Andy Cotgreave | A conversation on Dashboards and the Future of BI

0:00Andy Cockgreve.

0:01Many of you have no doubt heard of Andy.

0:03He's been at the forefront of our community for a long time.

0:06He's been an evangelist at Tableau and you've maybe seen him.

0:09Talking about the values of data viz, good practice.

0:12Maybe you've also seen him as a showman on the IronViz stage showcasing the event that we all come to love at Tableau Conference.

0:19Or maybe even you've

0:20purchased one of his books, two books actually, on dashboard.

0:23He's been talking about charts and dashboards for a long, long time.

0:27I wanted to take an opportunity to talk to Andy.

0:29We actually talk a lot in the background, but we just haven't hit record.

0:32So on this specific occasion I said to Andy before we say anything more

0:35dashboards let's hit record and I also wanted to bring back this topic of the cost of building a dashboard.

0:41This is really why I got him on the channel because I wanted to talk to him about how he sees this changing in the future.

0:47As ever, let's get stuck in.

0:49Andy, how are you doing?

0:51I am very well, Tim.

0:53It is really uh great to be on the channel.

0:55Oh, it's it's great to have you on the channel.

0:57I think it it it pains me to

1:00That it's taken this long, frankly.

1:02Like you have so much to share in this space and and yeah, yeah.

1:05We've talked a lot as well behind the scenes

1:07In indeed, well, I've been a keen follower of the channel for many years, so you know, better later than never, always a great opportunity.

1:15Here I am.

1:16Amazing, amazing.

1:18Um

1:18We'll get on to so many topics today.

1:20I I think it's always helpful for the audience to understand a bit about how you get into the roles that we all have.

1:28I'm a consultant, you've been an evangelist.

1:30I think it would be really valuable to just briefly touch on how did you get into into this field?

1:36Like what sort of how did you make that transition?

1:38What were you doing before?

1:40So uh alright there's two answers to this.

1:43So remind me I've got a second.

1:45So the first answer we go all the way back to my A level

1:48And I'm gonna do my entire career history in about a minute.

1:52A-level, left A-levels with a place on an R Foundation course.

1:56I wanted to go and draw a comment.

1:58I spent that summer in the Arctic and decided glaciers and geography were my future, so I did geography.

2:04Right.

2:05A geography degree I discovered spreadsheets and computers and the internet, so I switched to a manufacturing

2:12information systems and programming masters, got my first job as a software engineer, um writing uh window box software with RM.

2:21Through that so did a lot a lot of coding but also a lot of user experience, user testing.

2:26Then became a database administrator, so now we're about 2002.

2:30Got laid off from that, spent a year bumming around New Zealand on a bike, writing for magazines and doing a lot of journalism.

2:37Then came back from that, started getting back into IT consultancy, joined a business research firm interviewing CEOs of small companies, where I started.

2:47Building charts.

2:48I'll come back to that bit.

2:50Then got a job at the University of Oxford as a data analyst, discovered Tableau in 2007 and joined Tableau in 2011.

2:57Right.

2:58Amazing.

2:58There you go.

2:59Career history.

3:00But the point being, when when when people ask me how do you become a data analyst?

3:03Uh and this was something Andy Kirk did way back in 2012.

3:06He said the seven hats of data visualization and data analysis is you need to be

3:11A coder, a database knowledge person, a project manager, a communicator, a writer, and to an extent a creative, right?

3:19Yeah.

3:19Well not to an extent.

3:20I'm not going to shy away from that.

3:21You need to be a creative.

3:24Oh, you know, and then I look back at my career and I was like, well I just bounced around with no real strategy, but I acquired all those skills that I needed to be a great data analyst.

3:35So

3:36That's that's kind of how I got into the field, and if people you know want to take anything from that, those are the kind of skills you need.

3:43You don't need to be super deep in every one of them, but you need to at least be appreciative and maybe deep in one or two

3:49And then the specific the specific catalyst for me becoming a data analyst was at this this business research company.

3:57This was about 2005.

3:59Uh and basically we we need to run these league tables of fast growing companies.

4:04And we had a boss who let's just say

4:08Uh if you had a ticks box of um sociopath sociopathy sociopath yeah

4:16He might have ticked them all.

4:18Right.

4:18So he would he would walk into the research room and he would just tear down some or come down on somebody with absolute tirades of questions.

4:26That were impossible to answer.

4:27And I mean it was it was it was quite an interesting place to work, very high stuff turnover.

4:31And I was like, it's really annoying when it comes and asks me these questions.

4:35Why don't I just

4:36create a printable set of charts up that fits on one page in in Excel but can be printed out on one page.

4:43So if you have comes to me, I can be like

4:45There, go away.

4:46Yeah.

4:47Right?

4:48And so that got him off my back.

4:50And then sometime after that somebody said, Oh, you built a dashboard

4:55Bro was like I didn't know that.

4:58But through that, hacking Excel, really enjoying playing with the data, just thinking how can I communicate data to somebody to get them off my back.

5:05I built a dashboard.

5:07And that I mean that that was that that was a real catalyst.

5:10So then data analysis at the University of Oxford downloaded Tableau 2007.

5:15Yeah.

5:15And well the rest is history.

5:18The rest is history

5:19So yeah, so diverse skills.

5:22Don't hide away from the humanities, critical thinking and skills that are really important and

5:29Oh, I built a dashboard by mistake and here I am.

5:32Yeah.

5:33And I guess in a in a weird way, t so would you say Tableau was like your

5:38was your first tool into this space or would you were you dabbling with other tools like Excel and stuff along the way, if that makes sense Yeah, so it y you know it's it's

5:47Uh obviously sliding doors moment.

5:49So at that for the company prior to University of Oxford, I was beginning to get into the Excel community.

5:56You know, you know, there was loads of

5:58Excel user groups and you know calculations of macros.

6:01Oh so good.

6:02I loved VPA.

6:04Um so how do I not school Tableau map

6:07I would have become a Microsoft MVP, who knows?

6:10Um but then so I loved Excel and Access, Microsoft Access.

6:15But then at the University of Oxford, I got the job and it's like, well we have

6:18uh Oracle Student Systems is our tool.

6:22And believe me, I it still, I still sort of get a slight shivering coosebuck because it was the single

6:28worst piece of software ever, right?

6:31You know, controlled by our IT department.

6:33Our IT department were fine.

6:35Right.

6:35But it was just an appalling piece of software.

6:38And so I have frustration

6:40Uh in about September, October 2007.

6:44I was like, I I I got the Google search.

6:46I was like, data visualization software.

6:48And that took me to Stephen Fee's review of Tableau 3.

6:525.

6:52Yeah.

6:52And I was like, ah, save me from this OSS hell.

6:57Yeah.

6:57Uh and yeah, and then and then we create our own little shadow IT department running Tableau and Tableau Servo

7:04Uh uh uh much to the chagrin of our IT department.

7:08IT, yeah.

7:09It changed everything.

7:10Yeah.

7:11That's the story actually of early tableau, right?

