#035: Is Data Science the Future of Web Analytics?

Are you a data scientist? Have you pondered whether you’re really a growth hacker? Well…get over yourself! Picking up on a debate that started onstage at eMetrics, Michael, Jim, and Tim discuss whether a fundamental shift in the role (and requisite skills) of the web analyst are changing. You know, getting more “science-y” (if “science” is “more technical and more maths”). all in 2,852 seconds (each second of which can be pulled into R and used to build a predictive model showing the expected ROI of listening to future episodes; at least, we assume that’s what a data scientist could do).

Episode Transcript

The following is a straight-up machine translation. It has not been human-reviewed or human-corrected. However, we did replace the original transcription, produced in 2017, with an updated one produced using OpenAI’s WhisperX in 2025, which, trust us, is much, much better than the original. Still, we apologize on behalf of the machines for any text that winds up being incorrect, nonsensical, or offensive. We have asked the machine to do better, but it simply responds with, “I’m sorry, Dave. I’m afraid I can’t do that.”

00:00:04.00 [Announcer]: Welcome to the Digital Analytics Power Hour. Three analytics pros and the occasional guest discussing digital analytics issues of the day. Find them on Facebook at facebook.com forward slash analytics hour. And now, the Digital Analytics Power Hour.

00:00:25.79 [Michael Helbling]: Hi everyone, welcome to the Digital Analytics Power Hour. This is Episode 35. Here we go once more into the breach. All of us digital analysts are looking over our shoulders and we’re seeing the rise of the data scientist. And a lot of people would say that in five years, that’s our future. All of us will be doing data science or not doing this whole digital analytics thing anymore. And I think maybe there’s some merit to that. Maybe there isn’t. And so, This show is all about is data science the future of digital analytics? And of course, this topic was sort of inspired by somebody who asked a question on our last show at E-metrics, which is Boaz Velozny. He asked kind of this question. We covered it really loosely, and we liked it so much, we wanted to come back around for another go. So joining me on this journey is your very good friend and neighbor, Tim Wilson.

00:01:30.47 [Tim Wilson]: Taking, taking the position of yes is the answer. What’s the debate? What are the debates? We’ll get to that later. So you have pros and cons.

00:01:39.92 [Michael Helbling]: We have four in a game position. No, let me introduce our other host, Jim Cain. Hey, Jim. Hello. Awesome. And I’m Michael Helblin. All right. So guys, this is an interesting topic. Data scientists, is that what we’re all going to become? Is that the future of the analyst? Go. And maybe do we want to start with a definition of, no, we don’t want to start with a definition.

00:02:05.55 [Tim Wilson]: That would be just an hour and a half arguing about the definition of a, I don’t think it can be clearly defined at this point.

00:02:12.80 [Michael Helbling]: Just everyone listening, go ahead and insert episode two right here. No, I don’t remember which episode that was. Tim, I feel like maybe you are heading up the camp that says, yes, this is the future. Do you want to put some meat on those bones?

00:02:33.85 [Tim Wilson]: It’s the affirmative camp. That’s what it is. I never did debate. Oh, okay. Well, so I don’t think we can, I think some parameters of a definition, you know, there’s one school of thought that’s like, I saw a cartoon that was, you know, what’s interesting to data analysts and the data scientists, and it says about 30 grand a year, which, you know, there’s, hey, looking for a job, put data scientists in your title on LinkedIn. I get that there is all of the, what is this? It’s a fancy term. I think even Nate Silver sort of took a dig at it and was like, look, data scientist is just somebody who’s using statistics effectively. So given all that, that you could spend an hour and a half just arguing about what a data scientist is and recognize that no one is right because there is no authoritative place that can define it. I think we can draw some boundaries that it is kind of a more, it is more technical, it is more statistics, it is more data, more putting things into production from a model’s perspective than what I feel like I have primarily seen in my career when it comes to web analytics, which is adding value, but is more limited to, hey, we understand this core dataset and we’re answering specific questions with this this data set, but a lot of times we’re working with aggregated numbers. Yes, we have to segment and we need to slice it. So I just feel like there is increasing maturity and sophistication in the world of analytics or web analytics. And it’s just on a march towards is the, is the data gets more complicated and the tools get more sophisticated and the storage gets cheaper and more robust, it’s kind of inevitable that we’re going to kind of cross this boundary to where what we’re doing is more sophisticated, more science-y.

00:04:29.96 [Michael Helbling]: Yeah, but I don’t know that that means we’re changing in terms of what we should be called or we’re just embracing what’s always been there.

00:04:40.23 [Tim Wilson]: Yeah. Well, so I guess that’s what I’m trying to say is I don’t, I don’t know that I’m not saying that everybody’s going to have a job title of data science, data scientist in five years or 10 years. I just think we’re, when we look back at people entering, when programmers today look at back at people who were entering punch cards and feeding them into machines, they’re like, wow, that was a totally different world.

00:05:00.52 [Michael Helbling]: Yeah, and I think people will say the same of us. Wow, you tagged websites with JavaScript?

00:05:06.49 [Tim Wilson]: Yeah, but I think it’s more, wow, you were doing exports from a tool to do a trend line of traffic to your site that was aggregated data.

