Well, because some models you don't need to update. All right, Mo, what about you?
- [19:43]
D
I was going to, I was going to say though, just on Val's point, that that is one area that I feel I've definitely evolved my thinking, which is I'm not going to say that the answer was the answer or the measurement approach needed to be perfect. But I think over time I definitely index a lot more to the right answer or the right measurement approach for the business question. And that has to be at pace with the business. And I think previously, especially when I was the one doing more of the work, I'd be like, but it's not ready yet. It's not ready yet. I still need to look at this, I still need to do with that. And then I feel like now I see the other spectrum, which is like you work at it so long that the business has made a decision and moved past you. And so I wouldn't say it's like I've completely changed my thoughts. And I wouldn't say that it's binary, but on that spectrum I would say I wait a lot more now towards, let's use the measurement approach. That's relative to the size of the Business decision that we're making.
- [20:43]
B
All right. Yes. So what's fun is Tim and I in the background, just for everybody listening. Been working on creating sort of a custom GPT from all of our episodes in the past. And so I used that partially to come up with some of the ideas I found out in episode 10, which would have been way back in 2015. I said, nobody actually cares about privacy. And at that point in time, that was kind of true. But I think we've all been made to care. And certainly the tide has turned on that topic quite a bit. And even my own thinking on it has changed a lot over the years because as I've seen more and more, at first I was kind of like, all right, we don't need a bunch of regulations. We just all know what we need to do. But then you just see bad actor after bad actor after bad actor, and you're like, okay, well, never mind. We're not all nice people out here, so we've got to have some rules. So the road. But I did think that was pretty funny that, like, back in 2015, I was like, yeah, people talk about privacy, but nobody actually cares. And I was like, that said something.
- [21:48]
D
Similar to be fair, though. And I would still, like, again, it's not a binary thing. It's not either that people care or they don't. It's like, I would definitely say there's more attention and nuance to it now and folks are partially better informed.
- [22:04]
B
Yeah, well, there's still a long way to go, but I think because there's regulations now, people have to pay attention to it. And now are starting to ask the question of like, okay, yeah, are we doing things with privacy in mind and consent in mind?
- [22:20]
A
We're uniquely the non European set of co. I remember a very, very impactful presentation that I saw that was a European saying, yeah, you Americans, you don't fucking get it because you have not been so burned. It's deeply embedded in our DNA. And this is why the GDPR and the. What's the E. Privacy Act. I should know that. So I feel like Europe was out ahead on that front. But I have a. Like, related to that. I think I was on the board and it may have been what we were even talking about in episode 10, that there was a direct relationship between protecting privacy and the value you could get from the data. Like that tension of if you. The more you respect privacy, the less value you're going to get from your data. Because on one end you have total anonymity and no data even collected, therefore no value. And on the other end you have, you're following somebody around and tracking them online and offline. No privacy. And that's valuable. And I have radically changed my belief. I think we fetishize the user level tracking to the point that we get caught up in things like multi touch attribution. That's all this obsession with tracking a single user and then you're like, but wait a minute, what if we just have aggregate data or what if we do a mixed model? What if we run a controlled experiment where we don't need any of that? So I think I was marching along and I think there's still a lot of. It's an easy sell to say we're going to, for vendors to say we have a solution to basically violate people's privacy without violating a regulation or a law and ergo you will have more valuable data. And that last part is, I mean it's all bullshit. It's bullshit all the way along. But so I've radically, I've definitely changed. I'm much more on the let's do privacy by design. There's lots of value we can get without violating privacy or even getting close to it. Yeah.
- [24:46]
B
And for reference, episode 10 was that we were reviewing the CMO survey results. So there you go. Yeah.
- [24:53]
A
What's her name? Mary. No, what's the other survey? That was a CMO survey. I'm thinking of your state of the annual. There's some other big long report you read every year?
- [25:06]
B
I read a few, yeah.
- [25:07]
A
I used to believe that I could remember the things that Michael talks about all the time and now I clearly have aged out of that.
- [25:16]
B
There's going to be a GPT for that though.
- [25:18]
C
Yeah, don't worry.
- [25:20]
B
No problem.
- [25:21]
A
Also to clarify, when Michael said that he and I have been working on this, it is like 90% Michael and I just happened to have stumbled across having done something useful to plug into that.
- [25:31]
B
Well, I think.
- [25:32]
A
But it is really cool all the.
- [25:34]
B
Source material came from you because you rebuilt all of our transcripts. So I have to give you quite a bit of credit.
- [25:41]
C
Yeah.
- [25:42]
B
All right, who's got another one?
- [25:45]
C
Well, I actually had like a slightly small, just like story slash confession that feels like the right time to get this recorded confession and put out to the public. I had a neighbor that was a problem and so I researched this person and found out that they had a leadership role in marketing at their company that sold, we'll call them widgets. And so I may or may not have Created a link with some campaign parameters after I found out that they were using the Google stack that I asked lots of friends and family to hit, that included. We'll call his name Chad. That Chad. Chad drinks warm milk. Just because I loved envisioning, like, someone on his team being like, what is this, this campaign that people are coming from and why are they talking about our boss and how he likes to drink more milk? But, yeah, I don't know if anything ever happened from that. But, you know, just because we're talking about privacy and parameters, that's okay.