7:14Um

7:14Yeah.

7:14A lot of under-the-desk servers and you know uh land and expand kind of I was the epitome of the land and expand uh uh use case.

7:28I mean at that point I mean at that point Tableau had one salesperson.

7:32Well they had a b a few salespeople whose territory was in the US and then they had Todd Curry whose territory was the world.

7:39Right?

7:39I mean to give if you think back how big Tableau is now.

7:42But someone was just like, yeah, please buy one license of Tableau.

7:47One license is all you need to buy.

7:49And then you'll love it and then you'll use it.

7:51Incredible.

7:52And so

7:53I in your journey of evangelism, how would you say it's changed over the over the years, right?

7:57Because um like I guess when you've when you started the role with Tableau

8:02Did you feel like you were an evangelist, or is this something that you arrived at more reflectively?

8:07Actually, this is the role I've been playing.

8:09Let's call it that, if that makes sense.

8:11You know, that's a really good way to frame the question, because obviously I get asked about evangelism quite a lot, but did I ever

8:20No, I um I started blogging about Tableau.

8:25I mean I ran the first ever user group

8:27I mean Andy Krieble and now I ran the first user user groups on different continents at the same.

8:32Yeah.

8:32I will claim the first, he will claim the first, but yeah

8:35It's all so far in the past.

8:36Anyway, but the point is I was just driven to learn and share.

8:40Right?

8:40So I also started blogging about what I was learning about Tableau and being really active on the forum.

8:45This was about 2009.

8:47And at that point, there was just this sense of thrill of using this product which was clearly going to disrupt the entire industry.

8:56Yeah and excitement that

8:58There was a bunch of people.

8:59We were all learning this stuff together and it's like, well, well how can you do that?

9:02How can you fix it?

9:03Also, it was the start of Twitter.

9:05Twitter had just started, so yes, prime time.

9:08to really be able to actually build a name.

9:10You know, it's obviously a lot harder to do that now, but that was an important part of it.

9:15And so while I joined Tableau

9:17You know at that point Tap Gangit's Tableau was like five interviews with all the co-founders and really robust um

9:25Interview process.

9:26But I had one interview and a blog about a viz in viz tooltip, the Cot Gree tooltip, right?

9:31And a butcher with a hacks, right?

9:33I remember that.

9:34I remember that.

9:34Yeah, right?

9:36So th this was basically I made charts in tooltips using Unicode characters.

9:40Yeah.

9:41And they're faster than Viz tooltips.

9:43And um so I'd done all of these cool things and I'd run in user group and so clearly they already knew well this guy is just

9:50oozing Tableau already.

9:52Yeah.

9:53Uh and so I just I just joined in Tableau as a marketing yeah, you know, I had kind of a a wide responsibility.

10:00But I was always evangelizing.

10:01It was and it was again that the company was small enough.

10:05Yeah.

10:05You know, the dream it was just go out and get people to see Tablet.

10:08And then in twenty sixteen I formally became the evangelist.

10:12And again, that wasn't much of a transition.

10:15It was just I was given the leeway to go and

10:19Be passionate um about the field make content that informs, educates and entertains, and through that bring people to Tabla.

10:28And amazing.

10:29I mean and it I yeah, and you know what a privileged

10:34an amazing job it was.

10:35Yeah.

10:36But just incredible.

10:37Yeah.

10:37Yeah, no, it is.

10:38It is.

10:39I think um as you're saying it, I kind of I feel like

10:42To an extent, every every analyst in a company whose role it is to work with data, to an extent, has a f like a morsel of that evangelism role in that job, right?

10:55And I always wonder if

10:57Sometimes companies don't recognize that.

10:59We see we see sort of centers of excellence which kind of try and business operationalize the functions of an evangelist, but

11:07They sometimes fall short because they're too procedural, right?

11:09They they don't quite capture what I'm seeing in Utalk, which is the energy, the passion

11:14A little bit of the um spontaneity actually, just going in the direction of an idea because you can, right?

11:20And so um I often wish companies had evangelists for data to go out and tell people about the value of working with data, not just

11:29the value of working with the tools that they have.

11:31So so there's two really right, so uh one of my kids was diagnosed with ADHD a few years ago and we didn't

11:39We didn't pursue a diagnosis for me, but you know, I did all the things.

11:43It's like literally most of those trades.

11:47And whether or not I'm ADHD, certainly I am very

11:51Very scattery.

11:52My mum used to when my mum was like, When you were a kid, I think, but I just used to find you playing with Lego in the doorway.

11:58And I'd be like, Well, that's where I decided to play Lego.

12:01So why would I complete the previous task?

12:02I would just do the task.

12:04And

12:05You know, as I look, I mean I was I've I was a nightmare to manage, right?

12:09Because my best content was the stuff where I was, you know, I was doing something at a keyboard, I was out for a walk and

12:16You know, this idea would come it bounced into my head and I could not do anything but follow that idea to completion.

12:23Yeah.

12:24You know, which from a manager trying to do a strategy for evangelism, those poor guys, you know, it was a disaster.

12:29But

12:30The highest performing content I ever made a tableau was the ones that came out of those ideas, right?

12:36And so yeah, uh I I was there's something of a uh an unpredictable handful

12:44Um but back in the earlier days of Tableau, that flexibility fitted exactly where we were.

12:49And uh so I rode that way for a long time.

12:52Yeah.

12:52And it means if you are ADHD, those are superpowers, right?

12:56Yeah.

12:57Exactly.

12:57Spontaneity and focus is is an absolute superpower.

13:01And if you can find employees

13:05Yeah, absolutely.

13:06And um we call them data leaders, I guess, generally, um, when we talk about it.

13:11I always think like

13:12Companies or this the the the the corporate sector comes up with like refined terms for the things that already exist, right?

13:18So data leaders to me are just evangelists for the data practice or analytics capabilities in there.

13:24in their fields and they they might be sort of sector specific, but it's it's good.

13:28Now um in in sort of your um role as evangelist

13:33I'm sure you come across a lot of uh experiences and ideas and thoughts.

13:37And I I want to I want to assume that that is what sort of led you into publishing, right?

13:42So

13:42um you've written two books.

13:44Um but I want to sort of first talk about the first experience of writing a book and then we'll talk about the second one in in a minute.

13:50So like how how how w what were you doing as an evangelist that then led you to say, hey, you know what?

13:55I've got something to share here.

13:57I need to put it somewhere.

13:58Let's start with the book.

14:00I wrote an email to Alyssa Fink, who was the first CMO of Tableau in 2010, saying I'm gonna write a book.

14:11And um at that point I was I I looked at O'Reilly's cookbooks, right?

14:16That was the m the Microsoft Access cookbook was the best tech book I ever had because it was just

14:21It was just pick and choose recipes and I was like, I'm gonna do that.

14:24Um then I just got too excited about being in Tableau and it never arose.

14:29And then at Tableau Conference 2016, Austin

14:33Steve Wexler and Jeff came up to me and said, Oh, Andy, you know, can you can you can you would you would you be interested in writing a book about dashboards with us?

