Hacks in Tableau: Friday thoughts Episode 4

0:00Hey, Sam here. Today, I want to talk about

0:02a topic that drives me crazy. I've got

0:04nothing to show you. So if you're expecting

0:06me to show you something amazing in Tableau

0:07, this won't be the time where I do that.

0:10However, I do want to talk about hacks. And

0:12by hacks, I don't mean people like me who

0:14've spent too much time in front of a screen

0:16working on Tableau. By hacks, I'm talking

0:18about these tricks that we sometimes use to

0:21get ourselves out of sticky situations in

0:23Tableau dashboards and inside a Tableau

0:25workbooks.

0:26For a long time, people have asked me, "Hey

0:28, Tim, the mat players, you didn't do a

0:30video showing us how you can use it to do,

0:33you know, triple or quadruple axis charts."

0:35And there's actually a reason for that. I

0:38personally have an issue with teaching

0:40people hacks because of a philosophy that I

0:43think I've slowly developed over time.

0:46And I'm not saying that this is an issue

0:47for everyone. I'm not saying everyone

0:49should stop using hacks, because I think

0:51there is a time and a place. But there's a

0:53there's a deeper problem that I have with

0:55hacks. And it was sort of exemplified to me

0:58the other day during an exam, when a

1:00question came up, and it asked, "How do you

1:03get a user to switch between two charts in

1:07Tableau?" And it gave you a bunch of

1:09options. And it's essentially this

1:11capability here.

1:13I didn't even believe for myself that this

1:14was actually something that was in

1:16documentation, or something you should do.

1:19And this is probably a borderline hack.

1:22Okay, so you don't, it doesn't necessarily

1:24mean that this is a hack, because in some

1:25ways, this feature has gotten better over

1:28time.

1:29But essentially, what you can do is you can

1:31create a parameter, and you can then use

1:33that parameter in a calculation to control

1:36which view is being used. So essentially,

1:38your parameters then controlling whether a

1:41sheet is hidden or not. And that's

1:43essentially what happens. And essentially,

1:45you can then choose, "Oh, I'd like the map,

1:47I'd like the bar chart." And it's just

1:48switching off charts.

1:50And what's actually happening here is you

1:52're dragging all three charts into the

1:54container, and the parameters actually

1:56dynamically enabling and disabling the

1:59charts. And then the charts that is

2:01disabled sort of collapses. And there you

2:04go. Okay, so in real terms, the

2:07documentation makes this look like a

2:09feature, but in my view, it's a hack.

2:11Because for me, the right way to do this

2:13would be to give the user a capability in

2:15the user interface, along with things like

2:18containers and layouts, and control and the

2:21ability to be able to change that. And so

2:24let's get talking a little bit more about

2:26hacks and why they drive me crazy.

2:29The number one reason I don't think you

2:31should be using hacks is because of

2:33performance. Now, a lot of the hacks that

2:36we typically see, let's take the map layers

2:39options where you had essentially changing

2:41the capability where essentially map layers

2:44, well, layers are only supported in maps.

2:48So what people were doing is they were

2:49building a map, then disabling the mapping

2:51capability, the background, essentially,

2:53that turns into a scatterplot. And then

2:55from there, you can then start to visualize

2:57whatever you want. The reason I find those

2:59techniques a little bit cumbersome is

3:02because

3:03I think what they've allowed Tableau to do

3:05is essentially get away with not building a

3:08proper user interface for the product. And

3:11if you know me, I have absolutely ranted

3:13and raved for a long time about the lack of

3:16improvements to the layout system inside a

3:19Tableau to the point where I think most

3:21people in Tableau fed up with me bringing

3:23it up, and they're absolutely sick to the

3:25you know, to the nuts every time I bring it

3:26up every time we talk about it.

3:28But the layout system hasn't had any

3:30improvement for such a long time. And as

3:32people will tell you, this ends up being

3:35over 50 to 60% of the time, and all the

3:38spends just fiddling with things like

3:40formatting and layout and so on and so

3:43forth.

3:44And so I very much will literally rejoice

3:46the day we get those features. But the

3:48downside of that is because we're not

3:50getting improvements in the layout and the

3:52formatting, we're also not getting

3:55improvements in terms of the UX

3:56capabilities that we want to be able to put

4:00into these products, these analytical

4:02products so other people can use them.

4:04And so, yes, when you don't do any

4:06innovation or what's over on layout, people

4:09will try and bend the existence of what

4:11they use to be able to do stuff. A really

4:14good example is set actions. Now when set

4:16actions came out, I think the number one

4:18use case for set actions wasn't actually

4:21analytical use cases.

