Retail and Consumer Goods Tableau User Group 30 Sep 2026

0:01 Matt Huff: Um, this is, this is gonna be a very different format from our regularly scheduled program.

0:09 Um, really what we're, what we're trying to do is we're just trying to create a, a, a series of mini TUG sessions. We've got it blocked for an hour. If it takes us five minutes to get through everything… I don't think it'll take five minutes. Um, but if it takes us five minutes, then that's, that's what we get. And if it takes us longer than that, then that's what we get, right?

0:32 Um, so by all means, this is, this is, uh…

0:37 The, the whole intent here is to have, have a spot where we can, where we can chat and, and work through, work through the material. Um, today is gonna be very, uh, very much focused on before you get into Tableau, what are some things that you should be doing?

0:58 And it's, it's really geared towards helping us get a, a good foundation before we, before we get into Tableau. So, uh, we'll cover some of the stuff that, that Blake and I shared a couple years ago at, at Tableau Conference, um, how we think about design and, and problem-solving within, within my current employer, and then we'll also start to think about, you know, how do we, how do we think about designing dashboards in, in general?

1:30 And I'll be the first to admit, I'm not a, I'm not the strongest dashboard designer, um, but there's, there's a lot of good concepts that you can, you can take and apply. Um, but the, the long- the longer view here is that over the coming, over the coming weeks, this is gonna be a, a series of presentations. We, we start with basic, here's what's, you know, average analytics problem, and start working into more advanced Tableau capability.

2:00 Um, and, and so that's, that's the thought here, is, is really starting to build a, a curriculum that we can start to i- interact with and, and, and ask questions along the way, um, and, and have this be fairly, fairly casual. Sound good?

2:16 All right.

2:18 So with that, um, I can go ahead and start sharing my screen, and then I think, Blake, you're gonna, you're gonna kick us off, right?

2:28 Blake Feiza: Yes, indeed. That's perfect. And just a, a note on the format of these calls too. If you're in the call live, you might have noticed that we've made you all presenters. Uh, that's because we want this to be loose and interactive. So if you have questions on a topic or if you have thoughts as someone that has expertise in some of the curriculum that we're talking about, by all means, hop in on the chat, hop in live and cut us off.

2:52 Uh, everything is just to, to create content that will be valuable to others. So looking back at that recording, uh, if you've got something useful, hop in and say it.

3:01 Um, but without further ado, a quick peek at the goals for today's session. Uh, we're going to really dive into two big fundamental pieces that Matt and I talked about at TC '25. Had to think about that one.

3:16 Uh, the first of which being viz best practices, uh, and the second one being some of the formatting best practices, uh, that, that Matt had mentioned before. So, uh, if we wanna jump in, Matt, uh, let's see, two slides forward. I think we can probably just start with the content.

3:34 Matt Huff: Yeah. Right there?

3:36 Blake Feiza: Yes. That's perfect.

3:38 Awesome. So when it comes to building a visualization, whether it be a dashboard or, you know, whatever that final deliverable is, uh, typically you want to start with some form of requirements gathering. So in the case of my former employer, H-E-B, we always started off with something that looked like a stakeholder interview. So getting in the room with the people that are ultimately going to be your end users for this dashboard or deliverable and really start to drill into what the business problem is that they're trying to answer, and more importantly, why is that business question something they need the answer for.

4:12 Uh, if you take the time to do these interviews, oftentimes you will realize that the initial ask is just the surface level of what the actual business purpose actually is that we're trying to serve.

4:24 Uh, so by taking that time to really understand what it is that your business partners are trying to solve for, oftentimes the final dashboard that you make might be different than that initial request was because it wasn't covering the entire scope.

4:37 Additionally, there's kind of a second piece to this, which is the scope of the ask. Oftentimes if you ask a stakeholder what data they need access to, their initial response is going to be everything. They always want the data buffet.

4:51 Uh, so every field that you can imagine, if you give them a checklist, they would check the entire row. So with that in mind, I think part of the requirements gathering is not just expanding the scope, but also reducing it and making sure that we only need to provide the data that's necessary as the practitioners here, uh, and make sure that we get that business question answered in the end.

5:15 After getting those requirements gathered, uh, the next piece here is around getting inspired. So as that data visualization expert at your company, it's your job to figure out what the most appropriate dashboard should look like.

5:28 And if you wanted to reinvent the wheel every single time, I'm sure you could, uh, but I'm also sure that you'd probably be leaving some room on the table to iterate and refine how that dashboard looks in the end. Uh, there's no need to do that. Uh, we've got tons of resources, especially in the Tableau space on sites like Tableau Public, where there is just a plethora of excellent business dashboards and non-business dashboards, uh, available to pull inspiration from Uh, and not just online, but also your peers in your company.