14:43And they got me in right at the right time.

14:45I didn't know at the time I was the third person they had asked.

14:48You know, I was I'm always slightly dismayed that I wasn't.

14:51But anyway, I jest, I jest.

14:54Uh or do I just uh exact thank you thank you Tim.

14:59That that's the that that's the therapy I needed.

15:02Uh but that was the perfect timing.

15:04Again I'd just become evangelist, so Tableau were like, yeah, brilliant, off you go.

15:08And

15:09So that was Big Book of Dashboards.

15:16And it and and this one just hit a niche.

15:19You know, there was loads of books about data visualization at that point, but nobody had written a

15:23a newer book about dashboards.

15:25Stephen Fhew had done one a lot longer ago, but that was quite a negative tone.

15:29Negative tone and already a bit outdated.

15:32So we hit that neat.

15:33And

15:34And it was yeah, I uh that evangelism in me was like I have to write a book, I've got to write a book.

15:41You know, I realised I could write, I'd have knowledge, and it was the right time and place for that book.

15:47And so

15:48Yeah, I mean the whole process 20 months from from Yeah Yeah, about twenty months it took to from that 2016 to publication of the book.

15:59Hard work

16:01Uh serious yeah, seriously hard work can be very, very demoralizing at points, uh can be really frustrating at points.

16:10There's a lot of early nights, a lot of n a lot of early mornings, a lot of late nights, but not always brilliant.

16:15I mean delighted with the output.

16:18Yeah.

16:19And I think um I've worked in print magazine design, which is not quite the same as publishing.

16:24Publishing is like a very different thing, but very familiar with the editorial process, the actual I'm gonna call it the the the grift of putting together

16:33a production, which is what it is.

16:34It's it people don't sometimes it's not a matter of just going into Word and bashing out the pages.

16:39There are then sort of very

16:41functional steps, you know, screenshots, images, making sure they're the right resolution, um, getting editors notes back, being told half your images aren't good enough, you know all of this stuff.

16:53Back and forth.

16:55We um I mean I I I g one of the challenges of being well one of the challenges of being Andy, uh I won't attribute it to anything, is that I actually

17:04I don't like finishing products projects.

17:07No, so I like writing.

17:10Yeah.

17:10Of course when you write a book

17:12Your first draft is half of the process, right?

17:14And so that second half of the process, thank God for Steven.

17:18Uh who were who were good complete or finishers, but I was not.

17:22And images.

17:23Oh my god.

17:24Yeah, we we we were very

17:26We we asked all authors what's the what's the hardest thing about writing a book, images.

17:30So we asked Wiley, how do you want your images?

17:33They told us we delivered that and then they went they would we went to proofs and they went oh your images are all wrong.

17:40So we have to redo every single image.

17:44I can't do that, unfortunately, Jeff had the determination to do it.

17:49Yeah, yeah.

17:52I I used to design magazines uh in the university and so we like we had like fifty images in one magazine, so it was very easy to fix these kinds of things.

18:01But

18:01Occasionally this was back in the days before AI where you could do like AI enhance, right?

18:06Like so if there was an image that wasn't right, what we had to do is sort of creatively create pixels.

18:12So it's right, if this is not sharp enough

18:14I have to now go create a Photoshop scene and add stuff around the image and to make it to make it Oh my god.

18:20All right.

18:21Well hi hi here's the here's the story the audience your audience will appreciate, right?

18:25Actually.

18:26So

18:26So h hi here's hi here's a random image from the book, right?

18:29Um so tick lines uh right.

18:32Now there's a database designer, everything there is deliberate and with intent.

18:36Yeah.

18:37When we got our first proofs back, uh so we'd send a draft and this then Wiley sent us, here's your page layouts.

18:43And we were we'd printed them out.

18:44We were kind of reading the things and I think all three of us simultaneously were going

18:48Well their axis line's a bit thick.

18:50And is that the colour is that the actual colour we put and why's that gone there?

18:55And what they'd done is that the their layout people had assumed that our thin axes

19:01uh and our small fonts were accidents, right?

19:04Because they they they they they thought well they're not high contrast enough.

19:07So they had taken upon themselves to start redrawing

19:12half the images.

19:14Uh because it needed to be Yeah.

19:15And we're like, what what are you giving?

19:17It's like, well your images are all wrong.

19:19Because the

19:20The grey isn't black enough.

19:21It's like, I mean, no, then it's already sounded.

19:25Oh my god.

19:26So uh anyway, that was that was that was the moment.

19:29We're like you know, thin grey lines and thin grey lines in the background of a chart for a reason.

19:34Why don't you read the first chapter of the book and then you would have done it?

19:37Yeah, yeah, yeah.

19:39And then you get the book out.

19:40Um I think it's it's easily probably one of the most respectable books on dashboards.

19:45Um but I I get the sense you felt

19:48That you needed something wasn't said.

19:49You had unfinished business because you did it again.

19:52We did it again.

19:53So yeah, uh as of September twenty twenty five, yeah.

19:59Yeah, we did it again like like the band getting back together, we recruited

20:03a U band member, Amanda McCulloch, uh you know fiercely intelligent woman who's uh led the data visety for several years and does incredible time work focusing in health uh

20:17industry and well she became a co-host of Chart Chat with us uh back in 2020.

20:23Right?

20:23Yeah.

20:24And the the thing that was right, so the first book

20:28It's very much like here's 28 scenarios and a little introduction of uh data visualization laying dashboards out

20:37And people were like, well that's great, but how'd you build it?

20:40Well, what's the process?

20:41What's what what's the process?

20:42And you know, we didn't even touch on that in the first book.

20:46So the second book is all about the framework.

20:50Have I got the frame anyway that I haven't got it to but it's all about a framework a production framework.

20:55Um and you know Amanda spearheaded that based on her experience, but we've come up with

21:00this unique framework for dashboard development about you know where does the spark come from how do you work with the users to prototype development phase release maintenance and

21:11Yeah.

21:11Re-release or death of dashboards, right?

21:14Yeah.

21:14And yeah, that's what the new book is about.

21:19Because we've got asked it so many times and we're like, well, we've let everybody down with the first book.

21:24Um yeah.

21:25So it's a frame.

21:28You pique their curiosity.

21:30You piqued their curiosity, they wanted more.

21:36Yeah, so that's that's what led to the second book and Another Two Years of My Life.

21:40Yeah.

21:40And a Hell's OK.

21:42Yeah.

21:42But did you feel like the second time round the experience was um you knew what you were in for and so

21:47Um, I'd like to think also maybe the technology has changed over that time, you know.

21:51High resolution monitors, maybe the images were easy to take.

21:54I don't know, like

21:55I'm I'm making some I'm making some assumptions, but maybe I'm just wrong.

21:59It's just as painful as the first time.

22:02I d I

22:03Uh I don't think the process was any easier the second time around at all.

22:09Uh yeah, you you still have to put words onto

22:13paper or words you still have to put words in, you still have to make a structure.