- [26:48]
A
Most data doesn't ever get looked at. Yeah.
- [26:50]
B
Yeah.
- [26:52]
D
Attempted that to me.
- [26:53]
B
I think we've all done UTM prints.
- [26:56]
D
That's so funny.
- [26:57]
B
Yeah, I love it.
- [26:57]
C
Yeah, it's good. Good.
- [27:01]
B
Analytics.
- [27:02]
C
I'll never change my opinion on that.
- [27:04]
B
That's right.
- [27:04]
A
We use a URL shortener every time we post in the measure Slack largely to mitigate shenanigans. There's a step in our production process that is largely around not making our campaign tracking parameters is visible.
- [27:20]
B
Yeah. And frankly, there's quite a few people who might hear this and take that as a challenge. And we just urge you not to please do it.
- [27:27]
A
Anyway, we're not looking at the analytics.
- [27:31]
B
Go for it.
- [27:32]
C
It's going to take a lot for that. Anomaly detection.
- [27:35]
B
That's true. We don't care. We're not running into it.
- [27:38]
A
We had a whole episode that was the last episode around small data. And after we finished recording, Julie said, you know, we have a number of small data sets related to the podcast. And it was just occurred to me at the very end of the episode, Shade. Fair point.
- [27:58]
D
So I don't know if mine is a confession. It feels a bit like one because as I think about it, I sometimes feel like I'm betraying data people when I say it. And I said it to a group of, like, data experts at my work the other day. And, like, one person was like, yeah, that's fair. And I'm still. Anyway, so obviously I'm going to say it on a podcast where everyone can judge me publicly and please rate and review us. So I actually give a lot of credit to Kathleen Malley for this because I think, you know how sometimes someone gives you an analogy, and when you get the analogy, it, like, just like you're like, yes. So we were talking about Dada as part of a meal and the fact that, like, in some businesses, data is like the garnish that goes on at the end, or it's the salt and the pepper. Just to give it the extra like, nice taste. And in lots of companies, the full other end of the spectrum is like, they assume data is the meal. That if you make a decision without data, then it's a stupid decision. And I feel like I definitely fell prey to this about this idea that data was right or that if you made good decisions it was because you were using a data informed decision. And what I would say now that I strive for a lot more is data is part of the meal. So it might be the roast potatoes and then there's some roast meat and some vegetables that go with it. But it's one of the things that you use as a business leader to make a decision. And for some people that might take up a bit more of the plate than others. But I feel like it's really. Yeah, it's something that I found to be that I've really changed my opinion on because the context of this discussion is some folks were talking about like, well, this experiment result said that this thing was bad and therefore we shouldn't do it. And I'm like, but that's one part of the meal. And there are all of these other things and sometimes it's history, sometimes it's strategic intuition, sometimes it's like there's another form of research or there are so many other facets that come into making a decision and I'm going to get off my fucking soapbox now. Step down.
- [30:09]
C
No, it's good. You could take it further and be like, you know, grandma's recipe that was passed down from generations as like the intuition component of it. It. But I like that. Yes. For those, those mashed potatoes or whatever it was.
- [30:23]
A
Yeah, you've, you've hit on the, the, the people who are. I heard this anecdote when I was talking to a past colleague who said, oh, the CMO wants the AI just to, you know, in the, with their speaker, in their shower in the morning, you know, just go through the data and tell them what they should do. Right. That's again this idea that if you just have enough data and it's clean enough and you run it through the right tool or model, that is the start and the end. That is everything. And it never has been and it never will be. And I feel like that's. And it's not even a progression of like. Yes, but if Data was only 20% of how we made the decision now, then tomorrow it should be 30% and next quarter it should be 40. And eventually just the data, data will do everything. Is it Feels like a wreck. People are on. That is totally misguided. I think that's the same thing you're saying.
- [31:29]
D
So, wait, do people agree with me 100%?
- [31:34]
A
Huh?
- [31:35]
D
I don't know why I expected that to be controversial.
- [31:38]
C
Not with this crew. Unless Michael's, like, biting his tongue over there.
- [31:43]
B
No, no, no.
- [31:43]
A
He's like. Let me ask. Let me ask our custom GPT what it has to say. Y. You know, on episode 14.
- [31:51]
B
That's right.
- [31:51]
A
Mo, you said.
- [31:54]
B
Yeah. Let's not go digging into the. Into our previous comments.
- [31:57]
D
Please do not. Please do not. Contradiction to shit that I used to say, because I said lots of things that were wrong.
- [32:03]
C
That's the whole perspective of this episode as we evolve.
- [32:06]
B
There you go.
