#085: Moving Beyond Microsoft Excel

Some people (possibly even one of the co-hosts of this podcast…on this very episode) have been known to say, “People have this dependency on Excel, which is freakin’ weird!” We know it wasn’t Tim, because he wouldn’t have filtered his language! Whether it’s a symptom of weirdness, an illustration of inertia, or an invisible hand of inevitability, though, Excel remains omnipresent. Is that a good thing? Is it a bad thing? Is it merely “a thing?” In this episode, the gang dives into the topic: the good and bad of Excel, the various paths to a future where its ubiquity is no longer a given, and different strategies and considerations for moving towards that future.

Tools and Topics from the Show

Episode Transcript


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


00:27 Michael: Hi everyone, welcome to the Digital Analytics Power Hour. This is episode 85. Now, we are all sophisticated analysts. We use the latest tools. We keep up with the trends. And as such this show probably have no bearing on your life. Yes. For you this episode is something of a victory lap. That’s right. We’re talking about moving beyond Excel. Wait, what’s that? Your whole organization is still using your Excel? Your clients are still using it? Well, this is shocking. I guess we’ll have to just change our whole plan for this episode. I guess we better talk about it. And I am joined by a particularly angry Tim Wilson. Welcome, Tim.

01:18 Tim Wilson: I don’t wanna talk about it.

01:19 Michael: He doesn’t wanna talk about it. And our other co-host who is taking on the role for this episode of the Excel defender, Moe Kiss.

01:30 Moe: Which is blatant lie.


01:32 Moe: Because I’m so far from Excel defender. It’s not even possible.

01:38 Michael: No. Yes, we come not to celebrate.

01:40 TW: We’re gonna convert you, Moe, away from Excel.

01:42 Michael: We come not to mourn Excel, or whatever that line is in Julius Cesar. And of course I am Michael Helbling. Okay, let’s kick this off. For so many, for so long, Excel has been that de facto tool. I’ve even said before, Tim, that’s the first way that I found out about you in the industry, was reading all of the Excel tips that you would post on your blog, Gilligan on Data, way back in the day. But where have we gotten to, and why has Excel lost that place?

02:20 TW: I’m somewhat conflicted, because up until a couple of years ago, I would probably be the person who would be… As much as we were giving Moe a hard time, I probably would be the Excel defender. And I do still think it has its place, partly because we can’t just wave a magic wand and get rid of it. And so it’s utter ubiquity and having been for the last five or six years external working with clients that is… I’ve had clients that haven’t had Excel, but they are very, very few and far between. So from a ubiquity perspective and the fact that it has got a lot of power built into it for what it does, it’s hard to say outside of a magic wand waving it and moving on. There is an investment… Outside of moving to Google Sheets, which isn’t really a move, that’s the same paradigm, you’re looking at investing in technology. Unless you’re just diving into open source coding, which that’s not really practical for the business.

03:29 Michael: Yeah. I mean, it’s only the data scientists who are like, “Oh, it’s easy. It’s a two line Python. It’s no problem”.

03:35 TW: Yeah. [chuckle]

03:37 Moe: I confess, half of my organization still uses Excel, and we’re going through a big shift to how do we get people away from it. But my biggest question, Tim, is, when you’re working with agencies or clients on your agency side, and you’re using Excel, how the freak do you share it? Were you guys literally just sending Excel files around, and someone updates it and then you email it to someone else? And there is 8.1…

04:05 TW: No, no, no.

04:06 Michael: First off, us people who’ve been around a while, we understand how to create beautifully laid out pages and print areas in Excel.


04:18 Michael: So that you can print out and then punch holes in and then put in the binder…


04:26 Michael: All the information… You guys are laughing. This is not a laughing matter. This is like…

04:31 TW: The print area is a big… But we’ll keep in mind that. So literally in the new role that I’m in, I’ve actually been in Excel minimally. But in the last week, I was in Excel because I was building… If you take Adobe Analytics and Report Builder, which is very powerful and I have spent years of careers building stuff in Report Builder. I’m building a file that’s ultimately getting uploaded to a server and then that’s the mechanism that now the system is emailing out these files that people are opening. So put that in one bucket. And then the other uses of Excel have been, “Oh, I’m doing an analysis. They want the data.” I’m generally producing a deliverable in another means, which might be Power Point. Moe, I’m curious, put you on the spot. Well, I don’t know. Have you opened Excel today?

05:25 Moe: Do you know why I opened Excel today? This is a brilliant…

05:28 TW: To prep for this show. [laughter]

05:30 Moe: To actually, because I have some really killer points about data visualization I’m gonna bring up later. And I just wanted to re-confirm them by opening Excel and making a quick graph in them, which I did this morning. But the truth is, my main purpose for Excel which is potentially embarrassing, often it’s saving things in CSV format or like data dumping from R, reading into R. I’ve never even opened the CSV, it would just be a means to read something into R, or save a dump from R in case I wanna come back and look at it later so that I have that saved somewhere. But I’m getting to the point now where my use for it is exceptionally limited. And I still remember when my sister, Michelle Kiss, was talking to me about the…

06:15 Michael: You’re related to Michelle?

06:18 TW: That’s good to probably bring up again, and again.


06:22 Moe: Yeah. No.


06:26 Michael: Sorry. Carry on.

06:27 Moe: Anyway, she was talking to me about whether or not she should invest in a Tableau license. And I remember saying to her, I was like, “Well wait, how are you doing all your data visualizations if you don’t have Tableau?” And she’s like, “Excel?” I’m like, “Wait, people still do that?” So, I think we come from very different schools.

06:45 TW: Yeah, but I would be careful about being… I mean, and Michael’s intro was a little bit on that. I think, there is, even though Tableau gets a lot of play and their adoption is increasing rapidly, it is still a very small… Or even when I’ve worked with clients that have had Tableau, they’re the people who use Tableau but then the overwhelming majority of the company is still using Excel.

07:11 Michael: Yeah. ‘Cause it’s a learning curve.