4:23It was in fact user interface hacks that

4:24you could then do to build things like menu

4:27systems and drop downs. I'm not going to

4:29bother trying to show you those on screen,

4:31but I will go find tweets and maybe put

4:33links to stuff that you can sort of go out

4:35and check out.

4:36But it was a real shame because set actions

4:37were really, really powerful feature. And

4:40what ended up happening is it got hijacked

4:43by users who were just crying out for some

4:46user interface capabilities to be added to

4:49Tableau.

4:50Every single feature that comes out, I can

4:52't think of any time there's a new chart, a

4:53new capability. By default, our sort of

4:56first instinct is to try and find ways of

4:58doing things that we wanted to do with the

5:00user interface or with layout or with

5:02design that we haven't been able to do.

5:06And so I actually think that the more we as

5:08a community start to use hacks, the more

5:11Tableau is actually able to get away with

5:13it. Because if I look at this document here

5:16, I don't think this is the appropriate way.

5:20I don't think it's appropriate that it

5:21takes 12 steps to give someone a user

5:24interface to be able to switch a chart on

5:27or off.

5:28You should just be able to create a menu

5:29system and the button system, link the

5:31different selections to those charts. And

5:33when you click on them, they show up and

5:35they hide. Now, they did add something

5:37called show and hide controls, which now

5:39also works for tiled content. Check out my

5:41video on that.

5:42But nonetheless, it's a far cry from what

5:44we've actually been doing for years, and it

5:47's sort of crying out for masses of

5:49improvement. Same with layers. Actually,

5:51layers is a really good example. And I'll

5:53actually highlight how this highlights a

5:55specific problem inside of Tableau.

5:57But in essence, layers, map layers, and if

6:00you had put map layers in front of any

6:03author, the first thing they would have

6:06asked us, Hey, can I do this for charts as

6:09well? Okay. layering is not something we

6:11want to just do in maps. Maps are in many

6:14ways just a an elaborate scatterplot with a

6:17background.

6:17And so people thought exactly that. And

6:19they've sort of manipulated that capability

6:21and built a bunch of features out of it.

6:23And it actually highlights what I think is

6:26a core problem when we come to product

6:28development, especially in large companies,

6:31which is these kinds of features tend to

6:34have a home.

6:35And when there is no one home for the

6:37entire user experience, as it were, so if

6:39there's no product team whose job it is, is

6:42to look at something like map layers and

6:44ask the question, hey, shouldn't map layers

6:47be a feature that's possible in all charts,

6:50then essentially, things like this happen

6:53where one part of the product, maps, looks

6:56at that feature and goes, this is

6:58incredible.

6:59This is going to help us tell better

7:00stories with maps. And it has absolutely

7:03done that. And yet, let's for argument's

7:05sake, say there's a there's a histogram

7:07team. And they how did the histogram team

7:09not look at that and think, well, yeah,

7:11that would be useful for our charts too.

7:14And in fact, and how did anyone in the

7:16charting system or anyone who looked at

7:18that didn't look at layers and think that's

7:20an incredible capability? How else could we

7:23use that to augment our analytical

7:25capabilities when we're doing stuff? How

7:28could our users use layers to tell stories

7:31with their data?

7:32And you don't have to go far to find that

7:34out. If you go to Tableau Public, surprise,

7:36surprise, a free product, lots of people we

7:39can work week out, visit the days left,

7:41right and center that highlights so many

7:43use cases for layers.

7:45And so, to me, that's a real problem.

7:48Because, again, if you'd gone through the

7:49development process, you wouldn't have

7:51arrived at the solution and said, mapping

7:52is the only part of the product that

7:54deserves this.

7:55And so it's slightly hints at the fact that

7:57they are sort of not necessarily silos

7:59within the product development team, but

8:01they are sort of camps within the product

8:03development team. And they have to sort of

8:05pitch for resources.

8:07And if one team can pitch for some

8:08resources, then they get that resource

8:10assigned to their particular part of the

8:12project. But for some reason, that resource

8:14is not sort of a holistic one, you know,

8:16while they're working on my players, it

8:19would have been interesting to say,

8:20Well, could we spend a bit more time and

8:22bring this out for all chart types and make

8:24it more of a wow feature. Instead, you've

8:27kind of got this layout that's deliberately

8:29been designed for other charts as well, but

8:32just this hasn't been sort of kind of doesn

8:34't feel finished.

8:35Okay. And so that sort of highlights a

8:37problem in general for me, because so much

8:39of where we've got to today with Tableau

8:42now needs the product to come together, a

8:44lot of the features we need going forward

8:47to make Tableau the best product that can

8:50be needs

8:51multiple disciplinary teams to come

8:53together. If you think about the layout

8:55system that doesn't just need the

8:57formatting team, the dashboarding team to

8:59come together. It also needs the charting

9:02team to come together because most of the

9:04things we're formatting involves analytical

9:06charts that we're building.