5:57 If you're not the only designer in your company, talk to other people around you, because most likely the questions that your stakeholder is asking are very, very similar to those that your peer stakeholders have been asking. Uh, or it's a dashboard that may have already been built and can just be, you know, repurposed or redirected.

6:16 Uh, so consider all of these different resources and really get inspired as you're thinking about what your plan of action is going to be and how that deliverable is going to come to life in the end.

6:26 So just recapping these two kind of phases, the ultimate goal here is just to figure out what that end state needs to be. So when we say start at the end, it really means picture the final product, and then let's work backwards from there.

6:40 Um, and then Matt, we can flip to the next slide.

6:44 Matt Huff: Yep.

6:45 Blake Feiza: Awesome. Let's see, how much of this did I just ramble on the other slide?

6:48 Uh, who is your audience? Yeah, so as we think about some of those considerations for starting from the end, sometimes the audience isn't a stakeholder, right? Maybe it's a dashboard that is serving you as the analyst, and you just need a place to perform a recurring ad hoc analysis.

7:03 If that's the case, the design components and, and some of the visualization that you create might be a little bit different than something that would serve a stakeholder. So considering who it is that's going to be your end user, uh, is incredibly important, uh, because it does shape what that final product will look like.

7:21 Uh, additionally, how is this data going to be used? In some cases, it's a dashboard that that final deliverable is. In other cases, it might just be a table, right? An Excel export or something that you can serve up directly from, from an AI agent, right? It, it really is considering how that data is going to be handled and answering a question. Uh, and maybe a dashboard's not the right medium for, for that specific ask.

7:43 So considering that is also incredibly important.

7:47 Uh, and then next up, we, we talk a little bit about update cadence, right? So is this something that needs to be refreshed daily? Is it a report that maybe is only used every fiscal quarter?

7:57 Uh, is it something that needs to be near real time, and we need to figure out a way to set up a data source that updates every fifteen minutes?

8:04 What, uh, what does that cadence look like? Because oftentimes that will inform the, the underlying data structure that you need, and make sure that you even know if, if it's possible with the data that you have, uh, up front.

8:16 So all of these go into that why piece, right? Why do we need a dashboard, or why do we not need a dashboard?

8:23 Uh, just lots of things to consider and, and align on upfront.

8:28 And Matt, if you wanna jump to the next slide.

8:31 Perfect. So once we've conducted that interview, we've gotten inspired, we likely have some idea in our mind about what a dashboard should look like, assuming we're taking that route.

8:41 Uh, the next step that we would use, uh, over at H-E-B was creating a physical mock-up. So when we say physical mock-up, what we really mean is something that is not super polished. It's low fidelity. It's something to just get the ideas in your mind out into something that can be shared.

8:58 Uh, oftentimes, we would use physical sketches on either a piece of paper with a pencil or using something like an iPad and a, an Apple Pencil.

9:07 Uh, and the idea here is just to sketch out or whiteboard what that dashboard might look like that answers the questions you're trying to solve.

9:15 Uh, if you would prefer something where maybe you're grabbing screenshots from your Tableau Public dashboards that you've found, that totally works too. Just throw them into like a PowerPoint slide or something and stitch them together. The idea here is really to remove all of the friction that you can here and just get that idea out onto paper to share with others.

9:35 So in this phase, it can be super useful to refine that idea and maybe share with, with a small initial group around you.

9:43 Uh, but yeah, the, the next piece here for kind of this wireframing process is going to be something a little bit more high fidelity.

9:49 So depending on your relationship with your stakeholder and how involved they may want to be, they could either get involved in the initial phase, or maybe you want something a bit more polished to share with them.

9:59 That's where this digital mock-up step comes in. So as we start to think about how we can improve the quality of, of said mock-up, uh, we oftentimes want something that can be shared with other people, right? So if Matt's living in San Antonio and I'm living in Austin, and him and I want to iterate back and forth with one another, uh, using a, a pen and paper probably isn't the best way to do so at this point because we'd have to deal with USPS and everything.

10:23 Uh, it just would be a bit of a pain.

10:25 So thinking about something that can be shared easily and also something that increases the clear outline of, of what you're trying to accomplish, uh, is, is kind of this next step here. So we would use tools like Figma. Uh, oftentimes that was really useful if you wanted a custom background in Tableau because that Figma mock-up that was outlining the entire dashboard can ultimately be iterated on and create a background for your dashboard.