07:13 TW: So that’s where we have to be careful about getting high and mighty about, “Oh my God!” It’s like, “No, plenty of people… “

07:21 Michael: It’s sort of like in a certain sense, there’s almost a generational difference. You know, Moe, where you have become an analyst and developed your analytical skill set in a time where you didn’t have to use Excel for everything.

07:36 Moe: Wait, is that ’cause you’re much, much, much, much older than me? And experienced?

07:38 Michael: Yes.

07:40 TW: He’s less than a decade older than you. I am more than a decade older than you.

07:43 Michael: Less than a decade.

07:44 TW: Just to bracket that.

07:45 Michael: But that’s basic, in analytics years.


07:48 Michael: It’s like 14 jobs… 14 jobs later.


07:57 Michael: No, the point is not, like you’re not experienced.

07:58 Moe: No, I know.

08:00 Michael: The point is, it’s kind of the difference, like the first computer our family ever bought when I was growing up was when I was in high school and my kids have had a computer since they were born. So like, that is the kind of difference I’m sort of looking at and wondering like, “Interesting, is that a thing?” Because if you think about it, the future… I really believe, Moe, you’re the future, right? That’s what our future is. But for me personally, I was like, “Why don’t I use these other tools?” And to be fair, for reporting, our business has really kind of gone all in on running all of our reporting through Domo as a platform. And there’s other platforms you can use for sure, but we use Domo. So every week I’m in Domo and I’m getting better at Domo. But Domo is a reporting tool. It’s not an analysis tool. So when I need to slice and dice my data, my home is Excel. And I recognize that’s a limited place. That’s why everybody switched over to R, that’s why people are leveraging Tableau. But it’s interesting because for what I need to do, I can still do most all of that in Excel. Including actually, strong data visualization if you know what you’re doing. I’m just saying.

09:17 Michael: Well, the generational challenge is interesting because if you look at… Skip beyond Moe to the, you know, those who are in their 20s coming out of school, and you’d say “Oh, well they’re gonna be the next ones who were all like, ‘what the hell is this Excel?’” Except you’ve got the gray-haired, old people who are still saying, “Can you do this for me in Excel?” So there is a… And you look at it, it’s like the big consulting firms are some of them that are just like the crazy, like, “Excel is amazing.” Right? So they’re still taking the young crop coming out of college and saying, “Hey, you gotta do this stuff in Excel. Look at all this cool stuff you can do in Excel.”

09:54 Michael: Yeah, and pushing Excel way past its boundaries.

09:57 Moe: I still think you need to know it, yeah. And from that perspective of, it is very good for specific things and what you need it for. I’m just saying, it’s like people have this dependency on Excel, which is freaking weird, and also like unwillingness to move past it, which is frustrating. But it is… I mean if I wanna do a pivot table, I will move to Excel. I won’t even use Google Sheets ’cause sometimes I find that a bit annoying. But it’s improving.

10:27 TW: I will hazard a guess that over 75% of the people who actually use Excel more than three times a week, actually, their mind is blown by pivot tables.

10:35 Michael: Are you kidding?

10:37 TW: No, absolutely not.

10:38 Michael: Really? Oh, my gosh.

10:40 TW: I’m saying, not everyone. I’m not saying analysts. I’m saying…

10:43 Michael: I’m like, that’s about all I do in Excel anymore is pivot tables. [chuckle]

10:47 TW: But I remember what… One I remember going through and a light bulb going on and then it still took my a while to actually internalize what a good data structure was, but I’ve interviewed analysts who have said… Because that’s one of those tells, it’s like a VLOOKUP question, that if they say they use Excel, I’m like, “It’s gonna be VLOOKUPs or pivot tables.” And you will get people who’ll say, “Huh?” And if they say, “Huh?” on VLOOKUP, that’s a little shaky. If they say, “Huh?” on a pivot table, I’m like, “Oh, my God.”

11:16 Michael: See, I always go for the LEN and the MID and the FIND. Those are the ones that I was… I’m just kidding.

11:23 Moe: Oh, Jesus.

11:24 Michael: I like to mix it up and ask HLOOKUP.

11:27 TW: There is one fairly loyal listener of the show who took a class for me that was an entire day on Excel within the last three years and she knows who she is. There’s an enormous amount of… Especially if you’re connected in, if you’re using Supermetrics or… If you’re pulling in data automatically, and you wanna disperse this to a broad group where they want the data. But it’s limiting in that the light bulb… The way that I’m viewing the world right now, which who the hell knows may change by the time this episode comes out, is that the moving beyond Excel, there are literally two ways to move beyond it. And I’d be curious, Moe, if you agree with this ’cause you’ve moved along both paths where you are. And they’re not mutually exclusive. But one is about that data democratization, it’s Domo, it’s Tableau, it’s Qlik, it’s lets get richer visualizations, more connected to the data sources in the hands of more people.

12:29 Michael: Single platforms…

12:31 TW: Single platforms were used.

12:32 Michael: But not single files, right, which is Excel, but a single platform so, yeah.

12:38 TW: With governance capabilities, you can limit access, you can protect people from themselves. In a way that’s the throwback to PowerPlay cubes. You could, if you set up those structures right, it’s really hard for people to get into trouble. So that’s like one path. Then there’s this other path which is kind of the data science path, and that’s the R and the Python, the SPSS and the SaaS. And that is kind of the domain of the true, the deep analyst, the statistician, the data scientist. And Moe, you use R, you use Tableau, both. I think you do most of you’re visualization stuff in Tableau, right? You’ll crank the numbers.

13:20 Moe: Yeah.

13:20 TW: And then wait a minute, how exactly do you push that in to a flat file that you then bring into Tableau, or do you push it on to a server?

13:28 Moe: So at the moment a lot of what I’m doing is I’m aggregating stuff in BigQuery and putting it into a table, and then I’ll just read that table into Tableau.

13:37 TW: Okay, so you’ll connect to the table ’cause you’ll have a smaller…

13:40 Moe: Yeah.