9:06We also need the server team to get

9:07involved because we want to build those

9:09things and animate them and make them

9:11beautiful. They need to be performant and

9:13fast and load in seconds on Tableau server.

9:16So how many of the things we need to do

9:18going forward require a more holistic sort

9:20of joined up approach across multiple teams

9:22, the kind of approach you kind of see Apple

9:25where when iOS 15 for example has just come

9:27out a bunch of the features work right

9:31across the

9:31whole ecosystem of Apple, they don't just

9:33work, you know, in their sort of silos they

9:36work across the whole setup. And so, going

9:38back to hacks which is sort of how I've

9:41gotten onto this run in a way, hacks really

9:43highlight sort of the weaknesses of the

9:46approach

9:46that comes in in product development. And

9:48sometimes I have to say this, sometimes as

9:51a community we tolerate it, we actually

9:53accept the fact that these hacks are the

9:55accepted way of doing things to the point

9:58where it makes it to documentation a 12

10:00step process

10:01which between sheets actually makes it to

10:03documentation, and even into an exam, and

10:06in my opinion, that shouldn't be how user

10:09interface design is done it shouldn't be 12

10:12steps to show and hide a single chart

10:14inside a tablet, and that shouldn't be

10:18something

10:18that is held as an achievement or something

10:20like that. There's actually a video a

10:22colleague of mine has done on this feature

10:24and it's had 18,000 views and it's kind of

10:26makes me cry because 18,000 people out

10:29there think that this is the right way to

10:32do that.

10:32And actually there's a better way of doing

10:33things of course there's a better way of

10:35doing things if you asked half of those

10:37people, they could probably draw something

10:39on a napkin that might not necessarily be

10:41better but would work in their heads a lot

10:43better than what we currently have now.

10:45So, the hacks, the hacks really really I

10:47think damage the quality of the products we

10:50get long term because if there's, if

10:52something never ever sort of bubbles up to

10:55the top of the pile in terms of need if we

10:58don't cry about something sort of hard

11:01enough because there's a hack that kind of

11:03suffices, then essentially it just never

11:05makes it to any developer or any sort of

11:08product managers table to be fixed because

11:10new features and new requirements

11:12to keep up with the competition will always

11:14take priority. And so this is sort of not

11:17necessarily a plea, but I'm sort of saying

11:19to people don't put up with hacks.

11:22Because if you put up with hacks, what will

11:24end up happening is over time we'll have a

11:26very hacky product, which has lots of sort

11:29of nuances I mean, it's not too long ago we

11:31had to use sheet to make buttons to

11:33progress and that was again a documented

11:35way of doing things

11:37until someone realized that's awful let's

11:39just give you a button element. And because

11:41the reason that was done is because it was

11:43causing performance issues actually had an

11:45impact on the server, and it was sort of

11:47making things go crazy.

11:49And that leads to my final point around

11:50performance, all these hacks that we do,

11:52when you actually deploy them in an

11:54enterprise setup, if you just imagine let's

11:56say you build a hack in your workbook and

11:58it's fine in your workbook because in your

12:00workbook is just you and your content.

12:03But let's say you start sharing this

12:04practice let's say you get held as a

12:06champion, and someone asks you what's the

12:08best practice way of doing this, this and

12:10this, and you share that skill with someone

12:12.

12:13Now they go build the same thing and they

12:15teach it to more people, and that sort of

12:17keeps going and before you know it you have

12:19a tableau server where maybe a small

12:21percentage maybe 10 to 20% of the workbook

12:24have these hacks in them, but these hacks

12:27don't sort of surface themselves

12:29there's no like little part of tableau

12:30saying hey I used a hack these hacks are

12:32just hidden inside of workbooks and there's

12:34no real easy way to know which ones are

12:36being used and what impact they're having

12:38but they're sitting there, sort of just you

12:41know waiting to become a problem.

12:43And where it does become a problem is

12:44especially when tablet then decides to do

12:46something about it. And now you've got to

12:48go find those hacks and remove them and do

12:50them in the supported way, so that things

12:52start to perform a little bit better.

12:55Okay, so performance is always a sort of a

12:57key thing that's especially pertinent when

12:59you run on premise for tablet server. Now I

13:02have to say most of the hacks are maybe

13:04exaggerating most of the hacks don't cause

13:06the kind of performance issues that really

13:09sort of take a server down.

13:11Those tend to be sort of bad practice

13:13things like large cross tabs you know large

13:15tables, but nonetheless, they do have an

13:18impact and in some cases they do slow

13:20things down especially when they get led on

13:22to really complex data sets or really large

13:25data sets.