00:05:16.90 [Michael Helbling]: All right. What were you going to say, Jim?

00:05:19.02 [Jim Cain]: Well, I was just listening in. I mean, if we go back to what we were originally talking about, the Florida metrics, Again, my definition of a data scientist is more someone who takes an engineer or a software developer’s approach to problem solving. Currently, they’re being given problems that live outside the current boundaries of the existing tooling and infrastructure. If you can’t do it in Google Analytics, or it’s very complicated to do easily in Tableau with some basic data exports, then a data scientist goes out and they bring everything together. They use advanced statistics, they use software development skills, and they try and answer a complicated question in a complicated way. Does that mean that’s the future of the industry, or does that just mean at this particular moment, certain business requirements have eclipsed the ability of the tools to answer them?

00:06:09.07 [Tim Wilson]: But can I give you two things. One was kind of the technical getting, getting and managing the data. And that seems like a separate thing to me than applying the statistics or building a model. Not that you don’t use tools to do that, but are you, it seems like you’re kind of lumping those together. Like the, if we’re outpacing Adobe analytics or Google analytics, that they’re going to get to the point where They both handle those more extreme cases that somebody’s having to use Python or R or something else to pull it, and they’re going to be applying a layer of statistics on top of it, that those are both kind of an engineering problem.

00:06:49.34 [Jim Cain]: So what you’re trying to say is that data capture shouldn’t be part of, or getting the data shouldn’t be part of a data scientist’s job is just be manipulating data?

00:06:57.97 [Tim Wilson]: No, I’m saying that I think, well, one, I think probably depends, depending on who you ask, one person’s going to define it completely as the, oh, they’re just able to go out and write scripts to basically do ETL on the fly and get data, manage the data. Not so much the raw collection, but the pulling it from different sources. That engineering aspect of it of we got to get the data and clean it up and get it transformed to do stuff with it. But when it comes to statistics, and statistics, I’ve taken an intro to statistics class at two major public universities. And I still feel like when I head into something, I’m never quite sure when i’m applying them right there’s like there’s there’s math and there’s when is it appropriate based on the all the rules around it right so that’s it’s not one where you just say yeah you know i can do a regression on anything but 90% of the time that’s the wrong thing to do so there’s more knowledge around. the math, which, and I think sort of both of those can get lumped under data science. And in some cases, it’s the same person can do both. And in some cases, I think you’re like, no, you’re a Python Jackie, and you, you know, don’t know what an ANOVA table is, but you’re call yourself a data scientist. And in another case, you know what an ANOVA table is, but you need the data set handed to you so you can manipulate it in SAS or SPSS. I’m not, I’m not

00:08:30.21 [Michael Helbling]: anti that, but I also see those as like their own distinct areas of ability or profession. A statistician, because to your point, it’s the application of statistics at the right place at the right time based on the data that you have that makes statistics kind of really awesome. And when you apply it poorly, which is a, you know, that happens a lot, you come up with weird results. But that’s because you’ve got to have really smart statistics people, statisticians who are able to, to like see why the data is shaped this way. So these methods apply better than those. Like that kind of thinking is, I think, its own profession. I don’t know that that’s not a data scientist. That’s just a statistician, but it’s really awesome.

00:09:19.42 [Tim Wilson]: So Nate Silver, a Nate Silver quote, I think data scientist is a sexed up term for a statistician. Statistics is a branch of science. Data scientist is slightly redundant in some way, and people shouldn’t berate the term statistician. which is not to say that Nate Silver is the BL and all definition, but I think we are getting to that definitional part. I think that when talking about the future, I think analysts have to get better at statistics. I mean, if I gathered a room of 50 analysts, I am probably fairly, or 50 kind of web analysts, digital analysts, I’m probably about average when it comes to statistics knowledge. And that scares me. I think that’s really dangerous. I think that’s like 1960s level of what people kind of intuit about math. And if you gather a room of 50 web analysts and say, let’s see what kind of our technical engineering ability is to do, to grab the data and transform it and combine it and manipulate it, I think it’s the same thing. And I think both of those are going to, I think the people who are in the 80th percentile on either one of those fronts are the ones who are actually going to grow and be successful. I think the ones that are on the 30th percentile, and who knows, maybe I’m using the term percentile wrong. are the ones who are going to, one, they’re going to drag the industry down because they’re going to keep training people that, oh, what we do is we automate reports using report. We automate dashboards in Excel to show your trend lines, which that stuff needs to happen. But a lot of times that’s going to be the only representation of business has as to what digital analytics is. and there is stuff ripe for the mining, but it’s got to be a person with the skills. Not only do they have to have those skills, there’s the other piece of being able to identify what the right problems are. I’ve worked with people who have been great, really solid statisticians, and you just never felt like you could get a definitive answer out of them because everything was couched in you know, uncertainty and they struggled to figure out which were the problems worth chasing. So I think it’s just an overall sophistication level that the bar is just going to kind of move up.