13:41 TW: It’ll handle the Tableau. Which, and Tableau will get better, over time, with its in memory processing capabilities. But what do you think of that framing? Like that, I’ve been kind of marching down that path for the last few months?

13:54 Michael: I guess, the one and only thing I think about, Tim, in that context is sort of where do you see the analysis or the work of the analyst happening? Is it all happening in that sort of statistics and data science realm?

14:12 TW: I think it can happen in both.

14:15 Michael: ‘Cause it also, things like, Analysis Workspace from Adobe kind of like, pull on our definitions a lot. And in the old days, like Net Insight, you could slice and dice your data in the tool, in cool ways. Or Adobe Omniture Discover in way back or whatever, Ad Hoc Reports, whatever it’s called now. [chuckle] Ad Hoc Analysis.

14:43 TW: So one of the diagrams I had that I was going to use in my session here at Summit, was showing that Analysis Workspace is kind of weird because it’s still completely tied to the data that’s in Adobe Analytics. Not the data… It’s only the data that you manage to pump into that silo. So they’ll say, “Yeah, you can use external data, as long as you brought it into Adobe Analytics.” So, it’s challenging it in that way, that yes, it’s intended to be more slicing and drilling, but it does really crank down the expanse of your data set, which when you go to Tableau, Domo, Qlik, Looker, you name it, those are definitionally saying if you’ve got a key, bring that data in and you can look at those in concert with each other.

15:34 Moe: And that’s a direction that everything’s evolving in, is towards you have API’s or some way that you can pull out data and then connect it to something else. And the idea that you have one platform and you would push everything into it, from an analytics perspective, it’s just, that’s not the direction that the industry is heading in. It’s more likely, well from my perspective and I know what we’re doing here, is that analytics is one data set, and you need to be able to connect it to everything else that you’re doing.

16:04 TW: And you need to be able to push the data back in…

16:07 Michael: Yeah. Data flow has to happen both directions.

16:10 TW: Yeah, Excel… Now Workspace and even in Google’s web interface, not Data Studio, but the fact that you can make a segment and then use that for marketing, that’s starting. But you totally can’t do that with Excel. I mean unless it’s, “I’m gonna to email my file to Kinshoo or Murin, and they’re gonna do something with it.” I mean, that’s just fucking ridiculous, so…

16:33 Michael: Hey, there’s plenty that have been there done that.

16:36 Moe: There are still marketers doing that though.

16:38 TW: Yeah, and hopefully though they’re ones who are kind of frustrated and woke about it right? Because as soon as you step over into that path of saying, I’m gonna use a programmatic API driven… Not programmatic like programmatic display, but a program that can both pull data from an API, do stuff with it, and push data back out through an API, that just introduces like amazing automation. I am so conflicted ’cause I do have a special place in my heart for Excel.

17:08 Moe: Okay Tim, so you asked me this question before, what have you done in Excel in the last week? What have you used it for?

17:16 TW: So I had done very, very little, but then I had a case of a client that is using Adobe Analytics and they have a bunch of… It’s a retailer, they have a bunch of merchandisers, and they’re like… They care about this stuff. And I’m like, “Well, wow, with Report Builder… ” Because Adobe has done amazingly well at what they’ve… And I used to, the same thing for Sheets, with even the Sheets plugin… It just, you’d hit a limit. Like I’d get to where I had all sorts of formulas where I might be doing 27 queries and then doing some visualizations in Sheets, but even then I’m like it’s at least cloud based, they can… I had multiple times with… When that automatic data connection is there, that I would set things up, what I thought were one time but I went ahead and pulled the data automatically. Next thing I know, it’s still living six months later and people are asking me questions when it breaks. Which is kind of a good thing ’cause it means they’re actually getting some value out of it.

18:11 Moe: But I think that’s the thing that scares me. So when I started, I got given two huge Excel documents. I mean they had five or six years worth of data in it. So from like a security perspective that terrifies me. But there must have been 20 or 30 tabs that were all feeding into one main sheet, that had like the face of the report on it. And from a collaboration perspective, it was not impossible, but it nearly killed me to go back and look at hundreds of thousands of calculations, and figure out what the frig was going on. If there was… If something did break, trying to figure it out having not been the person that built it, was so hard. And if you compare that to a few lines of code, and to be honest like the same could happen in Tableau to be fair. But if you compare that to code where you can comment and say, “This is what I’m doing here,” you can actually future proof it, that the next person who needs to run that same line of code, can follow what you’re doing. I find Excel is exceptionally hard when it comes to collaboration or shareability. Even just from a like offline one document perspective, but then from another analyst trying to figure out what the hell you built?

19:22 TW: I will both agree and disagree. I agree, I also feel like I evolved over time to have architectures for my Excel work. And that was proven out that I had cases where I’d handed stuff off, the clients were using and maintaining three and four years down the road. But it took a lot of being screwed and having to decipher what somebody else had done. I would say, it takes… Like future proofing Excel, means that you have to consciously really be thinking about it. And the same goes… And the same goes, reality is R and Python, they say comment… I have a background as a technical writer, so I sometimes have more comments than I have in code. So I agree and disagree. Mostly agree.

20:08 Moe: This is why I love it when you share your code with me. ‘Cause I’m like, “Ooh, I know exactly what’s going on.”


20:12 TW: Which would be hard to do in Excel.

20:14 Michael: That’s right everybody.

20:15 TW: Yeah.

20:15 Michael: ‘Cause you can’t comment on your formulas in Excel.

20:19 TW: No you can put notes, right? I’ve done that I’ve put comments in cells.

20:24 Michael: That’s true.

20:24 TW: And it just feels clunky as hell.

20:27 Michael: Exactly. It’s not a good look.