10:51 Uh, so depending on your tool of choice, there is no right tool. The best tool is whatever you're most comfortable with. So if it's PowerPoint, then it's PowerPoint. If it's Figma, it's Figma.

11:00 Canva, like this slide presentation, um, and so on.

11:05 And last but not least, it is definitely worth seeking that feedback from everybody that you can get, uh, the perspective from. So, you know, we're all one brain, uh, and if we can bounce off and, and pull ideas from others, it can become incredibly useful. Uh, I know that for me, I always think my ideas are brilliant until I talk to somebody else and I realize that there are plenty of, plenty of things that I could improve on, and maybe it's a chart type or, you know, the, the structure of the charts and, and kind of the order that we want to drill down on the information.

11:34 Uh, the, the goal here is just to create something quickly, uh, that you can iterate on and essentially outline that final deliverable.

11:46 Awesome.

11:47 And we can flip to the next slide here.

11:51 All right.

11:52 So next up, just taking a moment to talk about the data model here that's going to support our dashboard.

11:58 Uh, so first and foremost, if you're not spending time thinking about the data up front before you start building in Tableau, you will likely get burned. Uh, and that's because the data model that you come into Tableau with will ultimately dictate how easy it is to build that dashboard you plan to build.

12:15 One of the big considerations that you'll want to keep in mind is whether or not your data should be tall or wide. So when we say tall versus wide, what we're referring to is the number of columns in that data set, and ultimately the number of rows in that data set.

12:31 So if we think about what a wide data set might be, maybe we take a dimension where it could be, I don't know, in Superstore, the category.

12:40 Uh, and instead of having one column that is category, we have one column for each of the three categories.

12:47 Uh, so it just changes the structure of your data, uh, and it, it becomes a little bit of a different problem to solve in Tableau.

12:55 Oftentimes, both will work in the end, uh, but one will be a path of less resistance. So planning that out up front can be incredibly useful down the road, uh, and making sure that you know exactly what you're getting into.

13:10 Uh, the next thing on the list here is densifying your data.

13:13 So if you're trying to create something like a, a more complex chart, like radial bars or something like that, data densification might be a prerequisite. You'll want to consider that up front, uh, and understand the trade-offs there between, you know, maybe a more bespoke chart type and also larger data. Uh, if you have a, a table with 100 or 1,000 rows, it might not be that big of a deal, but if you've got millions or tens of millions, suddenly that data densification can have performance implications.

13:42 So something to consider there as well.

13:45 Uh, and then also for densifi- densification, uh, you can also consider something like a date scaffold, right? If you need a calendar inside of your view and you're worried about days with missing data, uh, you might want to consider a scaffold so that there's a way to account for days with zero sales in the example of Superstore.

14:03 Uh, and then what will be handled inside of Tableau and outside of Tableau is also important. How much computation can you push further upstream to reduce the amount of compute that that Tableau workbook needs to do on the fly? Uh, ultimately to just improve that performance for the dashboard, uh, because no one wants a tool that, that buffers for 30 seconds every time you click a filter.

14:26 Uh, and then last but not least, collecting that data. Um, maybe how do we want to structure that data? Do we have a data warehouse through something like a Tableau s- a Tableau Prep or a Databricks?

14:36 Uh, and maybe we have a published data source that's up on Tableau Server, uh, we want to just connect to that. Uh, it's important to figure out how some of that ETL is happening and, and where that final table that we're connecting to lives, uh, to ultimately create something that is highly performant. You know, if we can use extracts inside of our workbook, that's obviously going to be more performant than a live connection.

14:58 Uh, so just creating the appropriate data model, uh, will ultimately help build that dashboard.

15:05 And now that we've spent some time thinking about what the data looks like, uh, we're now at that point where we can start building inside of Tableau.

15:14 Uh, go ahead, Matt.

15:18 Perfect. So a quick checklist here to essentially recap all of the pieces that I just talked about.

15:24 First and foremost, we go through our requirements gathering phase. So stakeholder interviews, understanding what it is that we need to build, digging into the why and not just the what.

15:35 Next up, getting inspired, doing the, the research on best practices and understanding what that chart is that, uh, you think will serve the why.

15:45 And then going into our mock-up phase, starting with something physical and, and low friction, highly rapid iteration.

15:53 Uh, and then going into something a little bit more high fidelity with a digital mock-up or a resource that can be shared around.

15:59 Uh, seeking feedback, so sharing that mock-up with your stakeholders, sharing that with your peers, getting a multiple sets of eyes on that product before you go in and you start to craft the dashboard.