22:17Uh and images are still a pain in the ass.

22:19Yeah.

22:20And then the other challenge is when you know with three people you've only got like

22:24Uh three three decision vectors you put in a fourth person and suddenly it just just making decisions becomes a lot harder, right?

22:32And and with with three, if you there's a democracy is great in three because somebody

22:36There's always a minority, whereas four you get split decisions on strategy.

22:41So you know, that's not Amanda's fault.

22:44That's just four people with strong opinions.

22:47Yeah.

22:47So uh,

22:50Yeah, we're all passionate.

22:52Yeah.

22:52It's harder.

22:53But that means that is a better product because you know, like with the example you gave with the the images that got sort of doctored by the design team.

23:01Um

23:02you know, those things existed that way for a reason.

23:04So actually um you can maybe say that the second book because it's it's had uh four of you involved, um uh like the experience of writing the first book, a little bit more time in the industry

23:15Um that challenge only leads to a better product, right?

23:19And and everything in the book is there is there explicitly for a reason.

23:22Exactly.

23:23I I mean obviously it's in the

23:25uh purview of people who read the book to make the decision on how good it is, but they can rest assured that more people have seen it, more experience has been put into it, and more arguments have been had over every single figure.

23:39And if you like it please put a review on Amazon.

23:42Thank you very much.

23:43Of course, of course.

23:44We'll put a we'll put a QR code up on on screen so you can just uh grab your phone and uh add it to your basket and

23:51Make sure you follow through the purchase.

23:52Right.

23:53So there's there's been this um theme and this is related to dashboards, still on the topic of dashboards.

24:00Around, well, two two concepts.

24:01So I think I'll just set the scene a little bit.

24:03In the last, let's say, five to six years, we've seen vendors introduce what I would call um

24:09Consumption pricing.

24:10So essentially paying to access your data.

24:15The biggest and best, easiest examples to talk about are Snowflake and Databricks, who have a consumption model for how you access your data from the database.

24:23Now in broad terms, I think this is actually quite useful because it allows IT teams to not have to invest into like let's say IT infrastructure for their databases, and it also allows them to scale up and scale down

24:36Um there is also I sound like a salesman for one of these companies right now, but but it does genuin it does genuinely offer advantages, especially for customers who have challenges at scale, like sudden scale, let's say

24:48Christmas, Black Friday kind of scale issues, right?

24:51Um, so that's one side.

24:52And that that's actually been quite a positive change in the industry.

24:56I see a lot of customers very happy with the way they use Snowflake, it enables data sharing and and all that stuff

25:01On the flip side though, I feel like this model has creeped into the topic of dashboards, right?

25:07So the idea that

25:09You can decompose the analytical queries between your database all the way to a dashboard and essentially put a pricing model all the way up that chain.

25:19And I feel like that is

25:20that is very much alive and well.

25:22And it's alive and well in two ways.

25:24Um the first way I see and Sigma Sigma Analytics is a is a is a good example of this where

25:29um they don't have a concept of an extract.

25:31They always connect live to the database.

25:33So um their product perceptibly looks cheaper uh on on on the price uh thing because um

25:40You know, it it's just a much simpler product than something like Power BI or Tableau.

25:44But you'll see the cost in Snowflake, right?

25:47Because you're using up a lot more consumption, you're using up a lot more.

25:49So

25:50You're sort of almost offsetting the cost in some ways.

25:52And that's not always true.

25:53I'm I'm simplifying it a little bit.

25:56But I think also we are now seeing um this idea of actually

26:00Um, we're gonna charge you for every query that your dashboard accesses from not just your database, but actually all the way up through the chain.

26:09And it made me sort of laser focus in on, hey, we're actually getting to the point where it's going to be possible to look at your bill at the end of the year for one of these technologies.

26:18And actually see that this dashboard cost us this amount of dollars, right?

26:24And at that point, this question of well, is this dashboard worth it really starts to ring home?

26:31And so

26:32There's two sides of discussion.

26:33First of all, like what is a dashboard worth and how do we assign value to a dashboard?

26:38And then

26:39uh sort of all the knock-on consequences from that, I guess, are like the the the c the conversation.

26:45So I've said a lot.

26:46I've set the scene.

26:46Yeah.

26:47I'd l I'd I sort of love your opinion at least

26:49on sort of what you've heard so far and then we can dig into it a bit more.

26:52I've written down one, two, three, I've written down four counterpoints to you today.

26:56So uh this could take a while you're here.

26:59This is why you're here.

27:00Yeah.

27:00And uh and now we we also need to put a link somewhere to the

27:06I've done a video about that.

27:08Yeah, yeah, yeah, yeah.

27:09W in which I felt you were being very

27:12Optimistic to the vision of Bell or Major.

27:16I always liked I like to be optimistic.

27:18Yeah, right

27:20So I can't be otherwise until until the doom has actually arrived, right?

27:26Right, okay.

27:27I'm really speculating.

27:28Here comes the doom, right.

27:29So I I think uh flexible pricing on the database level is advantageous until it becomes a surprise.

27:35You know, if you don't anticipate anticipate those spikes, that can actually then be a disaster.

27:41Um

27:42Let's talk about exploratory data analysis.

27:45In order to create a dashboard, you are exploring your data.

27:49The core founding principle of Tableau.

27:52was that the cost of curiosity was zero, right?

27:55Not just because we did it wasn't there wasn't a cost to query the data, but because drag and drop was so fast.

28:02Yeah.

28:03Anybody uses Used Tableau knows you can create a hundred charts in three minutes.

28:08Keep only one of them

28:10But you still got ninety nine perspectives of the data.

28:13Correct.

28:14Right?

28:14So it so you so you're getting a vast sense of the data, even if

28:18your output just so sales over time, you've you've you've understood geospatial, uh product-based things, all the other things, right?

28:26So

28:26In order to uh hone in on the best articulation of the data, it requires consumption and curiosity and exploration.

28:33So

28:34Charging people for that and saying, well, you know what, maybe me less curious?

28:39Uh I would say is antithetical to the founding principles of Tableau.

28:43Yeah.

28:43Optimizing a dashboard, you know, this is something um you you know you talked about this a lot in your video.

28:50Uh suddenly this was discussed at Tableau when they were talking about pricing.

28:54It's like well how do you design a dashboard

28:57to optimize for query-based pricing.

29:00And suddenly it's like, well, uh uh, you could put build a dashboard with 20 charts on it and 10 filters

29:09Uh you know, and there are a Stephen Few style dashboard, a big executive overview.

29:14That's a really effective dashboard.

29:16Print that out, you know, and a CEO could see an overview of every part of their organization.

29:21That's a disaster in com consumption based crop pricing because every single one of those is a query, every filter is an operation.

29:27And suddenly you're like, oh no, we can't do that.

29:29Perhaps we need to

29:31combine charts into one.

29:33So instead of using two bar charts, do side by side bar charts.

29:37But then oh my god, then you've created things that are gonna look awful, right?