20:28 TW: It’s… Yeah. And even compatibility, when an Excel file gets corrupted, like, “Oh Lord,” you’re opening it as an XML file, you’re trying to delete and you’re like, “Oh my God, if I have to… ” It is so monolithic. I have found that when it’s stable, it’ll live forever and it can be very complicated, with very convoluted formulas and it’ll be fine. But boy, if you’re marching down, mucking with stuff, and all of a sudden it craps out, you’re in a world of hurt that is gonna be very hard to recover from.

21:00 Michael: Yeah. So let’s talk about this split, because you kinda laid out that framework Tim and we’ve been talking about some of the features of it. But let’s talk about reporting. ‘Cause obviously, historically, reporting was one of the big features of Excel. We did all of our reporting there, we sent reports out to people in Excel, people look at their data in Excel. But let’s say we switch to BI tool of your choice of the day, Tableau, Power BI, Domo or Qlik, whatever it is, what are the benefits and drawbacks of making that migration?

21:36 TW: I remember when Sweetspot Intelligence came out. And they had… Actually Eric Peterson I think wrote a paper about, “Digital Insight Management,” I remember having some deep discussions with Sergio about it. Because I think what they were doing, what Domo has done, there aren’t that many platforms I’m familiar with of the ability to comment, and question, and discuss the data natively with the data. And I don’t know what the capabilities… Like if you have Tableau server and somebody’s looking at something, is there a communication capability, a commenting or collaboration capability in that?

22:18 Moe: That’s the one thing that I think Domo does better than Tableau, is that Domo allows you to really easily ask people questions or interact. And I think that’s particularly one thing that they’ve done really well with their pricing structure is that, people can absorb and comment, without having to have a license, which Tableau hasn’t cracked yet. And to be honest, that’s our biggest challenge at the moment, is that Tableau is pretty expensive. So as a business, we’re actually investigating an open source competitor which is called, “Super Set,” which I’m not gonna lie, it’s killing me. Because it’s much more technical, it requires you to have really good SQL skills and all sorts of stuff. But it is open source, and it is developing.

23:03 Moe: So we’re actually building some operational reporting in that, we’re doing some analytical stuff in Tableau for the time being and… Yeah, I think the ability to comment… But then I don’t know, I also beg the question… So I kinda thought this when someone showed me a demo of Domo, which was like someone put like a question saying like, “Oh, why did this go up?” and I’m like, I could just see me as an analyst, suddenly having this report with 50 questions of, “Why did this go up? Why did this go down? Why did this go up? Why did this go down?” Which to me says, the analyst isn’t doing their job, because as you’re sharing this data, you should be sharing those observations as well, and not just sending them like…

23:43 Michael: Yeah. ‘Cause this is one of the challenges, these tools give you is, do you create one view for the whole business? The whole point of Domo would be to create multiple views across the business. Like there’s not, every single person who needs to see the same data and even conceptually when we were all living in the world of Excel back in the day [chuckle] conceptually in our heads, we had frameworks for the dashboard for an executive versus the dashboard for a business operator versus the dashboard for an analyst or a set of data for the analyst because I remember when I used to be the analyst looking at daily reporting and I built my own dashboard for that just to give me a daily context for what’s happening with the business, that wasn’t something that I sent to other people, it was just for me to do my job as an analyst. It wasn’t for other people to look at, so this is the work and this is where I think it gets really challenging or rather migration is a lot more difficult than people realize, which is all this thinking that you’ve sort of been avoiding by using Excel and letting people sort of do this work “on their own.”

25:08 Michael: Is a significant level of effort. Who will create all those different views, who will maintain them? Who will fix them when they break? How do you make sure they’re all working correctly? Like these are, even just inside, Search Discovery is a pretty small company but we are hitting problems in the data and our data is very simple, but we will hit problems in the data and have to work with and luckily we have people internally who are focused on dealing with that but that’s a big commitment. We’re using a lot of somebody’s time to deal with and build out all that we wanna build out as a business around in our case Domo, but it could be Sweetspot, it could any of these platforms. So I think, it’s a little tricky or a little wrong to be like. “Okay move from Excel to this” without counting that cost up into the total so to speak.

26:05 Moe: We have this exact issue happening at the moment, where we have one of the analysts in my team who basically has 12 dashboards lined up that people want and she’s like “oh great, so I’m gonna spend the next six to nine months just building dashboards and nothing else.” And I think it goes back to the bigger question of analyst’s job is to understand the why not just kind of do static reporting on past performance, well in my view anyway. And also how do you keep someone really excited about work when… I mean that stuff, I know that there’s lots of people that do love it but for her that’s not a super big challenge and so you go to these tools and you’re like let’s move away from Excel and then it comes with its own paradigm shift of then how do you manage the analyst’s work and how do you make sure that they just don’t become this factory for spitting out dashboards too.

27:00 Michael: Tell you what, that’s where all the AI people should be spending their time is cleaning up data and being smart about it, somehow.

27:07 TW: [laughter] I’ve gotta share this, so Trevor Paulson from Adobe is… He and I do this are doing this session at Summit where we’re going through our stuff and we both have some examples we walk through and he makes this comment as he goes through it and he was like so basically we’ve just gone through a couple of hundred lines of code, that is really just the data preparation, heres now two lines where we’re actually running the model, he was like we talk about data science like its that and all the work is actually the cleanup but having said that I think it’s important to sort of conceptually separate the visualization, the dashboard and the underlying data that you’re working with and I think there is a, to Michael’s point, there is a challenge or an opportunity, if you build a really robust data model and you have your data and you’ve got the lines around the data of what can be linked up and joined where, then you build a dashboard, you build a few widgets and the people who are really interested, they head down to self-service training and they’re reasonably safe from getting into trouble.

28:24 TW: Like Google analytics, seven or eight years ago people just didn’t pull fucked up data. Like you just couldn’t no matter how you pulled the data, it was gonna wind up being the right data. You might still mis-interpret it, but you weren’t gonna get in this mess that you can now get into it with Google or Adobe or you name it where “oh no no no” you can’t sum that stuff together, you can’t take bounce rate, you can’t take conversion rate where you’ve got 17 dimensions and sum those together, you don’t have to know the data. So I think there’s the potential to say, it is goes to me back 10, 12 years of Cognos’ Powerplay building cubes where you said, what is the very rich set of data that we can construct a cube from and now we can federate that out to everyone in the sales team and we can control their access based on what they should be able to look at and they’re just not gonna get themselves in trouble and and turn up with a bad number.