16:11 Uh, jumping into the data model. So again, talking about how our data is structured, uh, making sure that we have the right fields to create this dashboard.

16:21 Uh, and then ultimately jumping into Tableau and getting building, the part that we all love to do here on this call

16:28 . Awesome.

16:30 And I will pass it to you, Matt, if you have any, any notes on, on this viz best practices piece, or if you wanna jump into the, the formatting.

16:38 Matt Huff: Yeah, no, I was just gonna… I, because I've got it up on a separate screen, I only have the one screen, or I've got the, I've got the, the presentation up. Are there any questions or comments, thoughts on, on this, you know, kind of what Blake has covered so far?

16:56 Blake Feiza: And for those hopping in late as well, everyone is on presenter mode, so if you want to hop in and, and say any comments on the mic, you're more than welcome to, or type in the chat. We've got our eyes on both.

17:08 Matt Huff: All right.

17:08 Blake Feiza: Sweet.

17:10 Matt Huff: Cool.

17:10 Well, I'll, I'll go ahead and get started then. Uh, obviously, if you've got questions and, and whatnot or, or comments, like, I can't see them, so, um, I'm gonna rely, Blake, I'm gonna rely on you to kinda stop me if I need to or, or anything like that. So.

17:27 Blake Feiza: You can count on it.

17:28 Matt Huff: Awesome.

17:29 So the… So we talked about, so Blake talked a little bit about the, the structure that is needed before you go and, and design a, a dashboard, right? And go build, go build a report in whatever form that is, whether that's a email that you're sending out, uh, a Tableau dashboard or a, a table or, or whatever.

17:50 You're, you're effectively likely going to follow something similar. Um, you're gonna have a, a mental checklist that you're, you're working through that is doing some of these things, whether, whether it actually manifests itself- Physically or, or not, right?

18:07 Um, now, one of the things that we need to think about is, as we're thinking about data storytelling in general, is how do we, how do we do that in a way that helps us communicate clearly with minimal, with minimal distraction and, and really starting to hone that, that story. And so the… Ultimately, as we start thinking about this, the, the, the question, like, one of the key questions that we have is, well, why do we build dashboards or reports?

18:41 And the reason we do that is, is really to communicate data and insights, right? And it's to help drive, it's to help drive a data narrative and to help drive strategic action.

18:56 That's really why we're doing dashboard, why we're building dashboards, why we're building reports, w- in whatever capacity we're, we're, we're doing that, right? Whether that's, whether that's through an enterprise, enterprise reporting ecosystem, or whether it's one-off builds that you're doing to help inform work that you're doing separately.

19:18 The reason we have dashboards, reports, data visualization in general, is to help drive that strategic decision. And w- in whatever capacity that is. If that's, um, in the retail capacity, that could be, "Hey, do I need to, do I need to increase pla- facings of a product in the store? Do I need to order more of this product? Do I need to shift some marketing strategy to drive performance?" Uh, dashboards and reports are the mechanism that we use to inform that decision.

19:51 Just like as you're driving a car, if you look at your dashboard, you'll, you'll see your speedometer. You'll know, based on where you're at in the road, you'll know whether you need to increase or decrease your speed. A dashboard is, is intended to help guide, guide a decision. And so that's why… That's ultimately what we're trying to do, is we're trying to build that data narrative.

20:10 We're trying to tell a story that then allows us to, and, and change some kind of strategic decision, okay?

20:18 So the key considerations that we need to think about here as we're, as we're talking through that is, how do we minimize that, um, ink-to-insight ratio? And what do I mean by that? So as we think about a dashboard or as we think about a, a table or a, a chart or whatever.

20:37 So if I, if I look at a table, and it's, you know, ten rows, ten rows long and, you know, six, six columns, now I've got 60 observations that I'm, that I'm, I'm putting out there, and I'm really leaving it to the reader to, um, generate the insight. So as you think about what that means, uh, that's a lot of ink j- used to create each of those, each of those observations, and the insights that you can glean from that are really hard to find.

21:09 Um, an effective approach would be to look at a, a bar chart, um, and highlight an, an individual bar in that bar chart.