- [32:08]
A
I mean, getting into, like, the confessional. The confessional term, like, this is like a trivial one. But I did. As somebody who deeply, deeply cares about effective data visualization, I did have a phase when I first thought I deeply cared about it, where I put a gradient background on every chart that I did because I thought that added class and panache and credibility, because I knew how to put a gradient background on a chart in Excel that that would make my charts stand out. And I'm happy to say it was a brief phase. It was a long time ago.
- [32:46]
D
You used to have presentations with lots of moving things, too. Do you remember that?
- [32:50]
A
That was the prezi phase.
- [32:52]
C
That was the prezi phase.
- [32:53]
B
We all had a prezi phase.
- [32:54]
C
We all had a prezi phase.
- [32:55]
A
John Lovett was right up there on a pedestal. And if John Levitt was doing it, then I was going to do it too. And if people got motion sick in the room, then so be it.
- [33:04]
B
Yeah.
- [33:05]
C
Oh, my gosh.
- [33:06]
A
They should take a Dramamine. What's the motion sickness?
- [33:10]
D
Yeah.
- [33:11]
B
Yeah, there you go. Yeah. I used to think that you could just sit down with your data and come up with insights. You know, you could just. Valuable insights were in the data. You just spend time with your data and you'll just come up with cool stuff. And it's interesting because, like, I used to be in the role as an analyst where my. My job was coming up with insights. And I thought that's kind of what we were doing. But really what we were doing was a little more nuanced. And as time has gone on, I've sort of learned like, no, you can't just sit in front of a data set and the data itself is valuable. It's like, no, it's. First there has to be a perspective and a question, and then There is potentially some information or knowledge the data can bring to you. But it was like. It was sort of like a process of learning all the steps that were happening, not just looking at the data, but there were other steps that were part of that process. But at first, I think I would actually think that, yeah, I could just sit here, and I do remember sitting at my desk being like, all right, just come up with something cool. Like, just, like, find something in the data.
- [34:19]
D
Oh, I've done that.
- [34:20]
B
Yeah. It's.
- [34:21]
C
I think we.
- [34:22]
B
Honestly. But, like, that's what I thought.
- [34:24]
D
Tim and I having a huge argument once about monthly reports. And I was. He was against. For that reason.
- [34:36]
B
I. Just one argument.
- [34:38]
A
I mean, I knew I was well past that. I think I was at that point, at one point, because that was kind of the draw to doing stuff with data is like, oh, I can poke around, and once I learn my way around our tools and our data, I can do that. But I remember a client telling me, and the fact that my head wanted to explode. He was like, look, we have a biweekly meeting with these stakeholders. And this was like multiple groups. He managed the team that everyone on his team with different brands had a biweekly meeting. And he was like, look, we have a biweekly meeting. We've got to just see. We got to shuffle through the data. We got to find something. We got to find something new to show them every time. Which totally predictably, if you asked who they were presenting to, they were like, every other week, we have to sit in a meeting and watch charts be thrown at us with the analysts looking at us like, is there something useful here? Is there something useful here? Oh, look, look, this number went up because we did X, which is not a surprise to anyone. But we didn't. We didn't do X two weeks ago, so we did it this time. And I was so. I was done. I was so done with that guy at that point. And then he continued to be a client for, like, another three years. It was rough.
- [35:56]
B
What amazing moments in time where that haphazard analysis somehow interconnected with the interest of the moment. And everyone was, like, so focused on what you're bringing to the table.
- [36:08]
A
Well, and that's what gets. That's what gets shouted from the rooftops. Expectation.
- [36:13]
B
And then, boom.
- [36:13]
A
That's the case to happen. Yeah.
- [36:15]
B
Every week. Yeah.
- [36:17]
A
And it's like, no, that was.
- [36:18]
D
I still believe in monthly reports.
- [36:22]
A
I'm not. I mean, I can mount my little soapbox very quickly on that, but I'm not Opposed to monthly reports. If mo. If you're remembering me saying monthly reports were.
- [36:33]
D
No, I think. I think what we talked about was this idea that, like, every month the data team are going to come and bring insights. And I feel like probably where we landed maturely, obviously was an agreement that that's not feasible or possible, but there is still value in meeting regularly with your stakeholders to talk about how things are performing and that sort of stuff and discuss, I don't know, priorities for the next chunk of time. But speaking of priorities, there is one that VAL put on here, which I also used to have as a deeply held belief.
- [37:09]
A
Do we have to go to another one? That is Tim having heated discussion off.
- [37:14]
D
Mic about ticketing systems. Do you want to say more, Val?