29:29 TW: They’ll still come to us and say “hey, I saw this thing that happened and now the analyst is getting asked a question that goes to if you’ve got Sweetspot or Domo where that’s being publicly recorded I’m asking this question, I think people are better about not asking the random curiosity bullshit questions and more the meaningful, I found this, I didn’t look at a dashboard you sent me and see something and wonder why you don’t tell me the why, I looked at data that you the analyst hasn’t actually stumbled across. It’s in my niche that I care about. I was able to explore it some way, which is back to Michael, your question about the two paths. Yeah, I think analysis happens on the path of the drilling in. I’ve now hit a point that I don’t know what’s going on, it could be X or Y and the data that’s available to me, I can’t figure that out. Now, I’m going to the analyst and I think that’s perfectly fine. Are you guys still there? Hello? Hello? Hello?

30:27 Michael: Even for you. No, so I think Tim, you are…

30:32 Moe: No, that was really long winded. I’m struggling.

30:35 Michael: I think you’re moving in the right direction, and in a certain sense, it kind of splits the role of the analyst from that “data administrator” kind of person. And that seems exciting to me. Can we hire somebody else to do that job and I just get to be the analyst full time? That would be sweet. The reality is probably we get to do both. So, don’t get your hopes up too high out there.

31:03 TW: Well, expectations will be set… The vendor that you’re moving to, the expectation will be set by the vendor that, the business will get this intuitive tool. The data administration will be automatic and the tool will be intuitive to use for all your business.

31:18 Michael: Oh, just click, click, insight.

31:21 Moe: But I do… That’s part of… I’m just trying to think about this in our context. We’re really lucky. We have an amazing data warehouse and people can access it via a cube in Excel which is why lots of people still use it. Or in Tableau or just directly in SQL, whatever your preferred method. You still need to know the data. You still need to know there’s this particular flag that you put on to say don’t count invalid or cancelled orders. You still need to know that you need to use net revenue provisions if you’re talking about return rate. There’s all these caveats that you need to understand to make sure you don’t come out with really weird numbers that are completely inaccurate. And that’s part of the analytics team’s job, right? Is to educate the business on how to use those things well. And the same goes for Google analytics and summing unique whatever whatever. It’s part of our job. We need to teach people that stuff and not just kinda leave them in a corner to figure out stuff on their own.

32:17 Michael: So, maybe if anything, this migration away from Excel for companies, doesn’t necessarily make the job easier but maybe makes the job aspects clearer. What do you think of that?

32:32 TW: It seems like you are… You start to have to define some roles, just like we used to talk about the analyst had to do implementation and analysis and reporting. And we’ve kinda said, wow implementation can be a role unto itself. Implementation and integration, there is… I don’t know, data administrator, data engineering, data, let’s wait for the DAA for another year or two and they’ll come up with another set of roles and responsibilities. Because I don’t think they’re really capturing that aspect of it.

33:05 Michael: I do think that it’s a data engineering role. That’s kinda what I think it is. And the ubiquity of data and the amount of data demands… Honestly, I think we should be elevating the data engineer. Everybody and their brother wants to be a data scientist, but actually a data engineer is what’s keeping businesses afloat.

33:27 Moe: Our data engineers, we have a team of four. We couldn’t survive without them. They’re utterly incredible.

33:34 TW: Hold on, let me pause. So everybody out there who hates Moe and that she has a four person team of data engineers and they’re looking around saying “What the fuck? I’m supposed to do the data engineering between 7:45 and 8:00 AM in the morning, I hate you Moe.”


33:52 Moe: We’re very lucky. Our team has gotten pretty big and we’re very lucky. But they do an incredible job and I remember when we were back at SUPERWEEK and we were talking about data accuracy and who’s job it is to QA numbers, that responsibility for us, it very much sits between our analytics team and our data engineering team. We collaborate so much on that stuff that if there is an issue between the two teams, it’s picked up pretty quick. And it’s never like this is your job, this is my job. It’s like, this is our collective responsibility and make sure that whatever we’re feeding into our data warehouse is correct and being used correctly as well.

34:32 TW: Well, and the reality is one of the big challenges is the no data collection is complete, and so then you wind up, in the pockets where the data is incomplete, there are cases where you can work around it and there are cases where it’s just a blind spot. And that becomes some of the real challenges. I think technology, vendors tend to see it as “Oh, you have a common key and you join them.” I’m like, “Well, what about the other 10% of revenue where that key doesn’t exist?” What do we do? How do we handle it? And the reality, the answer is well it depends on what you’re trying to do with it. And the analyst may need to kinda learn that, but that’s… I’ll pause and not be another seven-minute rant.

35:14 Michael: And then you work through a huge set of data engineering challenges like what do you do with the outliers? What do you do with in-store kiosks that are getting these renewed cookies like thousands and thousands of visits? It’s like, oh that data is worthless, throw that out. It’s like, well, what about all the orders that came through that? Those are the real world challenges and they exist in every… And that’s kinda why. But at the same time, as I’m saying it, I’m kinda like am I doing that thing where you thank a person who’s been in the military for their service, and then secretly you’re like but I don’t wanna be that.


35:57 Michael: You’re doing a great thing but I don’t want to do that.

36:01 Moe: So can I just ask? I’ve got a kind of pivotal question that I wanna get towards which is… So I feel to wrap up that we see pros of Excel and we see cons, yada-yada. The industry is evolving past it, so on and so forth. But if you do truly want to get your company off Excel, which is a journey that the iconic… Like I mentioned. There’s a lot of, well, not me. But there are a lot of people in the business still who do, they would spend, I reckon 90% of their day in Excel and we wanna move them past that. How would you go about doing that? Like truly getting people comfortable on something else.