21:17 That all of a sudden now points the reader to a specific observation that they can then ground against the rest of, of that visual, and your ink-to-insight ratio, um, improves dramatically, right? So that's…

21:33 So that when we're talking ink to insights, it's the extra stuff in our or, or, um, table that, that allows for us to develop insights quickly. And so as we think about what this could mean in Tableau, this could look like, um, Tableau defaults to, uh, li- li- light line background. You know, maybe it's removing some of those. Or if we've got,

22:02 if we've got legends data or in our, in our dashboard, can we, can we remove those legends but still get as we're, as we're highlighting, um, parts of the, the title or, or whatnot to, to allude to that legend without actually having to show legend. It's minimizing the extra that comes with a dashboard in, in order to make that, in order to improve that ink to insights.

22:29 Other things that we need to think about is how do we, how do we improve the speed to insight?

22:34 And so as we think about this, it's how do you simplify the, what you're presenting in a way that gets it so that you can, you can quickly point to the observation that you want the reader to, to take away, and you want them to be able to do that fairly quickly, especially in a corporate setting where you need to have, um…

22:55 You've got folks that are, are working through a report and don't have a ton of time to just sit there and digest it. If, if I can point to you quickly the, the observation that you should take away, I've got a couple of key points that outline what, what it is that you should take away from that, that chart.

23:12 That is much more impactful than, "Hey, here's, here's an Excel document with a whole bunch of data, um, and, and I'm just gonna leave it to the reader to figure out h- what to take from that," right? And so how do you, how do you improve that speed to insight and minimize that, that ink to insights ratio are incredibly important as we think about when we're formatting a dashboard.

23:35 We need to be cognizant of both of those things to make sure that we're help- helping our reader and, and not making them work too, work too hard. Okay?

23:45 Um, so as we're thinking about this, and I, I alluded to this a little bit earlier, but the… Only keep the needed formatting.

23:53 In Tableau, Tableau defaults to putting a lot of excess formatting into, uh, into a line chart, into a bar chart.

24:02 Uh, strip a lot of that out. Uh, you don't, you don't need a lot of that.

24:07 You, you may not even need, um, axes depending on how you're labeling your data.

24:12 And so it's h- just be cognizant of the, the formatting bloat that already comes inherently baked with the, the tools that you're using.

24:24 Um, the other thing is to give your data breathing room.

24:27 So this looks like white space in a dashboard, and so as we think about this, it's don't cram 15 charts into a 9, 9 by 13, you know, page, page report. Um, give them, give them a little space to breathe. And, and, and give…

24:45 And that helps the, the reader focus on one thing at a time. It helps guide them through the flow of your report overall. And then, the other piece here is color and size.

24:57 Uh, use, use those sparingly, and, and by sparingly, really what I mean is use those strategically.

25:05 Uh, use color, use color strategically throughout your, your visualization. So if you're pointing to one observation at the top of your page and, and you're tracking that observation through the whole report, use the same color. Don't, don't be changing colors haphazardly. Um, bold and size the, the text, uh, in, in a way that, that points to the same observation throughout, throughout the data.

25:32 Um, and, and keep it simple.

25:36 And then as we think-

25:36 Blake Feiza: And hop in really quick here, Matt.

25:38 Matt Huff: Yeah.

25:38 Blake Feiza: A little bit of activity in the chat. Uh, Külli actually mentioned about her starter workbook that she uses too.

25:43 Matt Huff: Yep.

25:43 Blake Feiza: So Külli, if, if you wanna come off mute and, and talk to it yourself, by all means, uh, or I can read the chat.

25:51 Külli Esko: Yeah. Well, uh, to make my life easier, I've just prepared myself so-called starter workbook where I already set up all the fonts, the color palette, remove all those nasty grid lines.

26:07 And, uh, also prepare like, um, a starter dashboard, removing phone layout so, uh, that I w- won't forget it. And now recently I had it… added this, uh, Ojo's trick of, uh, uh, putting extra floater contain- floated container to it to, to, um, grab all the, uh, filters and, and other stuff.

26:32 Uh, also, yeah, set up the size, uh, first floating, uh, text objects for, uh, title and source, and I have saved it as a starter workbook. So basically every time I start a new project, a new dashboard, I just open up this one, not just a new, uh, empty one, and it's already, like, so much more easier. I know I have everything set up, and there are no, no, no, no such, uh, things as grid lines.

27:06 I don't have to do deal with them anymore. So, uh, uh, just the thought that maybe it's a good suggestion for anyone else as well.

27:17 I like it very much. I was just thinking that how didn't I think of it before? But-… better, uh, later than never, so.

27:26 Matt Huff: Well, and-

27:27 Külli Esko: So, yeah

27:28 Matt Huff: … yeah. And I don't, I don't have one of those, but, like, that's, that's an excellent idea. I n- like, that's probably what I'll be doing.