29:42And and you're basically dumbing down

29:44Dumbing down the potential in order to save cost.

29:49So that where you know, I I would love the world to be when I design a dashboard or when anybody's designing a dashboard, they are designing

29:58the best uh asset for the end users to achieve their goal.

30:05Right?

30:05Yeah.

30:05And and I've got I've paid for my licenses, I've paid for my day round lists.

30:10And they know their stuff, so if that is a 40 chart or a 40 sparkline, five filter dense dashboard, then that is a success.

30:20If I then have to go, well, Mr.

30:22or Mrs.

30:22or Madam or Ms.

30:24Data Analyst, um you've also got a factor in cost.

30:30That just to me doesn't seem like a yeah a sensible solution.

30:33If anything, it feels like going back to the world of cubes.

30:39which one of Christian Chabot's first keynotes that I ever saw, he laughed his head off at Cubes because of what it does to data and design.

30:48Yeah.

30:49So I'm not a fan of anything that reduces anything that increases the cost of curiosity is a problem.

30:58I will add

31:00Um as analytics products become more as LLMs, rightly or wrongly, for good or bad, leak into or but

31:11barge into whatever adjective whatever the verb is as they get into the analytics thing.

31:19Right?

31:19And and if I if you take Ambi's mantra and go, yeah, do 10,000

31:25Uh quick prompts to LLM in this analytics session.

31:28Yeah.

31:28Yeah.

31:28That genuinely is a cost problem.

31:31Um so

31:33I adamantly believe we shouldn't be increasing the cost of curiosity and we shouldn't be adding cost to the right design.

31:41I appreciate there's a challenge.

31:43Yeah, no, definitely.

31:44And that that that point you bring out around LLMs is is interesting.

31:48I've I've always been a champion of this idea

31:51Um, let's go back three years.

31:54Tableau had machine learning in the product already.

31:56Yeah.

31:57We had explained data, ask data.

31:59We had the concept of metrics and uh RS data lenses as well.

32:04Um these concepts already existed.

32:05Now they used what I would call more statistically reliant sort of components of

32:11what we would now call AI.

32:14So these these things now fall under the AI banner and y there's even this uh feature in desktop where you can connect to a

32:21uh a table and it tells you what other people have connected to who've also connected to that.

32:27This was before AI.

32:28That would be now called like smart tables or something in today's AI world, right?

32:32So I I always sort of

32:35I always like to highlight that we had this level of of of feature set in the pre-AI world without what I would call the massive inference costs of LLMs.

32:47Yeah.

32:47And I just wonder if we've

32:49we've sort of um just replaced it and and and and not not sort of thought about where the useful places to have an LLM would be.

32:58And I've always challenged Tableau on um you know

33:02The AI features that maybe got focused on first to me weren't the weren't necessarily the best ones.

33:08There were so many other use cases that I've always thought, man, why didn't you start over here?

33:12Like this one is such a such an obvious example, right?

33:16Um but yeah.

33:17Yeah.

33:18I think ASData was not great on version one.

33:25Yeah.

33:25By the time when when we turned it off

33:27When we turned it off, it it actually began to be my entry point into datasets, right?

33:32I was like, oh actually you know you could type queries, the tokenization was great, and you could change the tokens in the query or change the bids.

33:38And it was actually pretty cool

33:40Yeah.

33:40And then we dis and then it was decided to turn it off because LLMs could do all of that.

33:46I'm not sure I'm gonna say whether I think that was a good or bad decision, but I'll let my face

33:50Yeah, right, go on.

33:52Well what do you think, Tim?

33:53I don't think I I don't think it was a good idea.

33:55Uh we we still then have that in j in the little

33:59Einstein thing on the right hand side nowhere near as good as what it used to be, yeah.

34:02And and and that ThoughtSpot was coming up with this with better tokenization and editable stuff like that.

34:08I'm now doing work with Vizu, you know, there

34:11the way you can add in the tokens in visa is great, right?

34:13So yeah so that that there were decisions to go, well hey, LLM's everywhere.

34:17Um I don't

34:19You said it.

34:20I won't say my opinions.

34:21Uh but I agree.

34:24So I think that's a challenge.

34:26And then um what was the other part of that?

34:28The

34:29The oh yeah, the features, right?

34:31The the the brilliance about the way Tableau used to do it was it was using machine learning to let the analysts do things more quickly.

34:40Correct.

34:41One of the

34:42Uh I don't know if you saw Adobe's recent tech conference, but a lot of the Adobe features they announced in the last uh it was a couple of months ago, I can't remember what I think.

34:54A lot of the AI stuff they did there was basically, here's a menu option.

34:59We can get you to do the thing under that menu option faster.

35:02Right?

35:02So it's it's it's going back to that brilliance of AI, going, well, I'm an Adobe user, I'm using Photoshop or Illustrator, and I'm still doing the work I do, but it's just all made more efficient.

35:12It's just driving.

35:13Yeah, that was the the those those great parts of those features we added to Tableau as well.

35:19And I I was really taken by that Adobe Conference because

35:22It's that recognition that if you just try and turn everything into an agent, well we could run the whole thing just by prompt.

35:29It's like well we can't.

35:30It's just it we've worked out over the last two years.

35:33It is

35:34It it's the context of every one of those operations is far too complex.

35:38And uh you know, so maybe Adobe recognizing that and maybe we'll see a swing back to the end user assisted by simple

35:47tangible a simple manageable machine learning AI-based tasks.

35:51Um yeah yeah.

35:52Yeah.

35:53Yeah and and I'd I'll also add that gener generative BI as I'd like to call it this idea of

35:59You type something into a text box and outcomes a visualization.

36:04It is only really enabled by what I would call really excellent semantic modeling all the way through your um

36:13your uh data infrastructure and uh honestly um not many companies have invested in that in the past right that's so that's such an understatement

36:23Right?

36:23Tim you want to find all the time.

36:25Have you ever found an organization that has managed to define a good semantic layer?

36:31No.

36:31Yeah, it's very tough.

36:32It's very tough.

36:33Yeah, I and um so I think

36:36Yeah, generative AI generative well generative AI B I.

36:41Good god, I don't even know what we're gonna write.

36:43Gen B I can call it, but yeah.

36:50It requires a semantic layer.

36:52Now we we've been banging on about semantic layers, right?

36:54You know, Brinton nineteen fourteen, hundred and ten years ago, he was talking about semantic layers and companies

37:00Companies don't do it because it's too hard.

37:02And businesses change.

37:04And you know, there's always Tim's in in publishing and he's got the he's got the Google sheet that actually runs the business, right?

37:10So it's just it's just too hard.

37:14It's absolutely fundamental to make this stuff work.

37:18And how do we make those two meet?

37:20I I don't know.

37:21The other aspects the the the other

37:23I remember uh you know a lot in Tableau marketing, we were like, but Google Analytics is driven by AI.

37:29Google Analytics does have a really good semantic layer, right?

37:33If you could if you measure your hits through semantic layer.