- [37:20]
C
Yeah, I believed that having, like a ticketing system. I'm using air quotes here, something like a JIRA like, interface for people to enter. The question of the moment was a great front end to the analytics experimentation team. With the huge caveat that I never thought there would be no conversation once ticket was entered before work began. As I'm like, bracing myself, I've definitely evolved on this. And that was my. When this was. I was kind of growing up in digital analytics and I was at the American Medical association, and we had 11 different lines of business that were being supported by a small but mighty team of three people. And so we were, like, doing a lot at first to, like, drum up interest and like, for people to understand what this capability was and the ways that we could integrate. And then all of a sudden it was like the interest was there and we were like, oh, holy, how do we handle all this demand? And so we're like, I've got a great idea. Which I think is like something that a lot of people turn to. Right? But again, not as a replacement for that discussion. However, Tim and I did go a couple rounds, as previously mentioned on, I think, at least two other episodes that some people got to witness because I think Tim had a different deeply held belief.
- [38:42]
A
It was one of those moments where, like, everybody around, like when you're sitting in a restaurant and there's like a couple having a heated. Like, the marriage is falling apart and they are intensely disagreeing and they have no awareness that they're out at a nice. I couldn't even tell you. I could guess who a couple of the other people who were witnessing were. But I think everybody got quiet and we were.
- [39:06]
C
It was just like everyone's forks were. Was like.
- [39:11]
D
Oh, wait, this was actually at a restaurant. Or metaphorically At a restaurant. Oh, wow.
- [39:16]
C
Very physically.
- [39:17]
B
It was.
- [39:17]
A
It was after we'd been with a client all day and the client had a ticketing system. And I was like, well, we gotta get that misconception off. And Val's like, I think it could work.
- [39:26]
C
And I think we have to make some edits.
- [39:30]
A
Oh, we should update the input intake flash. And I'm like, whoa, whoa.
- [39:36]
D
So, Tim, just to shorten it, for a person on my team who deeply believes in the intake system and is an avid listener, talk us through your reasoning. Why is your deeply held belief that you shouldn't have a ticketing system?
- [39:49]
A
Well, so not to pull back and say that we do have a document we're kind of referring to, but I want to throw it to Michael because there was, to me, a hugely related thing that he had and what he used to believe. And we'll see if I can get him to tee that up. And that's kind of my main as.
- [40:07]
B
It relates to ticketing systems.
- [40:09]
A
I believe it does. I think you could tell by the fact that I've made it 30.5.
- [40:13]
B
Yeah, no, I see the font in the notes. Yeah. So the thing I believed early on in my career, and I think there was a couple of unfortunate events that kind of made me go down that path for a while, was that business users knew what they needed from the analytics tool or what metrics. Metrics and reporting they needed. So I would just try to answer their question, like, exactly how they asked it. So it's like, oh, yeah, here is time on site for all these pages. Now you'll know exactly what to do with this, or whatever the terrible request was. And it's funny because really, really early in my career, what I did day in, day out was a lot of web trends, implementations, and I had a client client that I did the implementation for. And so when you do the implementation, you'll go build out a set of initial reports. And they had a list of reports that they wanted, and I made some extra ones that I thought were nice. And they called up the company and were like, we don't like these reports and we don't know what they are. So I had to go back out to that client and take those away. So that was sort of like a learning moment. But that was sort of one of those things where it's like, like, okay, well, just give people what they ask for. Don't ask too many questions. And as time went on, I learned more that, you know, there's some discovery that you need to do with People when they ask for digital data, especially because there's a pretty big divide or disconnect between what people are usually asking for and what they actually need. And so that was a learning process for me.
- [41:48]
D
I'm a bit scared to poke the bear. I'm scared to poke the bear, but I'm going to do it anyway.
- [41:52]
B
Okay.
- [41:53]
A
Okay.
- [41:54]
D
That could happen in a email, a Slack message, a ticketing system. Like business users not knowing or not asking for the right thing or not knowing the question, like that's irrelevant of how they ask. That's just the, like the process to do the asking.
- [42:11]
A
Well, yeah, but when, when somebody sends an email, when you put up. When you put a ticketing system up, I mean, I know how it happens. It's like we have so many coming in, they're incomplete. If they're in an email or if they're in a Slack, there's a problem. So if we just give the structure of what we really need and we put it in a ticketing system, then it'll come in. The math never works. You've got three analysts or five analysts or 10 analysts. If you have 10 analysts, they're supporting 200 people. 98% of those intake requests is going to take. Take two to five minutes to put the request in and one to 20 hours to resolve it. And the business user is doing the best they can. They're trying to be as prescriptive and helpful as possible. And they even believe because they filled out your damn form like you had seven things you wanted to know. What strategic goal does it tie to? What area of the site? Is this an analysis or a report? Is it a dashboard update? So they are. They are doing their best. But you've put a structure in front of them that implies that if they put this in, you will be good. And I don't know why everybody's laughing so hard.
- [43:35]
C
I'm not laughing at you, Tim. At all.
- [43:37]
A
I don't want to break the fourth wall. I think Mo or Val. I don't know who's going to pee themselves first there laughing so hard.
- [43:46]
C
No, it's.
- [43:46]
A
Thou says it's going to be her.
- [43:48]
D
No, I'm pretty close.
- [43:55]
B
I've never seen someone take longer.