27:36 Um, I'll probably, I'll probably end up doing that for Makeover Monday this week if I'm being honest, i- is, is building… You know, creating a, creating a template for that.

27:47 Külli Esko: Yeah. It's really, really useful. It saves so much time, and all those extra clicks you always have to do, so it's always, uh, like, up and running already. You are already on half a way, so I strongly su-

28:01 Matt Huff: And-

28:01 Külli Esko: uh, recommend, as I wrote in the chat.

28:04 Blake Feiza: Yeah. And, and from an enterprise standpoint too, right? If you think about having an entire team create dashboards that are styled the same, a starter workbook could be a great resource to share around that entire team so that everyone has that same dashboard configuration. Uh, it's an easy resource because no one wants to follow a 20 step guide on how to configure the dashboard up to style guide standards.

28:27 So another use case outside of Makeover Monday would be, you know, something like an enterprise starter, starter kit.

28:34 Matt Huff: Well, and, and if you go back to the conversation we had, I don't… Was it last week, around AI builds in Tableau, like, you could, you could set that up against a starter template, right? And now a lot of the design heavy lifting has already been done.

28:52 Um, and so, you know, that's, that's something that, that the agents tend to struggle with is, is some of, some of those capabilities. And if you already have something built out, that, that gets you, again, improvements, um, for your life. So cool.

29:11 Any, any other questions, Blake?

29:12 Blake Feiza: And we have a couple.

29:13 Yeah, we have a couple more questions. Uh, one of them is from Rohit who asks, "Does more marks cards, uh, reduce the database performance?" Uh, and so generally speaking, yes, Rohit. Uh, in some cases, if it's a very small data set that you're fetching, it might be such a small performance hit that you may not notice it.

29:33 Uh, but generally speaking, every time that you add more fields or more rules into the view, uh, the more that computation is going to get heavy and, and slow things down a bit.

29:45 Um, so rule of thumb would be yes, be mindful about the amount of, of things that you're pulling into the view. Uh, but in some cases the, the difference may be negligible, and if you need to create, uh, if you need to add more marks cards in order to create the view you're trying to create, uh, it's, it's weighing the pros and the cons there and, and picking what makes the most sense.

30:07 Matt Huff: Yeah. And I, I know on my side I've ha- because I haven't had to do a whole lot of, like, I- Enterprise reporting, I've had the luxury of being able to pull in small data sets, and so, like, what I'll te- tend to do is I'll have several small data set, small, relatively speaking, data sets, um, that I'm, I'm using to build a dashboard for specific views, uh, or components of the dashboard, and that has significantly improved performance as well because I'm not having to hit this massive data set for a BAN or, or something else like that.

30:41 Blake Feiza: Right.

30:43 Yep. And we've got one more question. This one is from Alex. Alex says, "Do you have any suggestions for how to accommodate stakeholders who ask for more than you know should be put on one dashboard or report?"

30:55 Matt Huff: Yeah. A, a lot of times it's… A lot of times what I'm doing is I'm, I'm working with them and kind of guiding them through the process, and, um, a- and it, it ends up becoming a, a dialogue, right? And a, and a conversation is, is how, how I've tended to approach it. Now, the…

31:17 For us, we end up running into, um, we end up running into a lot of problems where there's just an abundance of data and, and trying to get something in a meaningful form for, for anybody is really, is really tough. And trying to do… And for us, it's doing that at scale is, is a problem, and so we're filtering down to rather granular views very frequently.

31:44 Um, and so what we, what we did and the way that we bypassed a lot of that was aligning on the design upfront.

31:52 That allowed for us to say, "Hey, we all, we all hands in agree that this is the approach that we wanna do, and we'll refine it over time. We can always add things to it or, or take things away." Um, but by getting that upfront alignment, everybody kind of knew what, what they were getting and, and d- I would say it's, like, the other part of it is once a dashboard's been built, it, it's not, it, it doesn't have to be, it doesn't have to be sacred.

32:21 You can always, you can always tweak and refine it and, and adjust it if, if you discover that, hey, this observation or this whole section that, that we had doesn't really make sense or isn't all that useful. Like, let's, let's, let's kill it. The other thing that, the other thing that I've done is I'll create, I'll create a suite of reporting. So one is, like, an executive summary that gives you an overview of, of the overall business performance, and then if you've got questions, I can, I can, within that same dashboard, I can link you out to another dashboard that gives you the, the context for that particular question, um, that, that you may, that you may be w- concerned about.