36:43 Michael: It’s an interesting question. So we have this internal feedback forum through our company called Office Vibe, which I think I’ve mentioned before. And one of the people goes through and says, “Hey, it’s really interesting that when we’re displaying things in our town hall meetings and PowerPoint slides, we’re not using our central tool of record,” which in this case would be Domo. We’re using Excel. And it really made me think, because whoever put that in [A] like, “What’s their problem.” No. I’m just kidding. I’m kidding [laughter] I’m so kidding. But I was thinking, I was like, “Okay. Yeah. What about me?” Any problem I need to go solve quickly, my default is gonna be Excel for a long time until I take sort of that arduous journey of replacing my Excel skillset with a new set of tools, whether that be Tableau, whether that be R, whatever it is. And I’m like, “Well that’s really interesting,” ’cause I don’t expect that our CFO will probably ever move away from Excel in terms of the way that he probably puts together our numbers which is probably where those financial charts are coming from.

38:06 Michael: And it’s just interesting because it’s sort of like, “Yeah. How do you get past that?” Even for some… I don’t know that we’re super forward looking or anything. But we certainly are trying to walk the walk as we try to explain to our clients the benefits of these kinds of things. But even for us, it’s really challenging to kind of move there, because it’s like, “Do I feel comfortable doing this personally in a new platform? So could I slice and dice the data in the same way I would be able to do it in Excel and Domo? Can I get those views?” A lot of times, it’s sort of like… I have to send an email to someone and be like, “Could you create this report for me this way in Domo?” Or whatever. Right? And so it’s kinda like you’re in a different… I don’t know that added anything to our discussion, but it was just something that made me think a lot. The more that is your primary capability, the harder it is to put it away and pick up something else.

39:12 Moe: But Michael, for you personally, the fact that you… ‘Cause I’ve got a person in my team who’s like that. And he says, “I always revert to Excel because I will be able to do it in Excel the fastest. I know it the best.” What does it take to push you over that edge and say… Is it lack of time? What is gonna motivate you to jump off that cliff and say, “Actually I’m gonna give it a go in Domo today?”

39:37 Michael: Well, I mean, so it’s a little unfair because most of what I do from an analysis perspective today is pretty simplistic. And I do it in Domo. And so that’s kind of… But let’s say for instance if I was gonna do some longitudinal analysis of all of our time spent across all of our projects by person. I’m definitely gonna download that data from our internal tool and pull it into Excel and put it in a pivot table. ‘Cause there’s no better way for me to look at that data over time than in that way. Right? But for an analyst, I think, if I was gonna go back out and be an analyst again. Let’s say I just was gonna hang up my cleats and rejoin the ranks of the analysts, I think it’s on me to reup my skill set and say, “Okay. What are the tools of today? What makes an analyst great today?” Okay. Is being able to think about and utilize some of these other tools that not only provide better benefits, but also clarify the differences between what an analyst does, what a data engineer does, and what a data scientist does.

40:47 TW: I think if you’re an analyst who is committed to being an analyst, that path… And maybe that’s why I find that little sort of two-paths view helpful for me. I think there is… There are open-source platforms out there and there is a world of a community. And it’s hard. This is not like I’m gonna knock this out this weekend. But if you are a… Love to crunch the data, have you ever done a propensity model in Excel? Good luck. You know what? There are blog posts. You wanna try doing a Shapley value attribution with Excel? Good Luck. So that path, it’s easy. It’s like flip a coin, pick R Python, and start screwing around with it and you will wind up doing stuff that you could not do in Excel. So I feel like that is literally the motivation of the analyst to say, “I’m gonna pursue that path.” The path around to go to Domo, Tableau, Qlik, Datorama, Looker, whatever, that requires an investment, I think is actually trickier. Because that’s the one where it seems like an executive has to see a demo or fall in love and feel like that’s the path you need to go down.

42:05 TW: Tableau has the… Tableau Public you can do some degree, but you can’t… It’s not stuff you wanna do necessarily with your own data. But to me the one path is very easy, and maybe partly I lived that. I had limited funds to invest and said, “Screw it, I can follow this path.” It’s possible I had conversations with Moe’s sister, Michele Kiss, not the same person, about that as well saying, “Do I wanna buy a Tableau license and teach myself that or do I wanna keep using this open source service with some very nominal costs, if I wanna do some hosted stuff?” So to me that one path is literally the motivation, and the interest, and the excitement about what additional things to be done with the data on the part of the analyst who wants to go that route. The path to getting a company to shift… Although, really Power BI may be the path that… If there’s a company that’s all in on Microsoft then I haven’t really said Power BI, but that’s kind of heading down the path… That’s an evolution from Excel, that’s still in the Microsoft stack. And Gartner’s saying they are still super healthy and strong and you can pull R stuff into Power BI. So it may be that you can… For a while you could do Power Pivot. I’m not super clear on what exactly the Microsoft stack looks like.

43:30 Michael: I think Power BI supports R.

43:32 TW: Yeah it does. You can…

43:34 Michael: Yeah.

43:34 TW: Sort of. I didn’t actually get it to work, but it does in theory.

43:37 Michael: Oh, okay.

43:37 TW: But there is a migration… Microsoft sort of should be totally teed up for this for companies that wanna take an incremental step and say, “We’re not leaving the Microsoft world, but we’re gonna investigate Power BI and maybe that’s the path to go down. I haven’t really dug into it much, but it’s gotta take somebody who says, “This can’t… The stuff from the past can’t be the way forward.”

44:04 Michael: Right.

44:04 TW: Wow! I am not lacking in strong opinions.

44:07 Michael: I think this is as good a place as any for us to wrap up. So I’m gonna start down that path.

44:15 Moe: Before we wrap up. I just wanna bring up my homework that I did before the show, which I’m going to post. So last comment on Excel hating, the defaults.

44:28 TW: Oh my God!