13:26It just sort of multiplies the problem. And

13:28so, number one, I think it ruins the

13:30product development process for everyone

13:32because we don't sort of get the focus and

13:34attention we need on features and then

13:37number two, performance, you know

13:39performance is not something you can

13:40guarantee. Number three for me personally

13:42as a consultant is hacks are very hard to

13:44hand over. And the amount of times I've

13:46been handed over something and then it's

13:47been built with a hack and then you look at

13:49the hack and you go, really, this is not

13:52resilient, this

13:53is not going to withstand sort of

13:54resilience testing or not going to stand

13:57the test of time. Let's say I leave, I've

13:59built a hack and then I move on, unless I

14:01've documented that hack religiously which

14:04is not often the case.

14:06And then, again, that hack is sort of just

14:07living there now someone comes in to edit

14:09that they look at it. And it's essentially

14:12the same problem that Excel has had with

14:13things like VBA script where it's just

14:15basically a black box.

14:17And so, hack sort of just proliferate

14:19throughout the entire sort of community,

14:21and there's no real way there's no sort of

14:23common folklore about how things should be

14:26done.

14:27And so, listen, I'm not trying to be some

14:29some old granddad, you know, in my day we

14:32did things properly I'm not trying to be

14:34that kind of person. I love seeing the

14:36innovation I love seeing the energy that

14:38comes out of these little tricks, and

14:40actually sometimes

14:41features come out of these things is

14:42because people push the boundaries of the

14:44product that we discover features and we

14:46discover sort of capabilities, but in my

14:48view we don't get feature improvements,

14:50often enough, as a result of the hacks.

14:53Too often those hacks just become the

14:55common way of working. Now, I'm going to

14:57stop there because I think I'm just going

15:00to end up sounding like an absolute grand

15:02dad if I carry on, but nonetheless, and that

15:04's my opinion that's my view and so when

15:07people ask me

15:07Hey Tim why haven't you done that video on

15:09that little tips and tricks or how do I do

15:11this you'll see comments all the time,

15:13asking hey how do I do this or this, and

15:15often you might see me linked to someone

15:17else making that video.

15:19And in my view, that's sort of my approach,

15:21I don't make videos on these hacks because

15:23in my opinion, I think, as soon as

15:25something changes about them they're going

15:28to break, whereas supported features tend

15:30to get improved on so I can make another

15:33video, a few months later saying hey you

15:35know the

15:35map layers feature now that's available for

15:38all chart types and here's how it works.

15:42And instead you don't have to sort of

15:45figure out some sort of hacky solution that

15:48happened. My light just went out mid round

15:50that's how that's how long I've been

15:50recording so I'm going to stop there.

15:51Let me know what you think, am I being some

15:53old fogie, am I wrong, am I flat out just

15:55incorrect. Do you have an opinion on this I

15:58'd love to know, and as a hacker and your

16:01life or as a hack change your life I want

16:03to know about the good stories as well

16:06because let me be honest again

16:06hacks sometimes do get you out of sticky

16:08situations. I just don't like how much we

16:10lean on them in the tablet community,

16:13instead of actually pushing for a better

16:15product and having the time spent on things

16:17that we've been crying out for for a long

16:20time.

16:21So, thanks for watching if you've enjoyed

16:23this, check out some of my other Friday

16:25thoughts, and I did a video on tablets

16:28feature and that actually went down pretty

16:30well. Lots of you sort of got in touch and

16:33lots of stuff has sort of happened since

16:35that discussion

16:37and I've been watching some of my videos on

16:39tablo as well about that video so again,

16:41watch out some of that content in your own

16:43time it's the kind of stuff you can switch

16:45on in the background, and I do something in

16:48the background just listen to me talk

16:51because it's not often

16:52much on screen so I do check out these sort

16:54of Friday episodes as well. I'm going to be

16:56trying to get back into the videos I took a

16:58sort of a mini break what tends to happen

17:00after a big release is I tend to sort of

17:02take a short break, because when you record

17:04a lot of videos,

17:04you can't really spread out the new feature

17:05releases either because there's only really

17:07a short window where they're relevant, and

17:10then people start searching for them over

17:11time so what I try and do is get them out

17:13early.

17:14Now I've had one issue with slack, setting

17:16up slack for tab online for me at least has

17:18been really troublesome. I'm actually

17:20engaging tablet support on that particular

17:22topic to try and see if we can get it fixed

17:23.

17:24And once that's fixed I'll actually be able

17:25to show you how to set it up, and then how

17:27it works. That's probably one of the

17:29biggest features I haven't covered yet, but

17:31we'll definitely be covering that and then

17:33hopefully we can make hay on some of the

17:35other features

17:36I haven't done yet on 2021.3, but that's it

17:39for now thanks for watching and I'll catch

17:41you in the next video.

17:42Transcribed by https://otter.ai