33:02 And so I know the Flerlage twins have, uh, a blog post on, on how they think about, um, uh, they, they would use, I don't even remember what it's called, but it's effectively they'll have tiles in a master dashboard, and you can click on the tile and it will navigate you to a separate, to a separate dashboard, right? And so you can create kind of a, a landing page, if you will, that will pull up insights as needed and, and really focus the, the narrative on, on what it is you're trying to communicate in, in that report in particular.

33:37 And so, like, there, those are some things that I, I have tended to do in the past.

33:41 Um, I don't know if that, I don't know if that helps or not.

33:44 Blake Feiza: Yeah, I think that's great, Matt. And, and just hearkening on the, on the fact that, you know, some of the work that you can do upfront in that stakeholder interview and really boiling into, you know, not just what is the request and all of the, you know, the large amount of information that we wanna pack into a, a dashboard. You know, getting past that and digging into, you know, why is it that you need all of these fields and dimensions.

34:07 You know, what is the, the fundamental question that we're trying to ask? And if we can reduce that scope down, then to Matt's point, having some of those, those pieces that might get removed in the future, uh, maybe we can skip that entirely, right? And push it down the road to maybe that next, uh, that next dev cycle and build it as a separate dashboard or, you know, answer that question later and, and kind of prioritize and stack rank there.

34:30 Matt Huff: Yeah. And it goes back to what we were talking about earlier, right? With why do we build reports and dashboards in whatever form that looks like, and it's to drive a decision, right?

34:40 Cool.

34:41 Um, all right. So I think, so as we, as we think about, um, so jumping back into it, right? We talked about ink to insights.

34:52 Um, the other thing is, that we, we covered a little bit was speed to insights. And so this is where labeling comes into play. Um, double encoding your colors and sizes and, and fonts and that kind of thing really help drive attention throughout your dashboard.

35:09 Um, keeping, keeping things simple, uh, is just what we were talking about, right? It's, h- hey, we're gonna, this, this part of this dashboard tells us this story, and this is when you should use it, and this is why. And if you, if you need further, further context, like, here's, here's further context. Maybe that's in the tooltips or, or s- or maybe it's in a separate dashboard, right?

35:31 That you're, you're linking out to. Um, but it's really how do we keep that, that structure clean, simple, easy to read, um, easily interpretable, right?

35:43 Speed to insight is, is the, is the key takeaway there. And then as we think about this, like, as analysts, we have a tendency to want to throw everything into the dashboard, um, because, hey, I, look at me, I did all this great work. It, there's a whole lot of interesting things that are showing up here. A lot of times, a lot of times you can boil that down to a single chart or, or even a BAN or, or maybe it's, maybe it's just a bullet point.

36:10 And, and so as you're thinking about that, it's h- thinking about how to, how to minimize the, the technical, the technical burden on your stakeholders so that they don't have to go and do the heavy, heavy thinking because they've got enough stuff to think about.

36:25 Um- The other thing is, is using bands effectively.

36:29 Uh, so I'll, I'll use bands to, to highlight, um, in, in a marketing space, I may wanna highlight a conversion funnel. So I look at impressions and then clicks and then conversions and, and do all of that. And I can use bands to kind of give you that waterfall, right?

36:46 Um, the bands can be really helpful to get, to get quick reads on, on stuff, especially if I'm comparing multiple segments, uh, across a population.

36:57 Um, other things that we can use is, is color.

37:00 Like I mentioned earlier, double encode your colors throughout your, your dashboard. And then, and then really this is…

37:07 This last bullet point here, uh, is incredibly important, and it's communicate what you wanna communicate, and don't communicate what you don't wanna communicate. And so as you're working through, through a dashboard or a narrative, you do… Make sure that what you're contributing to that framework actually contributes to that framework, and it's not a red herring.

37:30 And by that what I mean is don't… Like, if I'm doing a sales dashboard, I don't necessarily need to, specifically in the context of Superstore, if I'm doing a sales dashboard, I might not actually care about the, the sales team member's performance. I may just care about the overall sales trends and trajectory of my business overall.

37:52 Um, but if I'm looking at, if I'm looking at performance reviews for, for my team, if I'm within the context of, of Superstore, I may not care about total sales performance. I might just care about the performance of the individual sales reps.

38:08 And so

38:11 if, if I include the s- total sales dashboard, that might be useful, um, depending on the narrative that we have, but it m- it most likely isn't. And so it's only communicate what it is that you wanna communicate, and make sure that you're removing the excess, the excess, um, excess data and, and content that doesn't actually communicate the story that you're, you're trying to share with your stakeholders.

38:36 Okay.

38:36 Blake Feiza: Yeah. And, and to hop in really quick, Matt.