- [43:58]
C
I've never seen someone take longer to 6F in color.
- [44:08]
B
I listen. Oh my God. We're not under the kind of time pressure to do it. I was just.
- [44:13]
C
All we can do hear is.
- [44:18]
D
Highlight un highlight bigger, smaller back space bullet. I tried so hard to ignore it.
- [44:28]
B
I'm so sorry.
- [44:29]
C
I saw Smile. I'm like, is she watching the same thing?
- [44:32]
B
I am.
- [44:36]
A
Enough of a tear.
- [44:38]
B
Okay.
- [44:39]
D
I really needed that laugh. That was the best moment of my day.
- [44:44]
C
I wish we could have played that back.
- [44:47]
B
Yeah, much, so much.
- [44:48]
C
But that was a well put kind of summary on that, Tim.
- [44:51]
A
Yeah, I mean, when you put the gate to the point of saying, after you've put it in, then we'll reach out to it. Like, you've already put a burden on them to in their mind structure things and like, this is what I filled in your fucking form. You know, like, why are you now coming back and wanting to engage with me? Like, you're actually making it harder on the analyst to say, no, what do you really want? So I think that's, that's to me, like, way, way better.
- [45:17]
C
To counterpoint there though is like, oh, I don't want you to like, feel like you need to schedule time with me or whatever. Like when, when you're in your flow, in your moment. I just need three critical pieces of information or whatever I had at the time. And if you can complete these things, we won't lose the thread that like, you don't have to have the mental weight of remembering that we don't have to, whatever. So it was more about like, how is it? And then I don't. As the manager of the team, I can. All the things that come in and kind of think about a way to prioritize and triage because we were working so collaboratively to support the business. So just to say that, like, I understand what you're saying there, but we did always go back and that was always a good, valuable conversation. Use the time. If we went back.
- [46:02]
A
But you're always setting that, that that flow is coming in will be a. You're never going to be on top of it. So you're then setting up the organization to always be in a point of saying, this is the stuff we're not going to get to because it's not, not a priority. And when they say, I just want these three critical things, I read it. I skimmed an article that said these are things you should. It was like, never say in a work context. And I was like, oh, this is related to one of them. Which is, this will just take a second. Which is the same to me of saying, hey, I just have a quick. Can you just pull these three critical things? Well, yeah, one of them's on our dashboard. I can pull it. The second one's super easy to get to. The third one is damn near impossible. I Could get close if I spend 12 hours on it, I can't pull those three things. So the business users, or the flip side, the business users ask for what they know is readily accessible. And there's something else that's readily accessible that would be way more.
- [46:56]
D
That's the job of a good data scientist, is to realize that.
- [47:00]
C
I don't know.
- [47:01]
D
I think one of the arguments that I often get is that it does introduce some friction, which is a good thing, is because stakeholders are stupid. Shit, sorry, I'll be nice. Sometimes, like, they'll ask questions that are interesting but not a good use of time or exist somewhere else or something else. And by creating this friction of forcing someone to fill out a form to give thought to what they're asking for, to not drop a sentence, which I do to my poor team all the time of like, yo, can you find me this thing? And they're like, mo. It stops that behavior, right? Because you have to fill it out and be like, what am I asking for? What? Like, it's. It does require thought. And so, yeah, I don't think there's a wrong or a right here, but I just keep poking.
- [47:47]
A
Oh, there's a wrong.
- [47:49]
C
Oh, okay.
- [47:50]
E
There is a wrong.
- [47:51]
A
I mean, so let me paint what I think is realistic and practical and it goes a little bit to the center of excellence. Like, much, much better for there to be something of a mind meld or at least a deep enough relationship so that there is a analyst or a data scientist, depending on the role, so deeply kind of collaborating with their business partners that instead they're kind of coming up with what do we really need to know? What are our ideas? Like, it's starting at that level. And I get it. There's the. I mean, it's such a knee jerk, easy thing. I've got presentations that are available online. Sometimes people just need a number because they're heading into a meeting with executives. Sure, that's a. There are times where I just need this quick number. So put those aside. But I think almost everything else would be much, much better served by having a thoughtful coming together of what are we really trying to do here?
- [48:53]
D
But Tim, they don't have time for all those conversations.
- [48:56]
A
They don't have time to not have those conversations. That was intentional.
- [49:00]
D
That was intentional.
- [49:01]
A
Okay, okay.
- [49:02]
D
Sorry.
- [49:02]
A
You were like, I'll push the this button and watch that.
- [49:05]
E
You guys are mean.
- [49:07]
B
You're mean. What I hear you talking about, Tim, is the need for a data sommelier. That's exact hoes.
- [49:14]
A
I was afraid you were going to say an analytics translator. Translator Data sommelier.
- [49:19]
B
Data sommelier. But sometimes you just need a beer. And I think that's the thing is that we're getting into is like, sometimes you just need a Bud Light and then other times you need a Chateau de Neuf. Pap. Whatever.