00:11:41.76 [Michael Helbling]: So I think the only things that are different now or into the future from here into the future versus when we were doing the first few years of analysis that we were learning how to do is probably that there’s more data and it’s coming from more places and this needs more ETL to merge together into something meaningful for analysis. And there is a need for statistics and better statistics and that’s growing. But the reality is is like, and there’s more devices. Right? So those are the things that are changing our world. Like if you look at externalities or like just what it is that’s changing about what it is that we do, we’re not web analysts anymore. We’re digital analysts because we are trying to merge multiple devices and multiple different data sources and things like that into sort of a, I’m going to say a single view and I hate that I’m saying that because it’s so trite.

00:12:38.57 [Tim Wilson]: You mean like a 360 degree view?

00:12:40.35 [Michael Helbling]: Yeah, 360 degree view. 357 degree view of the customer.

00:12:49.10 [Tim Wilson]: Go ahead.

00:12:50.03 [Michael Helbling]: The thing of it is, is that the need for us to be statistically savvy has not changed one bit. It’s just that we are more able to get to the data to apply our own sense of statistics to it than we have historically been in the field of web analytics, specifically as driven by a group of vendors who did a lot of the statistical slicing and dicing before they ever handed us a report, and then we were interpreting those reports.

00:13:22.19 [Tim Wilson]: I don’t think they were doing statistical.

00:13:23.73 [Michael Helbling]: I mean, if you really go back. They’re not doing statistics. I’m kidding about that part. But they are doing things like sessionization. They’re doing things like sampling occasionally. Those kinds of things are happening in those data sets. And if you go and talk to the big data guys of the quote unquote today’s data scientists, they scoff at those tools. It’s like, I need the raw data or else I can’t do anything with this. And that’s great. I can fully understand that. But there’s plenty you can do with a rolled up set of reports if you know what you’re doing. We’ve all lived that.

00:13:57.89 [Tim Wilson]: Yeah, but I think going back, if you go back 10 years, there was genuine actionable information to come out of a top pages report. What are my top pages? And let me look at entries and bounces because there were, I would say the web was newer. There were some horrible experiences or even just finding why are people actually coming to my site? There was kind of a simpler, yes, a list of the top or the bottom, or maybe even making kind of a calculated metric where I kind of weight some values. But to me, I see it much more as getting to where there is some level of a machine, right? I mean, you can throw machine learning into it. Throw in an ability to say, I actually want to look for, I don’t want to just, Keep guessing and trending two lines together and see which one seemed to move in tandem. I remember when Adobe rolled out Adobe Social and that was one of the big things they showed is, look, you can trend your Facebook reach on the same chart as your website traffic. And there’s something to be said for that, but I think it’s much better to say, wait a minute, I’ve got device type. I’ve got new versus returning. I’ve got this maybe RFM data about some users. I’ve got voice of customer that’s feeding into that. And really what I want to do is say, here’s my outcome. I want to drive more revenue. Now I want to have a tool where I with some smart, some informed, damn it, Mechershoff gave me a line. I’m going to have to find it that I now want to actually crank through way more than I can do with Excel or the web interface of one of these tools. and see where are things that start to become more correlative that I can then say, is there causality? Can I affect it? And that just to me is fundamentally a different thing.

00:15:58.99 [Michael Helbling]: I don’t disagree with you, Tim. I remember doing analysis and getting to the end of my own math skills, if you will. And I know there’s something here. I just don’t know the right ways to find it. And it was, you know, how do I tear this apart in a way that I can see what’s going on? And so to that extent, I think just like having technical skills as a digital analyst, you’re never going to be upset for having some of those skills and developing your skillset as a statistical analyst, as somebody who can manipulate data sets. But I don’t think that means a successful analyst has to be a data scientist.

00:16:42.37 [Jim Cain]: Because I mean, that’s the title, right? Is Data Science the Future? Am I wrong? Is that the title? Yeah. The Future of Web Analytics. Maybe.

00:16:51.90 [Tim Wilson]: We’ll see. Look at you pretending that we were bandying about the final title more than 12 seconds after the show ended. Right?

00:16:59.88 [Jim Cain]: It’s adorable. So if we were to work with that 10 years ago, it could have been is back end developer the digital analyst of the future because it was the only way to get the tools deployed and everybody was fighting with web trends and moving into JavaScript deployed tools and blah blah blah. And the tools eclipse the requirements and we’re just starting to get to a place now where you could be a pure web analyst without having to have development chops.

00:17:24.86 [Tim Wilson]: There’s never in the history of data been a place where you don’t need to actually have a good intimate knowledge of what is actually generating the data and how it’s being captured. But this is where we’re getting all wrapped up in the fucking title again. I mean, this is no, my work.

00:17:43.47 [Michael Helbling]: There’s a difference between a data scientist and an analyst. And I think that difference will persist into the future.

00:17:50.04 [Tim Wilson]: It’s great. The analysts can go and kind of have dead-end careers being cranking out repetitive stuff. I mean, that’s not even remotely true, Sam. You’re saying it is not a continuum from whatever we have no definition for an analyst and we have no definition for a data scientist. I think there is a spectrum from what most analysts that I see and work with and do myself today and what from an aspirational of what I picked up in the last couple of years, some other select people are doing, that I think that’s gonna flip. It is 95% doing one and 5% doing the other. And I think it’s gonna pivot around to the people drawn into the field or the ones who have the skills to be in what that 5% is now. And they’re gonna automate the crap out of some of the stuff that analysts do. They’ll certainly do it a lot faster.