44:28 Moe: I’m sorry, this tool has been around since what, 1984? It comes up with a default color palette which spits colors at you like yellow, red, blue, green and purple, which don’t even get me started on that. And then it also doesn’t follow Donna Wong’s rules of thirds, which I will also share in the show notes. So if you are listening, and you work on Excel as a product, get with the Data [44:51] ____, man, honestly. Anyway, I’m done.

44:53 TW: I will say that a… A few years ago, I ran, and this was maybe when Excel… Not the latest version, this was a couple of versions ago, ’cause I was curious whether if I ran it through a color palette checker… I’ll put a link to it, it’s on the Analytics Demystified blog. I think it was maybe before I worked there. But it’s on their site. I was curious whether they actually were following color blindness best practices, and they are. They actually… Their colors are actually readable as I recall with the newer versions at that time, which was five plus years ago, that if I ran it through kind of the top three or four color blindness, they were actually distinguishable colors. They’re terrible, they’re horrible, the Excel defaults is…

45:40 Michael: Many many years…

45:40 TW: Inexcusable and I’m pretty sure, if I can find some… I have had vitriolic rants about that as well.

45:47 Moe: I’m gonna test it. My product manager is actually color blind, which has been amazing for me personally because it’s made me be much more aware of what I’m using in my graphs. I’m gonna test Excel and I’ll let you know how I go by…

46:02 TW: So, June 18th 2009. Data visualization that is color blind friendly. Excel 2007, there is a detailed analysis of how it works with deuteranomaly, deuteranopia, protanopia and protanomaly.

46:22 Michael: What I used to do, Moe, back in the day was use the Juice Analytics Clean Charts macro that they built, which was exceptional at removing a bunch of clutter and junk by default. And you’re 85% of the way to having your chart look presentable.

46:41 Moe: Nice.

46:43 Michael: But I don’t know that it…

46:45 Moe: Still is supported, ’cause it’s so old?

46:47 Michael: No, it’s probably not supported anymore but…


46:50 Michael: Although they do have a zip file still available on their website.

46:53 TW: Excel defaults, that was a very good call out. Excel defaults generally blow.

47:00 Michael: Yes. Yeah they do.


47:01 Moe: Okay. I’m done. I’m off the soapbox.

47:02 Michael: Okay. Hey since you’re in the mood to chat, Moe, one of the things that we love to do on the show is go around the horn and do a last call. And I’ve heard that Tim has a lot of them. So we’ll start with you.

47:19 Moe: Well maybe Tim can just cover everyone off.

47:21 Michael: Just do everybody this time? No, come on Moe. What’s your last call?

47:25 Moe: So, I’m gonna share a blog by Trevor Stephens which Matt Gershoff kind of inspired me to revisit decision trees, mainly because… His point is about how easily understood they are by the wider business. It’s something that’s quite easy to digest that often we use models that are really complex and people can’t understand them. And decision trees is something that people can understand and visually is quite easy to follow as well. So I’ve…

47:55 TW: It is extremely difficult to actually do the analysis required for a decision tree in Microsoft Excel. However, fairly straight forward, yeah.

48:03 Moe: Yes. Which is why I used one in R. Now I’m pretty sure his blog did cover other languages, but I obviously focused on the R one so I’ll share that in the show notes.

48:14 Michael: Outstanding. And before we get to you Tim, I’d just like to share my last call. So this one is kind of a silly one. But I ran across it recently, it is somebody’s little project called, The Silicon Valley Job Title Generator. And it’s hilarious because you know, you could either be a webinar mastermind or a long form engagement custodian. But the reality is that being a Twitter visionary might be your job title or an in-house type mentor. You just never know. So anyways, it’s a little bit of a funny. Okay. Tim Wilson, what is your last call [S].


49:01 TW: So one, I’ve had one on my list of potential fall back last calls for so long that it’s actually gotten updated, and it’s so appropriate to this topic so it has been around for a long time except there are updates. So a professor, at the University of Texas named Clint Tuttle, has done parody songs around Excel. So in his class, he… The first one was he performed live in front of his class and it’s on YouTube, a parody of Justin Bieber’s, Love Yourself and it’s Learn Excel. And the guy actually has vocal chops and guitar playing chops. That one got so much good reviews that since I originally had found that, there is one that is off of a Justin Timberlake’s, Can’t Stop the Feeling. He made a music video and it’s a parody of that which I think it’s, Can’t Stop Excelling. So they are just amusingly well done Excel homages.

49:58 Michael: Wow Tim.

50:00 TW: That’s number one.

50:01 Michael: Keep going.

50:02 TW: Is that no good?

50:02 Michael: Wow. That’s what I was saying. Wow.

50:05 TW: Okay.

50:05 Michael: What a value… No, I’m just kidding.


50:10 TW: So second, at the risk of pissing off Moe, because she never does this, but I’m gonna do it. I’m gonna plug that Columbus MeasureCamp coming up on April 28th. So if you’re in the Eastern half of the United States, our last ticket drop is coming up very soon. Tickets have gone kinda like crazy, so we are looking to have a couple of hundred people in Columbus on Saturday, April 28th. I’d love to see you there. I’m not running it. I am on the organizing committee.

50:36 TW: And then my real last call that was something I stumbled across that was like, “How do I not know about this,” is the Pudding. Do you guys know The Pudding? Pudding.cool. It is… It’s literally amazing digging in data visualizations on kind of cultural topics of the day. So I stumbled across them because they had done a deep analysis of gender around This American Life which in the US if you’re a podcaster, NPR fan, you’re familiar with This American Life. It is an amazing show that he’s had a predominantly female gender staff, yet the analysis they did it was actually female voices on the shows was under represented and they actually went to the guy who created This American Life. So it’s kinda like FiveThirtyEight.com but a narrower initiative they just tried to hit cultural topical issues and they dig in deep and do some really cool visualizations. So that’s about all I’ve got. I’m gonna skip next show entirely.