37:36Yeah.

37:36It it does have every detail.

37:38So there's an example of a really strong semantic layer um that can be AI'd to the max really effectively.

37:47But in real businesses, it gets very messy.

37:50Yeah.

37:51So if I if I bring it back to sort of this this topic on on on the on the value of a dashboard, it feels like sort of several things are happening all at the same time.

38:00Businesses are realizing the importance of the semantic model

38:03Um there's sort of I I see data engineering being probably look if if you were coming into the data space today and maybe this is controversial to say and you asked me should I be a data analyst or a data engineer, I would

38:16Push you towards the data engineer space because I think you will land a job faster and you'll you'll have way more runway than a data analyst today

38:24That's not to say that data analysts are going anywhere.

38:26I think they still exist.

38:27If anything, data analysts haven't talked enough about how much data modeling they do.

38:34It's kinda hidden the problem from their senior colleagues and now they've they've they've clocked on that the role for that is data engineering when actually many analysts were sort of shadow data engineers anyway.

38:44Yeah.

38:44That's obviously gonna

38:45spin up a bunch of um you know investment in technology and infrastructure so adding to the cost we've then got this um let's say you sort out your semantic model let's say you do start to have AI in your dashboard

38:57We still have this sort of un unrealized challenge of measuring measuring the value of a dashboard.

39:05So we talked about the cost of a dashboard.

39:07That's very separate from the value because the value can

39:11outweigh the cost.

39:12And I think in in your in your book, when you when you talk about dashboards that deliver, the thing that we're in I think you're sort of

39:19Maybe maybe it goes on said the thing you're delivering is value, all right?

39:23It's it's it's sort of intrinsic value in the quality of decision making or in the quality of um people's ability to solve a problem.

39:31And so if I sort of challenge you a little bit, how how would you I mean we can talk about this together as well, how would you start to how would you start to help someone tackle

39:44Tackle how they answer this question.

39:46Like what are the not necessarily how do they calculate it, but how what what parts of their business would they look to to help

39:52answer that question, who would they be talking to and sort of how would they be thinking about this?

39:56Because I think this is the thing that's unsaid today.

39:58If I'm gonna evaluate Tableau next and I have this dashboard and I know it's gonna cost me five thousand dollars a year to run this dashboard

40:06Like who do I go to to say is this of value to the business, right?

40:11So I'll preface this answer.

40:14with I still don't think I have a great answer to this.

40:18I think this is I agree this is still one of the hardest things to do.

40:22Right.

40:22So in dashboards to deliver we interviewed loads and loads of people uh about how do they measure value of dashboards.

40:29And there are loads of different answers, right?

40:31Some people in in the very early development phase they they go, you know, what is the decisions that will be made from this dashboard?

40:39What

40:39is the benefit y you know, they they get the stakeholders to quantify the business value of those decisions.

40:46That's fine.

40:46And then a disciplined team working together will follow up and

40:59The other common method is the

41:03It is is in some ways a really tedious method because the answers are qualitative rather than

41:11Quantitative, right?

41:12Right.

41:13I can't necessarily tell you that the dashboard I was using in tablet marketing for the last three years made me make any you know, I g I can't show you a binary field in a database that says that dashboard was worth six million dollars.

41:25But I can tell you, I went and used that dashboard every day and it gave me a really good sense of what content and what what content succeeds and doesn't.

41:34Right.

41:37Surveys are so boring and tedious, so nobody answers them, right?

41:40And and what if Tim says it was great but Tim's colleagues says it was rubbish?

41:43It's like well he's right.

41:44And then my other

41:46One of my favorite ever quotes about a tableau dashboard was uh this we used to be on the story was uh Jonathan Kappa, who works at Arlingus.

41:53He's still there, he's n he's the data whisperer.

41:55He has leeway to go and find insights.

41:59But

41:59He said to us back in 2014, it's like we were looking at a dashboard, we spotted a problem with engine production.

42:06Bang, bang, bang, two more minutes in Tableau, we changed production and saved millions of dollars, right?

42:12So I love the quote because bang bang bang two minutes and tableau.

42:15I mean that is that is exactly what yeah that's what people want.

42:19It's like

42:20Yeah.

42:20Something unto uh something we didn't anticipate was exposed in a dashboard and I got to a new answer and an actionable insight very, very quickly.

42:27Brilliant.

42:29Was that dashboard designed for that decision?

42:33I can't assume it was, right?

42:35But it's it's that thing.

42:36So

42:38I've Right, so I'm just gonna come out with it.

42:42Is measuring the value of your dashboards like measuring the value of your toilets

42:48If you took all the toilets away from your business, you would kind of you would recognize that they have value.

42:55Yeah.

42:56But you couldn't ask anybody in your business.

42:57It's like how much is that toilet worth to you because an employee?

43:01And and and I I I don't love that.

43:09It it it feels like a cop-out if I was a CDO and I tried to say that to my CFO.

43:18Yeah, right.

43:19But then the toilets do have a price too, right?

43:22But they are an infrastructure asset necessary to run an organization that has employees on site.

43:31I'll yeah, I'll I'll I'll I'll that's the best of that.

43:35Yeah.

43:36Toilets.

43:38Yeah.

43:38No, it's it's a it's a it's a good analogy because the the the

43:44If I if I if I look at procurement teams and I look at just conversations that I've been in around um you know let's just call let's call it price the price of

43:56Not just the price of an analytics store, but this this this idea that I think Tableau talked a lot about for a long time, total cost of ownership, right?

44:03Not just how much it costs to put the software in

44:06but also the cost to train people, the cost to deploy it, the cost to IT to manage it, and then do all the things you need to do um to make it work.

44:15Just because you have Tableau doesn't mean you you can't ignore like your network infrastructure to make those data connections work from laptop to database.

44:22Like all of that has to go in.

44:24But we don't

44:25put that on the price of a Tableau license.

44:27You just Tableau can only tell you the price of what they believe their product is worth, right?

44:31The value they perceive to you as a business.

44:33And but that has always been a moving goalpost.

44:35We've seen it from seat-based licensing to subscription pricing, and now we're at, you know, this sort of consumption pricing slash subscription pricing slash bundles slash lots of things.

44:46Which all in all makes it very hard to track these things as well.

44:50I'm not gonna ignore that fact.

44:53But um there are just some things, like in your car you need wheels to get about, right?

44:58If I took your tires or gave you a flat tire, you'd suddenly realize the value.

45:02of a tire, right?

45:02Yeah.

45:03But it's by no means uh the biggest cost.

45:05It's probably the most easiest thing in a car to replace.

45:08Like for a hundred pounds or a hundred dollars you could replace a tire like that and your car's back on the road.

45:13Without it it's useless.

45:14So

45:15It's very much a um I think it's a it's a it's an evolving problem and I think it's about to get even harder because what generative BI seem to suggest is that

45:27not only analysts will be building dashboards, but actually generative tools will be taking those and and running sort of running away with them.

45:36Right?

45:41Uh do we even need dashboards, right?