00:18:44.43 [Michael Helbling]: Sure, that’s all great, but I think You might be following victim to sort of a, hey, a bunch of people I’ve met in my career who call themselves analysts have not been really, really smart. And a bunch of people I’ve met who call themselves data scientists are really, really smart. Therefore, the future is data science.

00:19:03.98 [Tim Wilson]: That’s a load of crap. I don’t even know what you just said, though. I don’t know a single person who calls himself a data scientist, actually. Chris Berry calls himself a data scientist. Yeah, there you go, Chris Berry. Okay, so maybe Michael Healy, does he call himself a data scientist? Yeah, he’s a data scientist. I call him one. So I mean, this is, it’s a little, I have looked a lot at what the discussion of, of what data scientists are, and it’s all over the map. So when you start saying, I mean, no, definitionally, yes, I am saying there is a level of sophistication. There is a reason that I am trying to learn R. I’m trying to figure out ways to get better statistics, all of this stuff. And I don’t see it as, well, I’m just dabbling and I’m thinking about shifting my career to something else. I think I’ve looked around and said, I’m looking at company after company that has analysts and they have data sets and they’re not doing dick with them. And you know why that is? Because the analyst doesn’t have the capability. And the analyst, if they did, it’s not like they’re going to be leaving behind some wonderful, this glorious, high value stuff they’re doing. If you could swap them out for somebody who had the higher level of statistical knowledge, the higher level of technical chops, they’d spend like two months automating a bunch of stuff and then they would just be off to the races actually finding things. And I guess the third thing is because in This is the other thing is that it’s not somebody who’s coming with a pure math degree who can just crank through this stuff because you still have to have the business knowledge and the understanding, which is why a lot of the data signed, I think it’s fair to say it’s a little bit of a unicorn. All we need is somebody who can program and do advanced statistics and work with the business and have deep business knowledge. And oh, you know what? They need to do storytelling and communication as well.

00:20:59.42 [Michael Helbling]: Well, and that was going to be one of my other points, which was there’s so much that goes into taking analysis from the data through its formative stages from a hypothesis to proving it out, to then putting it into a place that people can consume it and then driving it through an organization to make the organization react and change to it. I think the way that I define an analyst, they have a part to play through that process. But the way I define a data scientist, they don’t necessarily go through and hit every part of that process. And not because they don’t have value, they have extreme value. And to your point before, there’s parts that analysts have been trying to do that needs to be the realm of the data scientist. Analysts have done a poor job, frankly, because they just don’t have those skills of trying to cover off on the statistics and the hard math, the real math that goes on. But I think there can be a division of labor, too. So I just don’t know that if you’re an analyst today, you need to become a data scientist. I don’t know that I would be 100% on board telling somebody that about themselves for their career. I think I would say look at it. I think you’ll, I would go back to if you learn R or Python or any of those skills or any of those tools, you will never regret that if you’re an analyst. And if you get better at statistics, you will never regret that. And I’ve certainly regretted not having some of those skills in my career at various points in time because I kind of hit a wall and then I need to go get help. But what I do around that is I go build teams that have those set of expertise so that I can continue to go forward. But I’m personally not going to try to go out and learn every single piece of that. I’m, I’m not, well, I’m too. So what’s the analyst?

00:22:45.04 [Tim Wilson]: What is the analyst? Not the analyst manager. No, no, just the analyst. What’s the analyst? What is the analyst?

00:22:51.93 [Michael Helbling]: I think the analyst is the bridge between the business and the data.

00:22:56.63 [Tim Wilson]: As we talked about this on an earlier episode, I agree that there is a role of somebody who has an ability to do some level of data manipulation and data crunching, but has the relationship building, the communication skills, maybe even the ability to set up reports and automate. They’re somewhat experts on the data. And I had this model at one spot and we called them program managers, which we knew was not a good title, but we kind of distinguished those people from the analysts and the analysts were the ones who were more down in digging into the data. So there’s a little definition of like, well, if the analyst is kind of the generalist, where they’re really trying to scope the business, understand the business problems, understand the data that’s there at a broad brush level can say, this is how we would want to solve that. Hey, we need to use, this is something where a predictive model is needed, or this is something where we need to do some sort of whether I don’t know if it’s a cluster analysis or a correlation or regression, but something, is that what an analyst like they would stop at that point and say, now I need the expert who I can draw that box of this is what we’re trying to solve.

00:24:13.36 [Michael Helbling]: I don’t know that I see the boundaries as that distinct because I feel like it’s a collaborative thing. I think there should be ideation from both sides. Every person is a little bit unique in their ability to kind of hit on various aspects of the spectrum.

00:24:33.65 [Tim Wilson]: You’re the one who keeps telling me exactly how I’m defining, how you define data scientist, and that is the way to define it. So if you’re drawing the boxes here, I keep saying not about titles.

00:24:44.63 [Michael Helbling]: There’s functions. There’s functions that have to be filled in a business for analytics to work. Would you agree with that? No.