- [49:32]
A
Yeah, I mean, it goes to most. It goes to Mo's point of saying, sometimes giving an imperfect answer quickly is better than giving the perfect answer too late. I just totally paraphrased. I think I might have said something shorter than you said it earlier, which has never happened in my life. But you can only do that if you've got the trust in the relationship and the understanding of the question behind the question to be able to. To do that. And I would. Yeah, I don't think it's more time consuming. I think it's net less time consuming.
- [50:07]
C
I actually, I. When we had originally implemented this, when I always had, like, worked with clients who were using it as we. I saw it as a way of removing friction. So it's interesting and I, I understand why you guys are. Are saying that, but again, like wanting to keep it like within their flow. But it was so interesting the number of times when we would get a question and they would say, like, can you look at all the content in like slash whatever, you know, section of the site content is popping or, you know, whatever kind of generic question. And so then we would look at that and say, like, look at it like, forensically. Like, I wonder if they're thinking about revamping the content within this section of the site. Or I wonder if they're thinking about, you know, launching a new campaign. And so we'd have all these hypotheses about why they even asked the question just to go back to be like, so what is this really rooted in, right? To kind of like, you know, start the conversation there. Which, which is funny because if we had just started there and not with the interface to Tim's point, we wouldn't have to play that little game. But yeah, I guess again, intent, you know, being one thing, not always there, showing up an execution. Sometimes people do forms because they think it's reducing the friction. Because the worst thing is it just goes away, just off into the ether. Right? Like, that was what I was always trying to fix.
- [51:25]
D
I do also think though, for more junior folk, the reality is when you have a team of 40, you have to have some way to manage this stuff because otherwise you have some data scientists that are just getting DMs constantly and they tend to be the folks who are giving it great thought. But it's also partially about teaching younger data scientists what are the questions we should put energy and time into? It helps them sometimes say, yeah, I hear that you've got this question, but I'm going to focus on, like, it does help with prioritization.
- [52:00]
A
And I absolutely believe there should be a log of what's the work that's been done at the proper level of fidelity to know to build that little data set. To say, these are all the requests we fulfilled. Which, if you take that is a nice upside of having a ticketing system, is you have a data set being built and your requestors, your business partners are helping you build. But I think that can be built that doesn't require an intake form to track that.
- [52:27]
B
As a data sommelier, you keep a log of all your tasting notes. This table has exquisite robustness. Anyways, okay, we have to start to wrap up because. Oh, yeah, we've been going at this for a little while now. Yeah, I know.
- [52:50]
D
There's so many that we haven't discussed.
- [52:53]
B
As is always the case. Yeah, we should do a part two.
- [52:58]
D
Okay, one. One last one. One last one. I'm just gonna say it out loud, and I want, like, a quick, quick, hot take. You should test everything.
- [53:06]
C
Thumbs down.
- [53:08]
B
Yeah, that is a good one. Because it's so easy to agree with. Like, it's like, especially because so many of the big Internet companies are like, we run 10,000 tests, those kinds of things. And so you're like, oh, well, then testing velocity is the thing we should really be trying to do. Like, oh, we just gotta test more.
- [53:27]
D
And then do you know, I think you should test everything replaced. You should track everything.
- [53:32]
B
Well, I think people who think you should track everything also think you should test everything.
- [53:37]
C
I mean, there was a book. Wasn't that a Chris Goward book? Which I know that he's like, you know, since, like, moved on from. But you should test. That was the title of the book. That was like a. It was a pretty big, big concept at the time. I do remember that's, like, when I was really getting into. I was like, yeah, that's a really good one.
- [53:53]
A
I am. I am three weeks away from moving on from analytics the right way that I'll. I'll move on from my book and.
- [53:59]
B
Say, oh, well, let's save that for part two. All right, well, we do have to start to wrap up. And that makes me delighted because now it's time for a quick break with our friend Michael Kaminsky from Recast. There's a media mix modeling and GLF platform that helps teams forecast accurately and make better decisions. And as you know, Michael's been sharing some bite sized marketing science lessons over the last couple of months in the next couple months to help you measure smarter. So, over to you, Michael.
- [54:28]
E
Let's talk about one of the most misinterpreted terms in analytics, statistical significance. Many people say that the results of an experiment were statsig, but that abbreviation is misleading. Instead you should say the full sentence. We found strong evidence the difference between treatment and control was different from 00. Assuming all of the model's assumptions are correct and allowing for a 5% false positive rate. And even when communicating statistical significance in this way, you should be aware of its limitations. First, it assumes that the underlying model is the true model. In the case of analyzing a simple two cell experiment, that might be the case. But in more complex models, we're often operating in a world where we know that our model isn't the true model of the world. Just an approximation. So P values tell us very little. Second, it assumes that 0 is the most important comparison. Is it 0 or is it not 0? In business, it's very often the case that we care about something other than zero. We might care about whether or not the difference is sufficient to make A versus B more profitable. And simply knowing if the difference is greater than zero doesn't actually help us make the right decision. So in general, I'd say focus on reporting confidence intervals which give a range of likely outcomes instead of P values. And if you're going to show share P values, make sure you communicate your assumptions and what you're comparing against.