45:44You know, I think I think we we anybody in a position of

45:48data strategy leadership should also validly be questioning.

45:51Should we need a dashboard's the right asset, right?

45:53Yeah.

45:53Clearly.

45:54I just wrote a book about them.

45:55I have a s I have skin in the game.

45:57But if you have a business user, a task and an output, then a dashboard is still a bottleneck, right?

46:04Now if they can self-serve, then that dashboard becomes less of a bottleneck.

46:08But, you know, the dream, if I could just type into an AI, or some s some jump AI thing going, I am

46:16I need to create a new marketing campaign.

46:18What is the hot trend of our customers right now?

46:21And it tells you the answer without you needing to go to a dashboard.

46:25Then hey, let's kill all the dashboards.

46:27That's great.

46:29We are not we are we are not there.

46:31We are not there.

46:32But that is an aspiration and that gap is probably shrinking.

46:37You know, there are thousands of AI startups trying to

46:41reduce the time of that bottleneck, right?

46:44So again, one can measure the impact of a dashboard if one is a dashboard building

46:51team or standard of experts, but you need to be measuring the investment in your data and the decisions being made, which is even harder and beyond

47:01Maybe the third book.

47:03Maybe the third one.

47:03I'm opening the door.

47:05Yeah.

47:05Yeah.

47:06And and you touch on something there, which is I will say

47:09I really enjoyed um I've really enjoyed the emergence of what I would call the metric, right?

47:14So this idea of

47:17Wh which actually has always existed in Tableau.

47:19It's existed in Tableau in the sense that in Tableau I can take a dimension, um, I can take a uh a measure

47:27I can pin it to a time dimension and write a calculation and put it on the chart.

47:32And that is what we know today as a metric in Tableau Pulse or in other tools.

47:36It's basically the same concept.

47:37And I really liked this concept because what it what what I think it's allowed for is this little opportunity for people to say, I don't need a dashboard.

47:46I just need to know this one answer.

47:49And actually if you can give me a data model which Tableau and many other tools have invested in, I think the Power BI's had it for a long time.

47:56Uh you know, Sigma's just launched it in their capabilities.

47:59So

48:00Data modeling is now prevalent in pretty much all BI tools.

48:03This idea of being able to enable this this way of thinking, this sort of atomized way of thinking, I think is really helpful.

48:10I think it helps take the pressure off a dashboard.

48:12A dashboard can be better because metrics exist.

48:15Metrics can also help be

48:18uh citizens in a dashboard without you having to go and rebuild them from scratch, which makes makes spending time on a dashboard even better.

48:25So this this to me has been a positive change and I'd

48:27I'd really love to see more in that direction, sort of atomizing the components that we we call value and letting people sort of recompose them to borrow tableaus, you know, composable uh data kind of analogy.

48:41I um

48:42I the I love the irony, one of Tableau's senior sales leaders, when we launched Tableau Pulch two or three years, you know, about a year into Pulch is like, with this update, uh, with this update

48:53We no longer need dashboards, we have Tableau Pulse.

48:56And then he scrolled through a set of pulse metrics, which are a set of bands, some spark lines, and some colour indicating, so a collection of charts.

49:05Together to facilitate understanding.

49:07I'm like a dash metric is a dashboard.

49:12So the definitions, I mean uh the interesting thing about is the data analyst role

49:17disappearing.

49:18I think that's a really interesting question because I I I kinda have that sense that in a way if you can engineer great semantics and expose those to natural language then in a way that does put the analyst

49:31role at risk, but that means the engines building those views and outputs

49:40Two end users need shit hot data analysts because they're the ones who are going to be served, because it'll be these guys surfacing the charts, surfacing the stories and descriptions.

49:50So yeah, it might be the the analysts

49:53Move into the LLMs and the Gen BI providers.

49:57Yeah, yeah.

49:58Because somebody's still got to get everything we learn about cognitive science right.

50:02Yeah, correct.

50:03That knowledge hasn't got.

50:05I've always um uh everyone who knows me will know that I'll occasionally bring up Alteryx from time to time.

50:12Um trics have this concept of a curator in their ecosystem.

50:16And I I actually really like it as a term for where I think data analysts are heading in that um for a long time data analysts have been builders, they've been engineers um of dashboards and and data sets to to answer these problems.

50:30I think going forward they become curators, they become uh very deliberate um sort of a l artisans of bringing together the right data sources, the right semantic models to be able to validate that these things can answer these questions.

50:44Not so much

50:45build the question, but they they they sort of create these sort of walled gardens where people can come and be free with the way they do data analysis.

50:53And then

50:54The other side of that is uh have them listening to signals that I think AI tools can pull forward to them.

50:59For example, hey, what is someone typing in that LLM window that we don't know we can't ask yet, right?

51:05And

51:06And showing people that question, um, standardizing terminology within an enterprise.

51:10I've worked in consumer goods and one of the scenarios I've worked in when analyzing promotions is there there's like

51:1615 ways of calculating return on investment depending on the type of retailer, supplier, and all of that jazz.

51:23And to one team, it's just ROI.

51:26That's all they care about.

51:27But

51:28to the CEO, there's fifteen different ROIs that they that he cares about.

51:32Um but there is also one global ROI number and that

51:36That syntax challenge, that semantic challenge, really underpins probably what the analyst has been solving for.

51:44for a long time.

51:45Yeah.

51:46And I do feel with an LLM, a language model, we are maybe closer to actually having the technology to solve how we align that problem.

51:54It still needs the semantic understanding and it still needs good data analysts.

51:58So curator to me feels like where I'm heading.

52:01Maybe not as many data analysts, but still a really important core body of them working alongside data engineers.

52:08I think I think the I think you need to bring the word context in here, which is subtly that you can define a lot of context in semantics, but you can't put in

52:18There's no way you can put in 100% of the contact where it has to interpret.

52:23Yeah, so I think yeah, that's an important word to add to that quick.

52:27Well businesses change and innovate and they're by by definition

52:30Like I I said this yesterday to someone on a call.

52:33An LLM is only designed to look at past behavior and from that infer what might happen.

52:39They're they're like a really complicated statistical engine.

52:42If I say I'm going to

52:44It's going to figure out right, it's uh eleven o'clock.

52:47Uh Tim's not had lunch yet.

52:48He's probably saying I'm going to lunch.

52:53That is literally all an LLM is.

52:54It's nothing more.

52:55And so

52:56By definition, they can't help businesses intrinsically predict the future.

53:01They can they can infer, as inference is a topic in LLMs, but they can't

53:08They can't tell they can't suggest something that hasn't happened before.

53:12They just can't say.

53:12I mean that's an opinion as well.

53:14I'm not I'm not like an LLM scientist or engineer to be able to say that with

53:19But that same thing that same thing applies to dashboards.

53:21I mean it used to frustrate the heck out of me when we go to customers and they'd be like, Well, we've built the dashboards, we've kind of done our BI investment.

53:28I'm like wow you did you build a dashboard so I mean it's well November twenty twenty five you built it you mean you're not gonna have any new business questions from now on?