00:24:53.96 [Tim Wilson]: There are no functions that need to be filled. I guess we’re further off than what I thought. Businesses should not exist. Full stop at enterprises.

00:25:07.56 [Michael Helbling]: Marketers should just use their gut instinct. Trust your intuition, people.

00:25:14.65 [Tim Wilson]: Expert intuition was the phrase that Makershoff gave me, which I think is.

00:25:18.98 [Michael Helbling]: No, but that’s what I mean. So data scientists have expert intuition and great facility with understanding how a business should respond to various things and things like that. The way that I see them interacting, and maybe it’s the way it’s interacting today and in the future, it won’t be this way, but there is, well, I think one thing we all agree on, digital analytics is more complex today and the possibilities require greater mathematics, technical and statistical skills to really plumb the depths and capture as much value as possible. Is that, is that an agreement? So I think that’s, that’s something we can all agree on. The other thing I think we could probably agree on is while marketers I think are getting more sophisticated and digital, they’re not that sophisticated, right? So there’s always going to be a gap between the brainy guy in the corner and the marketers who want to go you know, five years ago without Groupon was amazing. But that might sound like I’m insulting someone’s intelligence and I’m sorry, because that’s not my intent.

00:26:26.70 [Tim Wilson]: Well, but don’t you think, don’t you think though that marketers, when you take the tools, and this is kind of maybe to Jim’s point earlier, when you’ve got Tableau, when you’ve got Domo, when you’ve got Sweet Spot, when you’ve got Inside Rock, and when you’ve got these various platforms that are allowing kind of more sophisticated, more robust, little sandboxes of sorts to play in, that the marketer of tomorrow is probably getting more exposure to data in school. And as they’re coming out, they’re getting internships. They’re being told to go dig into our Google Analytics and figure out what we can change on the site. So I think marketers are increasingly comfortable with data, still agreeing that they are not going to be at the point of an analyst. But I think what many marketers are doing now when they talk about bounce rate and landing pages and segments and hypotheses that they are kind of moving up the maturity chain into the world of a junior analyst.

00:27:29.84 [Michael Helbling]: That is absolutely happening. But one of the things that you just described is that people are creating sandboxes where those marketers can work. And they’re creating smaller versions of the problem or more closed systems so they can get their arms around what’s really meaningful to the business. I would call that the creation of a data product, whether that be merging data sets together into a place where you can analyze it simply like Tableau or Domo. But that is, I think that is the role of the data scientist is to create those spaces By thinking over or arching everything, well, what will it take for us to meaningfully analyze? With input, of course, there’s communication, but the build of those particular things, that’s what’s going to make or break a business in the future. Let’s see what’s making data products simple for marketers to use with great complexity and thought that goes into the architecture and underpinning of those so that When marketers pull the levers they pull to run the business good things happen because someone has thought through all the implications and all the statistics about how that all flows through to the underlying data.

00:28:38.49 [Tim Wilson]: So the funny thing is describing that and I’ve got a. Couple of cases where now i’m doing it again and again and it’s becoming expected of them and i’m doing it with excel and report builder for the most part but it’s an automated weekly or monthly thing and it’s a file so you know those out there who are going to flip out about excel spread mart but it has a decent level of interactivity you know these are the these are these five metrics. pick which one you want to sort by, pick whether you want to sort ascending or descending. It’s got slicers in it, so they can actually explore. Because I say, look, I don’t know your business well enough. You’re going to want to see this is how you want to optimize. This is how you want to inspect your content. All it is is the landing pages, and this is what we’ve set up. I never would have defined that as being hacky early, early, early stage data science, although if I had built something like that in a web environment where it was real time, it wasn’t being delivered in Excel file, maybe so. I mean, is the data scientist building cards and what do they call pages in DOMO? That can’t be right. No, but they’re building data flows and architecting how the data I built a data flow today. Oh, sorry.

00:29:50.13 [Michael Helbling]: You’re a data scientist. Congratulations.

00:29:52.83 [Tim Wilson]: I’m changing my LinkedIn.

00:29:54.89 [Michael Helbling]: There’s so much more to it too. And like we should get a data scientist to tell me I’m wrong because I’m probably wrong. But I also think that’s probably why so many, at least data scientists I’ve encountered have such a low opinion of Excel because it does such a bad job of what a data scientist is trying to do for a business. You know, it’s not giving a business a way to really interact and decide with the data in a meaningful way. I mean, and or the ability to manipulate the data in that environment is poor historically. Whereas I’m just pleased as punch to kick around all day long with pivot tables and VLOOKUPs and I’ll go as far as my little legs can carry me.

00:30:37.75 [Tim Wilson]: Well, and maybe, maybe that’s some indication that Microsoft doesn’t have, they’re not, they’re an easy punching bag, but you know, as they have, as their products have gotten more mature, Excel has more pivot table slicers gives you a crap ton of interactivity that the user didn’t have to have any idea. They’re just clicking on little boxes to say, I want to filter the data.

00:30:56.90 [Michael Helbling]: Well, and, and the tools are emerging now inside of the tools like analysis workspace with Adobe and design, what’s it called the 360 version of the same, the visual.