38:38 Matt Huff: Yeah.

38:38 Blake Feiza: We've got one more question in the chat. Um, so Rohit again with another great question. Uh, he says, "How can we effectively represent high cardinality fields?" As soon as a response to some of the advice with using color effectively and, and things like that.

38:51 Um, I, I can jump into the first step here.

38:54 Matt Huff: Yeah.

38:54 Blake Feiza: Which is there's, there's two different approaches that come to my mind immediately.

38:59 Uh, the first of which is if we have a high cardinality field that we're trying to visualize, using something like color and trying to assign a bunch of different colors is obviously just going to become a massive spaghetti chart-… uh, that is essentially useless, uh, to read.

39:15 So as we think about using color effectively with a high cardinality field, it probably looks something more like highlighting a specific, uh, value in that field. Uh, so maybe all of the other lines in a line chart are gray, and then we have one that is blue, so that you can kind of see how that specific value is performing relative to the rest of the cohort.

39:36 Uh, or alternatively, it could be something where we have a filter built in, right? And we can select, uh, a set or a, a subset of records from that field, uh, to again reduce the amount of, of squiggle and cognitive load that you have in that chart, uh, to make it useful. So it kind of depends on what the question is you're trying to answer. Uh, but there's different strategies for still using things like color and pre-attentive attributes, uh, but not overloading the visualization.

40:05 Matt Huff: Yeah. And those would, those would be how I, uh, would approach it as well. And it, part of that's gonna depend on, hey, can you add interactivity? And if I can't add interactivity, a lot of times what I'll do is I'll, you know, I'll color the, the biggest observations i- in the data set. So even in high cardinality data sets, the, not every observation is going to be as useful as other observations.

40:32 And so this is where principles like the, the Pareto principle comes into play where, you know, 80% of your, your value is driven by 20% of your, your observations. And so a lot of times in situations like that, I'll, I'll highlight the, the 20% or, you know, be able to filter down to that 20% and really just kind of hone in on, on those observations and provide the remaining 80% of observations as, as context, but it really, like, starts to focus on what's really gonna move the needle, right?

41:08 Cool. So, um, just as an example, I'm gonna a visual that, that I've got. This is an old Makeover Monday, and it's far from great.

41:20 Um, I've learned a lot since then, but as you, as you look at it, there's a couple of things, right? I've got, I've got narrative over here on, on the left. I've got, I've got double encoded colors kind of all throughout, and, and I've got, you know, simple charts that are, that are tied to specific observations and, and the, this narrative that I want that, that text to, to support, right?

41:48 So the text should support the visual, the visual should support the text, and it should be self-reinforcing kind of throughout, throughout the visual. So, um, and then as, you know, as we're looking at this, you know, you can see how I, you know, partition off the different components and, and how I've got a l- could, could probably stand to add a little bit more white space, but, um, you know, th- those are the kinds of things that, that we're, we're starting to show and, and talk about and, and kind of reinforce throughout, throughout this, um, throughout this session.

42:22 So any, any questions on, any questions on this that we haven't covered yet?

42:28 And so with that, um, there's no other questions. That's effectively the format that we're going to do for these mini TUGs is we want to have just a, just a bit of dialogue, have it, have it very casual. Wanna give you guys the opportunity to participate, share your, share your key learnings that, you know, Külli, appreciate the, that you had earlier.

42:52 Um, those are, those are the kinds of things that we're hoping to get out of, out of these sessions and, and really start to focus on, um, just, just building up some skills overall. So with that, we'll, we'll be back in a couple of weeks where we're gonna talk through, you know, blue versus green and some other basic Tableau functionality. Um, this, this session in particular is, is more about how do, how do we get prepared to do any of that, right?

43:20 Before you go into Tableau and, and start going down that rabbit hole, um, it's really helpful, uh, to kind of ground yourself on what does it take to be a, a good, good analyst and, and a good analyst in the context of, of Tableau.

43:34 And so that's, that's what we're, what we're hoping to do is, is have a series of sessions where it's fairly casual and, um, maybe even end a little early sometimes. So, awesome.

43:47 So with that, I'll, I'll give you guys back 16 minutes and appreciate, appreciate everyone joining. We'll get this posted on YouTube and, and, uh, and share it in the Slack channel. So thanks everyone.

44:01 Blake Feiza: Awesome. Thanks guys.

44:02 Külli Esko: Thank you. Until next time. Thank you.

44:04 Matt Huff: Yep. See you. Bye.

44:05 Blake Feiza: Bye-bye.