51:41 Michael: Yeah. There you go. So I mean, I guess if we ever did an analysis like that on this show, it would not go well. Tim.


51:51 TW: Yeah. You got it.

51:51 Moe: Well, cause Tim talks so much.

51:53 Michael: Okay.


51:56 Michael: Hey, Tim. No, we have no time for you to respond to that, Tim. We have important things…


52:05 Michael: You’ve probably been listening to this episode and you’ve probably been thinking, “Ugh they’re getting it so wrong on these important and salient points. And we would love to hear how.” And the best way to do that, I love for you to engage with us on the measure slack or on Twitter or our Facebook page or even our website analyticshour.io, although once the transcript loads into the page, those comments are very below the fold.


52:35 Michael: So some of those other options. The other thing that our listeners should be putting in there little heads right about now is what about MEE? And by that I mean the marketing evolution experience which is coming to Las Vegas this June. Just go type it into your Google to see what that’s all abou. The Digital Analytics Power Hour will be there in person and we’re pretty excited ’cause we don’t get to do that too often. So we’ll be looking forward to meeting a lot of you at the Marketing Evolution Experience in June. More information in podcasts to follow.

53:19 Moe: We do have a discount code to share with any listeners that wanna come along.

53:23 Michael: We will eventually even have a topic much to Tim’s delight, [chuckle] by the time this airs, we’ll probably have a topic.


53:32 Michael: And… But we’re really excited so most of you know but some of you don’t, this is the next step in the evolutions of analytics and marketing conferences here in the United States e-metrics was a part of our lives for so many years, but the organizers of that conference have taken a step back, they’ve looked at what the market and the industry needs and they’re trying something new and that is the Marketing Evolution Experience. And we’re excited to be part of it, and we look forward to participating in it with all of you and the DAA is also having the [54:08] ____ awards I believe, at the same time. We are very thankful that you’re listening.

54:14 Moe: Still.

54:15 Michael: We look forward to hearing from you.


54:20 Michael: For my co-hosts who are frankly, a little bit rude right now.


54:25 Michael: I want to encourage you all, no matter what tool you’re using, to keep analyzing.


54:35 S1: Thanks for listening and don’t forget to join the conversation on Facebook, Twitter or Measure Slack group. We welcome your comments and questions. Visit us on the web at analyticshour.io facebook.com/analyticshour or @analyticshour on Twitter.

54:55 S?: So smart guys want to fit in, so they made up a term called analytics. Analytics don’t work.


55:03 Moe: Wait. Why you, so you’re putting [55:04] ____ in the title, in the intro.


55:09 TW: I said, that was my joke. Let’s go. Let’s go.

55:12 Michael: No time to change. Let’s go. Let’s go.


55:16 Moe: Seriously, now? Oh, we missed Tims whole tirade. Can you… Sorry, Tim, why are we doing this topic…

55:21 Michael: If you play your cards right Moe…


55:24 Michael: I think we can get him to go again.


55:27 TW: Yeah. I think everybody agrees, the lightning talks are a bad idea.

55:32 Moe: So by everyone, you mean you two?

55:34 TW: Yeah.


55:38 Michael: And that’s the show everybody. Thank you so much.


55:44 Michael: Okay.


55:47 Michael: Wait, are you in a massage chair?


55:51 Moe: I’m in one of those giant bean baggy things.

55:53 Michael: Well, you have… Oh, that’s your microphone.

55:56 Moe: Yeah, yeah.


55:56 Michael: You were getting all comfy and I was like, “she’s getting a massage right now.”


56:01 Michael: Alright, sorry Tim, back to you.


56:06 TW: Oh, the days when 64,000 rows was our limit.


56:11 TW: Finally. Episode 85.

56:14 Michael: That’s right. 85. The true Tim Wilson.

56:17 TW: Finally came out of my shell.

56:18 Michael: Finally emerges.


56:20 Michael: Tell us what you really think Tim.


56:24 TW: That sounded kinda obnoxious.

56:27 Moe: I personally would like to start a Twitter war on Excel and pros and cons, so feel free to jump in.

56:32 TW: Okay, Excel… #excelwar.

56:34 Michael: Excel war, #excelwar, start that up with Moe, me and Tim or Tim and I. I don’t know, I’m not good at grammar. Shout out to Benjamin Gaines, he’ll know what I”m talking about.

56:49 TW: Because my last call was not long enough, we are just gonna…

56:52 Michael: We can edit all this out…

56:53 TW: Okay. Carry on.

56:55 Michael: And Tim loves awards. He loves them.

56:57 TW: Oh my god.

57:00 Michael: He loves to be nominated, he loves to be considered, he loves everything about awards and award shows.

57:06 Michael: Oh you guys are killing me.

57:10 Michael: Rock flag and pivot tables.


One Response

  1. Zack Gore says:

    GREAT podcast. Touches upon so many relevant topics / internal debates that I have amongst team members and other Analytics professionals. In general, much of the company wide reporting is done through excel/google sheets, however, I am in the process of rolling out my own reporting and am contemplating the best tool to do so. Due to the large data set that I’m working with (will be over 1mm records within ~4-6 months or reporting), I am scoping Tableau Desktop as my reporting tool. I already have a license and am relatively familiar with functionality, but my question is whether Tableau reporting can be shared with end users who do not have a software license (either Desktop or Reader)? If they do have Reader installed, does that mean they can at least apply (check/uncheck) filters? Any other thoughts or considerations relevant to this decision would be much appreciated!


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

#243: Being Data-Driven: a Statistical Process Control Perspective with Cedric Chin

#243: Being Data-Driven: a Statistical Process Control Perspective with Cedric Chin

https://media.blubrry.com/the_digital_analytics_power/traffic.libsyn.com/analyticshour/APH_-_Episode_243_-_Being_Data-Driven__a_Statistical_Process_Control_Perspective_with_Cedric_Chin.mp3Podcast: Download | EmbedSubscribe: RSSTweetShareShareEmail0 Shares