00:31:09.72 [Tim Wilson]: Google, Google, Data Studio, Data Studio 360.

00:31:12.36 [Michael Helbling]: Yeah, 360. So don’t worry, Google, I will learn the name of that just so you know. Okay. I’m not, I’m gonna get it. It’s only better.

00:31:21.93 [Tim Wilson]: 360, 357. 357, if you do the customer.

00:31:25.77 [Michael Helbling]: Because that’s 357 Magto. America. But don’t you love my view of it? Isn’t it sound so great? You want to buy into it?

00:31:41.47 [Tim Wilson]: I was ready to completely blow off the product building data products, but as we’re talking, I’m realizing in analysis Workspace is another great example. I mean, I sat on a call with a client late last week where I built her a little Analysis Workspace project. And then she kind of took and ran with it. And then she was like, can we hop back on a call? And here’s how I want to use this. And I’ve got, you know, these five panels, I think I only need three of them. So I was kind of coaching her, but ultimately she is, she’s not, she is an analytically minded user. Absolutely. She’s not in a role of an analyst. And I guess I was helping empower the business by, by building better than a sand, but not a sandbox where I’m hauling my bag of sand over and then saying, go and play with it. She’s working with it in the live environment, which is what a Domo does with somebody with, with shiny and R with Python. So it’s kind of the level of how much technical Yeah. I mean, Domo’s data flows, I’m like, shit, I can connect to two things and join it and need to have enough snap to make sure I’m not screwing something up. Yeah.

00:32:45.61 [Michael Helbling]: And of course, the level of complexity goes way through the roof and you need really high-end skills to be able to really take advantage of some of those things. But just on the level of some of the base products today, which again, sort of goes back to your point, Tim, which is there is an increased sophistication and a need for increased sophistication, but the people who Put together that cohort or define that segment and make sure that segment pulls the right visitors from the data set that’s that’s the function data scientist that wall it’s that is headed towards that direction and I would say the analyst is sort of the data scientist when they’re doing that.

00:33:23.54 [Tim Wilson]: Okay, I was gonna say, that’s empowering the business, right? To me on that bridging gap is saying, you know what, rather than giving you a crappy static Excel file, I’m gonna figure out a way.

00:33:33.76 [Michael Helbling]: Teams of analysts will work with the sandbox tools that they’re given by data scientists to go and solve business problems. I mean, the reality is is why we’re all moving outside of our traditional set of web analytics tools is because A, they’re not comprehensive in covering everything that we need to cover as a business, right? Because there’s external data as well as multiple device data and so on and so forth. And they’re aggregated. And so there’s certain analysis you can’t do on the reporting itself until you get the raw data out of those tools, which is why BigQuery exists for analytics 360 customers. and why Adobe Premium exists and you can dump all your raw data into that and do all kinds of analysis on it. I think it’s been a long time coming, but you’re starting to see the need to disaggregate or de-aggregate. I don’t know which one is the right word. Out of our standard reporting tools, a new layer of reporting tool is taking its place in the form of Tableau or the form of DOMO, the form of sweet spot intelligence or something of that kind.

00:34:40.26 [Jim Cain]: Jim? I don’t know. I’m enjoying watching the ping pong game with you guys. I hadn’t thought about the concept of data products being an output of a data scientist. I always see them kind of, like I said at the very beginning, out there on the cusp of where the UI-driven tools stop. I think it’s a valuable product of the data scientist to build things for other people to use that have their own kind of user experiences that don’t need to dig around with statistics and are in big data tools. That said, I still think it’s to a large extent a level up on being an analyst. It’s a direction to get into a business stream for people that are more technical, but I think the value of someone who’s an expert at listening to someone ask a question and bringing an answer out of the preexisting tools isn’t going away because the tools are getting better and better and better all the time.

00:35:32.71 [Michael Helbling]: Well, and that’s why I think there’s actually more demarcation between, and maybe it’s not called an analyst in the future, and maybe it’s not called a data scientist in the future. But I think there still continues to exist separate roles because it’s going to be too hard. It’s going to be too hard to find somebody with the math and science skills to really go and get and dig into all the data and then be able to turn right around and craft a narrative around that and present it to executives in a business. And there are people like that who exist and probably the most, the data scientists that we all know the best are probably some of the examples of the people who are covering that really, really broad spectrum. But even analysts struggle with this, right? I mean, what’s the number one complaint of the analysts today is, well, people don’t listen to me or I work in a silo and I can’t make anything happen. And it has everything to do with narrative creation and communication.