- [55:41]
B
Thanks, Michael. And for those who haven't heard, our friends at Recast just launched their new incrementality testing platform, Geolift by Recast. It's a simple, powerful way for marketing and data teams to measure the true impact of their advertising spend. And even better, you can use it completely free for six months. Just visit www.getrecast.com geolift g e o l I f t. That's getrecast.com gift geolift and you can start your trial today. So go check that out. That's actually really cool. All right, now we want to jump into last calls, something that might be of interest to our listeners that we go around the horn and share Mo, why don't we start with you? What's your last call?
- [56:25]
D
Well, we know that I binge listen to things in a quick succession, so I'll go Through like a podcast or a series or an author that's just kind of like how I work. And I have been binging Work Life by Adam Grant. And there are two episodes in particular that I thought were really interesting. One was called the Truth about the Attention Crisis. I don't know if that's the exact name, but that was the topic and it was with a historian, Daniel Imaware Imawar. I'm going to butcher someone else's name on a podcast again. This is what I do. But he wrote a really interesting article about like the fact that he thinks it's a myth that we can't pay attention to anything. And he talks about examples like video games and opera when first opera, you know, was the, the, the cultural pastime and books. And yeah, it was just one that I thought was like a bit of a different perspective on like the usual phones are killing our brains type of thing. And there was also a really interesting one on the case against personal branding. But the one that I probably most enjoyed was one with Linda Babcock and it was about protecting. And she specifically talks about a no club that she's in and how women take on more non promotable tasks at work and what are some strategies that you can employ to guard your time. So highly recommend checking out the podcast.
- [57:46]
C
That sounds good.
- [57:48]
B
Awesome. All right, Tim, what about you?
- [57:52]
A
So I'm going to start off by acknowledging, to paraphrase Mike Birbiglia on his last latest special, most of my last calls are for you, but this last call is for me. I'm going to Last Call, a podcast that is no more. But I don't think I realized exactly how much I was deeply into this podcast until it actually went off the air with the last episode being with Barack Obama. So this is really more for that handful of listeners who, who had slowly come up becoming WTF with Marc Maron fans like I had over the years. I remember he'd been doing a thought for a long time when he had Barack Obama on the first time.
- [58:43]
B
He.
- [58:44]
A
Has had his slow burn. And his last episode was almost exactly a month ago. I've listened to so many. I feel like I've learned about storytelling, I've learned about addiction, I have learned about thinking deeply. And I will warn you now that at the very end of our outtakes, there's probably going to be my final homage to Marc Maron. And for all of you who are like, what the hell are you talking about? That's fine, just skip it. For those handful of you who Are like, I fucking love WTF with Marc Maron, and I, too, am grappling with his departure. This is for you and it's for me. So that's my last call.
- [59:28]
B
Very nice. Very nice. All right, Val, what about you? What's your last call?
- [59:33]
C
Yeah, so this is another UX Collective Medium post. And so this one is, like, from that design world. I love. I love this one a lot. It's a good follow. But this was a specific article by Fabriza Osceolo called Building Trust in Opaque Systems. Why? The better AI gets at conversation, the worse we get at questioning it, which is something that we've heard and talked about before. But this take, I'm telling you, is really good. About halfway down, there's this concept where they're talking about scaffolding over crutches. And I'll just read this one sentence so that this reveals the fundamental flaw in the current approach. They're asking us to bear the weight of responsible use while providing tools designed to discourage the. The very skepticism that they require. And so there's also this whole section at the bottom that's like, reimagining this world about what if it was asking you to have curiosity about the answers provided when it was appropriate. And I thought it was just really interesting and well thought out. And I was like, yeah, they should absolutely do some of this because you think, like, oh, you know, you just can't trust it. But, like, what does that actually mean in practice? And so I thought it was really interesting. Interesting. And then my, like, 0.5 plus additional last call, because it made me think of. It is a podcast that I've been addicted to for years is the Daily. And there was one in mid September called Trapped in a chatgpt Spiral, a look at just how close and dangerous our relationships with AI can be. And everyone's heard, like, you know, getting too close or using it as a therapist and this crutch or, you know, having these, like, relationships, and you think, oh, that's just for, like. Like, X type of people. Like, whatever your brain kind of assumes is the type of person who would be susceptible to that. But I thought that this podcast was really interesting because it actually interviews some of the people who have fallen victim to that, and they kind of express, I'm just not that kind of person who could get duped. But I still got caught up in it. And so, anyways, that's a fun, quick, interesting listen as well. So that's my 1.5 last calls for today. How about you, Michael?
- [61:49]
B
Well, you know, if you're like me, you're constantly reading articles and things online, or you. Somebody sends you something, you're like, oh, yeah, I'm definitely going to get to that and read it at some point. And then it stays in an open tab for about three weeks and you never actually get to it.