00:36:28.33 [Tim Wilson]: So I had shared with you guys, there’s a guy, I just found out when I was Googling around earlier, the Stephen Garinger, the data science Venn diagram 2.0, which is one where now he drew data science around the entire thing, but he sort of tried to draw a Venn diagram with computer science versus subject matter expertise versus math and statistics and said, you know, when computer science meets math and statistics, you’ve got machine learning, you know, when subject matter expertise meets computer sciences, traditional software and so on and so forth. And kind of his, I think part of the point was where all of those intersect, it’s, it’s a unicorn, you know, which is definitely a phrase we bandy about that there’s, there’s probably an, we’re moving towards more inherent specialization because it’s, it’s a narrower set of expertise is still as much as one person can, can ramp up on. But there’s probably what’s maybe critical is an awareness of what all of those roles are back in the day. This would be five or six years ago, or if I was Jim Cain, I’d say 15 years ago to make it sound longer and more grand. 10 years ago. On average. 10 years ago. Look at my analyst skills coming out. On average. were people who were great in statistics and they just didn’t want to acknowledge the need for communication. They sucked at communication. They didn’t want to acknowledge that as needing to be a role. The classic, the data speaks for itself. Why do I need to, you know, build relationships and communicate effectively? So I think maybe we’re getting to a point, the DAA right now with their competencies, they sort of broke it down into two things, the analyst and the implementer. I can’t remember what the two things are, but it’s basically one bucket analyst and one bucket is the operational side of things and maybe that’s where we’re heading is that we’re going to wind up with three or four or five sort of disciplines not the right word because that’s kind of defining an entire discipline but it’s kind of like these are your five things and you need to be expert at two of them and recognize the value and the importance of the other three and either figure out if you’re the sole analyst, figure out how to get by in those. And if you’re managing a team or part of a team, figure out how you’ve got them covered. But it’s probably gone to more breadth, all of which has increasing depth to cover that. And maybe that’s the issue is that when you say data science, we try to just throw everything into it that you know, that we personally don’t do or that we look around and see some analysts not doing. And maybe that term will just go away entirely and we’ll talk about, we’ll just talk about statistics or we’ll talk about modeling or we’ll talk about machine learning. All those things are necessary parts.

00:39:23.79 [Michael Helbling]: Like there, we need to totally embrace them. And yeah, there’s precious few unicorns out there. Like if I’m anything, I’m probably like a Devil donkey. Not a unicorn.

00:39:38.13 [Tim Wilson]: Oh, dear.

00:39:38.71 [Michael Helbling]: Do I need to go to Urban Dictionary for that one?

00:39:40.71 [Tim Wilson]: No, I hope not.

00:39:41.45 [Michael Helbling]: Oh, boy.

00:39:42.05 [Tim Wilson]: Did you just make that up?

00:39:44.59 [Michael Helbling]: Yeah, I made that up. No, I mean, just like, I’m not, I’m no unicorn when it comes to this stuff and nor I know my limits, right? I’ll never be that. But there are people out there who have those skills. And the one thing I get concerned about as it pertains to this kind of a track is that I think that schools and businesses and organizations are working hard to meet this need, the data science need. And I don’t see an equal and opposite response to fulfilling the other half of the equation of mapping it into the business and making it work and being able to do that. And maybe I’m wrong because maybe that’s what MBAs are supposed to be doing.

00:40:24.74 [Tim Wilson]: No, I think, I think you’re, by the way, there is a at devil donkey Twitter handle Patrick Ormond. Well, thanks Patrick. Following 23 people out of Abilene, Texas. I’m sure. But I think, I think there’s also, I mean, with academia, I think there’s tends to be a presumption of cleaner data than is reality. Right. I mean, I think that’s Well, yeah, I’m kind of marching towards the niche. It’s you have a big, clean, glorious data set. Look, we’ve found this sample data set that no one’s ever going to give two shits about. And that’s what you’re going to do all your assignments on. Oh, no, look, now you’ve got web analytics data. Oh, that’s messy. Good luck.

00:41:01.87 [Michael Helbling]: Well, right. I mean, and Gary Angel speaks a lot about the difference between sort of traditional BI and digital analytics and very compellingly as per usual.

00:41:12.64 [Tim Wilson]: He just says it’s all about the four Vs. Oh, no, wait, he says it’s not about the four Vs. Not about the four Vs. Yeah.

00:41:18.84 [Michael Helbling]: Yeah. Anyway, all right. Well, we should probably wrap this up. I love where we ended up. Jim and stun silence and Tim and I in sort of agreement. How’s your moment, Jim? I was stunned. All right. Well, let’s wrap up. I don’t know if you want to go around the horn. Anybody learn anything new or think anything new?

00:41:40.48 [Tim Wilson]: I can say this has been super helpful for me since I’m at E-metrics in Chicago in early June. I am quite sure there will be some elements of this discussion which will come into my presentation where I will be making the trying to make the case for why analysts should be learning R, which is not something that I’m

00:41:59.87 [Michael Helbling]: put my little icon on any slides that I helped you with there. No, and I think it’s really good, because I don’t think this is the last time this conversation will be had. And I certainly think a lot of definition is needed. And for me, I always take a view of what does the organization need to have inside? have going on in functions to define what then should the role be for each person and should there be overlap. And the beauty of it is, is there’s such seamless transitions. Data scientists can be an amazing analyst, and an amazing analyst can learn to become a great data scientist. I don’t think they’re mutually exclusive.

00:42:41.69 [Tim Wilson]: And like I said, there’s not, there’s not a generally accepted definition for either one. Well, it’s wide open. I’m going to LinkedIn. You log in, you edit profile, and then you can seamlessly change your role from analyst to data scientist. That’s right. Michael Helbling. I think I’m doing that right now.