- [62:06]
A
Or.
- [62:06]
B
And then you. Or you do read a great article and you're like, where was that article? I can't remember what it was. So I recently stumbled across something that I've been looking for for a long time, and somebody built it, which is a personal data la, which is a little bookmarking tool that you just go in and say, oh, yeah, I'm reading this. You click it and it saves the entire page into your own little personal data store, which you can then search later and also connect with mcp. So you could even do some AI stuff with it if you want down the road. So anyways, it's called idexa and so we'll share a link to that because I just started using it and started saving some pages. The first page I saved was. Was Recast Geolift page at Get Recast Geolift. No, that was just accidental. But I did save that page anyways. Little plug for our sponsor there. Anyways. But anyways, it's super helpful for me because now I can close like 30 tabs and know that they're still saved somewhere where I can get back to them at some point. So that's nice. And as you've been listening, maybe you've got thoughts on stuff you used to believe and found out later on some new way of looking at it in the data world. We'd love to hear from you. The best way to do that is you can reach out to us on LinkedIn or the measure Slack chat group, or through email at contactalyticshour IO. And if you're going to listen to the show, leave us a review or a rating. We really appreciate it. And you can also request stickers for the show because Tim loves mailing them out. And you can do that on AnalyticsHour IO. There's a form you can fill out and we can send you stickers that you put on your laptop or on your phone case or wherever you put stickers for your favorite podcasts. So, yeah. All right, well, thank you all for doing this show. We didn't have, like, a specific guest. You're all the guest today, so congratulations. But I think this was fun. I saw a lot of fun, so thank you. And I think I speak for all of my co hosts, Mo, Tim, and Val, and I say, no matter what your priors and how they're changing over time, keep analyzing.
- [64:19]
A
Thanks for listening. Let's keep the conversation going with your comments, suggestions, and questions on Twitter @nalyticshour, on the web at analyticshour.IO, our LinkedIn group and the Measure Chat Slack group. Music for the podcast by Josh Grover.
- [64:36]
E
Worst.
- [64:37]
A
Those smart guys wanted to fit in, so they made up a term called analytics. Analytics don't work.
- [64:44]
B
Do the analytics say, go for it.
- [64:46]
A
No matter who's going for it.
- [64:47]
B
So if you and I were on the field, the analytics say, go for it. It's the stupidest, laziest, lamest thing I've ever heard.
- [64:54]
C
For reasoning in competition, Tony, let this hit the cutting room floor.
- [65:02]
B
Let it hit the.
- [65:05]
A
When Michael's the guest.
- [65:07]
B
Well, I mean, we can be like, you've got 60 seconds to teach us something. Go.
- [65:11]
C
That could be no pressure. I think he'd be like, no problem.
- [65:19]
B
Yeah. Yes, Mo, you can. I'll start the show. When you get back. It'll be your turn to say, how you going? I'm just kidding.
- [65:28]
A
There is a loss in that, right?
- [65:30]
B
Nope.
- [65:30]
A
That's not the.
- [65:31]
B
Nothing at all. It's perfect.
- [65:36]
C
The act of twitching.
- [65:38]
A
Some people thinking I was just on my computer the whole time, but I was like, I'm the one who has.
- [65:42]
B
The best Power Hour buddies yucking it up. It's a thing we do. It's called the Back Channel. We just make fun of anybody who's speaking.
- [65:52]
C
If I ever see the two owners of who is the. The company that shut this down, I was like, I will literally throat punch them. That, like, that's the level of aggression that I feel. She's like, oh, never expected to hear something like that come out of your mouth. And I'm like, and I'm not joking.
- [66:09]
B
That's the Chicago way.
- [66:11]
A
So then there's the question. Can that go in the outtakes?
- [66:16]
C
Yes.
- [66:18]
A
Okay. So, you know, Tony, clip it down to whatever you think is the funniest, but definitely throat punch.
- [66:25]
B
That.
- [66:25]
A
That definitely needs to be in the perfect.
- [66:28]
B
So, yeah, okay. Definitely gonna throw it to you.
- [66:30]
A
Your intros are getting, like, more concise, and mine are just getting longer.
- [66:35]
D
I was thinking that too. I observed that.
- [66:41]
B
I don't know. It's just an idea.
- [66:42]
C
So you're saying that Tim sounds.
- [66:44]
B
No, no. AI.
- [66:45]
A
I'm sorry.
- [66:46]
B
Saying that I use EM dashes and.
- [66:48]
A
I am, like, barfing out.
- [66:50]
B
What I'm saying is basically an AI in terms of a super intelligence. Okay.
- [67:00]
A
Content for, like, one medium length blog post, and it became an entire book. Like, that's. That's what you're saying?
- [67:08]
B
Well, that's not encouraging me to start reading analytics the right way.
- [67:13]
C
I was reading it, but.
- [67:17]
B
At least I'm honest.
- [67:28]
A
Rock flag and cat angels everywhere.