00:42:57.77 [Michael Helbling]: Michael Helbling is a data doctorate. Wait, no. A data, what’s better than a scientist? A data surgeon. There you go. Drew it. A data druid. All right, anyways.

00:43:10.99 [Tim Wilson]: If we not said growth hacking, because there are some people who said that data scientists and growth hacking, there’s a whole other trend.

00:43:17.46 [Michael Helbling]: If we’re all becoming data scientists, what will growth hackers do? Some people who are growth hackers for real, maybe if you wanted to drop us a line, we’d love to hear from you.

00:43:32.04 [Tim Wilson]: I got my entire quota of the word phrase growth hacker in one session at a recent conference I attended.

00:43:38.53 [Jim Cain]: Any closing thoughts, Jim Cain? I just wonder if some existing analysts who don’t go data science just start to move into a more traditional business analysis stream if you guys are correct, if the discipline is getting more technical.

00:43:49.58 [Michael Helbling]: It’s quite possible. You know, that was the thing I realized after being an analyst for a few years is that really I was just a business person who was trying to use data.

00:43:58.55 [Tim Wilson]: Although I, maybe I got burned. I, I feel like I worked with a lot of business analysts who were basically just gathering requirements for IT to, to build things like business analysts. It wasn’t like the definite, like what you would think the words analysis of the business was it was. So I think that would have been a great term for, but it, somebody got to it before us and used it for something else. Use it for evil.

00:44:23.37 [Michael Helbling]: All right. Well, obviously we figured it out, but if you’re listening to us and you think we’ve missed pieces or you’ve got it figured out better, we’d love to hear from you and we don’t care which side you’re coming from, analyst or data scientist. If you’re in R every day and you want to pop up on our Facebook and tell us how wrong we are, Tim Wilson is here for you. And if you’re in Adobe Analytics all day long, then I’m here for you for that. And if you’re an Google Analytics 360, then Jim Cain is here for you for that. So give us, drop us a line. We’d love to hear from you on our Facebook or on Twitter or on the Measure Slack, which is a great community. of like-minded people who are probably both data scientists and analysts all at once in the happy unicorn valley that is our digital community that you should join. So, uh, drop us a line. Love to hear from you. For my other two co-hosts, Tim and Jim, this is Michael saying, keep data-sciencing. Oh, keep analyzing. Keep analyzing.

00:45:32.02 [Announcer]: Thanks for listening and don’t forget to join the conversation on Facebook or Twitter. We welcome your comments and questions, facebook.com forward slash analytics hour or at analytics hour on Twitter.

00:45:54.11 [Michael Helbling]: Oh yeah, let me just jump in here with a wacky opinion about a notable industry figure.

00:46:06.14 [Jim Cain]: So I’m going back and forth between like weird metal and old school hip hop and jazz and shit. So then it goes, I don’t know, Bieber? Do you want to hear some Bieber?

00:46:15.79 [Michael Helbling]: Do you ever have music that makes you happy?

00:46:18.02 [Tim Wilson]: What about my persona makes you think there’s anything that makes me happy? That’s okay.

00:46:24.86 [Michael Helbling]: This was a show topic idea that we should never write down.

00:46:28.78 [Tim Wilson]: No, let’s just chat about it and slack and forget about it.

00:46:31.51 [Michael Helbling]: Exactly. Never mind. We’re recording. He’s just saying.

00:46:39.99 [Tim Wilson]: Old Gil leaves the sale. That’s Jack Black.

00:46:45.48 [Michael Helbling]: There you go.

00:46:46.46 [Tim Wilson]: Did I get that without good length?

00:46:48.24 [Michael Helbling]: You did. You just got a 12-year-old pop culture reference.

00:46:52.91 [Michael Helbling]: Congratulations.

00:46:56.23 [Michael Helbling]: and an analyst guru.

00:46:58.17 [Tim Wilson]: Ugh, I hate that so much. If this scares you, because it’s really hard, even though it’s really powerful, and you’d rather just mail it in and be a touchy-feely relationship person. Yeah, we’re gonna be a little short on the old outtakes, I think. It’s gonna be a… You know what I mean? Just to put in, go to searchdiscovery.com slash careers. Go work for Michael Hublin. Must have been a hard day for Ford Jim. It was. rock flag and data science.

2 Responses

  1. […] web analytics podcast, Analytics Hour with Michael Helbling and Tim Wilson, where they asked “is data science the future of web analytics?” Panelists were skeptical, saying that the web analyst will remain a web analyst, only […]

  2. […] R yet? No? Well, then Tim is disappointed in you. Or, maybe that’s totally okay! Way back on episode #035, we asked the question if data science was the future of digital analytics. We […]

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

#274: Real Talk About Synthetic Data with Winston Li

#274: Real Talk About Synthetic Data with Winston Li

https://media.blubrry.com/the_digital_analytics_power/traffic.libsyn.com/analyticshour/APH_-_Episode_274_-_Real_Talk_About_Synthetic_Data_with_Winston_Li.mp3Podcast: Download | EmbedSubscribe: RSSTweetShareShareEmail0 Shares