53:36Yeah.

53:36Your business is now defined and static.

53:38I'm like ha brilliant

53:39Don't want to be in your business, because of course they change, right?

53:42Yeah, exactly.

53:44Keep the analysts, keep the humans.

53:46Yeah, exactly.

53:48Um right, before we finish, I'd love you to talk a bit more about the role you're doing today.

53:52Uh is is it Visa, did I am I saying that right?

53:55Visu, yeah, V.

53:56Visu.

53:57Yeah.

53:57So so what yeah, what is your role there?

53:59Well, so I left Tableau uh

54:02Primarily dashboards of the deliver was coming out.

54:05I'd been in Tableau 15 years.

54:08I felt like I'd achieved everything I needed to achieve a Tableau was ready for a new thing, right?

54:13And, you know, the goal was, well, I'm I'm mid fifties.

54:16What am I doing with the rest of my work career time, Tip?

54:19Well, didn't I ask you, I asked myself and my wife and my family and my friends.

54:23And it's like, okay, can I go out and be

54:26Be do this independently.

54:28Um, and that's the goal, right?

54:30So the goal is I've got another book to write.

54:33Talk about that when uh the time is ready.

54:36Um what I want to do.

54:37Speaking, writing and advising.

54:40Things are better in threes, so I need to drop one of those.

54:43And and another big goal is, you know, I I have a lot of ethical issues with AI

54:49Right.

54:50But at the same time it's here and there is some really exciting stuff happening.

54:54And the analytic the analytical app landscape is

54:58is fascinating at the time at the moment.

55:01So I'm like, well what what are all these what who's the who's doing what and who's doing it well?

55:05Um and so what I'm doing with VZU for the you know it's November twenty-five so that'll be for the next couple of months is just

55:13uh exploring the product and sharing it to my channels and these you know that they they are not paying me to just say positive things right but I will say

55:28It is the closest experience I have had to using Tableau back in 2007.

55:35Right.

55:35When I was like, oh, this is doing things that nobody else is doing in a really powerful way.

55:41And that isn't because it's a brand it is a brand partnership.

55:44I'm getting paid.

55:45And I'm not saying that because of that.

55:48So hopefully people will see that over the videos.

55:52It's got a long way to go.

55:54Uh, you know, the there's there's th there are things that I'll I'll share some of the frustrations as well.

55:59But yeah, that's that's the goal.

56:01So that's

56:01I'm not that's not a job.

56:03I'm not employed by them, but it's uh I'm working with them and you know obviously doing more things like that.

56:08I'm I'm really keen um next year to

56:12I'm I'm gonna I say lift the lid on what I is what is that I I'm gonna say a can of worms in the analytics space because back in the day if I just rewind five, six years

56:23There were really like, you know, three or four big players in in the B in the BI space.

56:27So you had Tableau, Power BI, uh, and then you had um sort of what I would say uh like a cabal of

56:35uh you know, Click, MicroStrategy, all these sort of companies that have been in the space for a long time.

56:40I kind of just grouped them into one group of other

56:44Um and then you have like you know challenger tools, very few of them.

56:49Today, I feel like they've all graduated to an extent, like you know, Sigma, uh Omni, uh, Count, Hex

56:58Like the the list just goes on.

57:01Yeah.

57:02And and it almost speaks to the commoditization of BI in in a very in a in a very sort of real way.

57:09Um it speaks to the challenge that obviously Microsoft and Power BI have because lots of people are trying to enter the data space and these tools are perfect because new users to data don't have as much requirements.

57:21So actually these new tools are very good starting points, right?

57:24And so I'd love to lift the lid on this next year and just go go go on a journey and just say, Hey, if I was learning analytics today and I didn't have the baggage of having to deal with the sort of enterprise weight on me and all these things I've gotten sort of very comfortable with and I just said, Hey

57:38I have nothing but Excel or Google Sheets.

57:41Like what could I elevate to?

57:43What does that look like?

57:44Um yeah

57:46That's that that that's that's largely my goal, you know what I mean?

57:50Yeah, I'm not sure what's gonna happen next with Visual, but I I I'd love to do this with other vendors as well because there's some great potential.

57:57And and I think that having the

57:59Uh having the length in the career now that I have, I realized that the wisdom I've got, right?

58:04In the I draw tableau when we when we first appeared on the Ghana BI Quadrant, right?

58:09And all the big guys

58:11Laughed at us.

58:11They were like, you can ignore Tableau.

58:13They're a toy.

58:14They're doing nothing.

58:15Buy our big enterprise software.

58:17And oh how we laughed as we disrupted the entire industry and changed everything.

58:22And then the challenge Salesforce Tableau and the others, you know, all those leaders find that they are now incumbents saddled with a huge baggage of legacy code.

58:32And so it's really hard fitting AI into those tools because you've got these infinite anti machines that have to interface into

58:40The the fifty different connections and they could build something completely new, both of which are disadvantageous against

58:49the startups who are down here in the quadrant who have got none of that baggage, they're super agile.

58:54They don't have the scale.

58:56Uh, you know, and they might not be enterprise ready, but Tablet wasn't

59:00twenty years ago and hey, we turned the world upside down.

59:02Um so it's really fun being in the being around at this point, having been well I think for anybody, but having been through that myself.

59:10I'm excited.

59:11So yeah, I will uh join you in opening that can of worms.

59:15Maybe maybe I'll invite you back in six months to s to like

59:18Just just go through the tools we've used and tried and uh yeah see see what's in there.

59:22And I also have to add like you know your Google Sheets and your Excels

59:26they're weirdly becoming very powerful as well, right?

59:28Because a AI's allowing them to do far more than they were ever designed to do.

59:33So with with with super high risk, uh if anybody the the Jagged Frontier contact by Ethan Mullock

59:40is that you know they could that they AIs can do re can do really complex things really easily

59:48Some really easy things it just can't do, and you're like, why can't you do that?

59:53How?

59:54But that frontier is invisible, right?

59:56Yeah, right.

59:57How many R's are in strawberry, right?

59:59That frontier is invisible that edge of knowledge is invisible.

60:02As human beings, we know what our own frontier in knowledge is, right?

60:05And that's immensely frustrating.

60:06The co-pilot experience, I think, has that.

60:09When you hit something and it does it, that's amazing

60:12It's really annoying going.

60:14You're like just changed this color to red and it doesn't understand it.

60:17Yeah.

60:17Um and

60:21I've seen I've asked Copilot to do fairly basic Excel formulas, it does them, and they're wrong.

60:28And you're like, well who's so when you then tell oh Tim CEO we should sell apples, not bananas

60:34And that was based on a copilot error.

60:37Who's responsible for that?

60:38Well it's me.

60:39You've talked about that.

60:41You know, you've you've said in previous videos, it's like it's taken analysts years to build trust.

60:45Yeah.

60:45And it takes minutes to lose it.

60:47I thought that was great

60:49Yeah, absolutely, absolutely.

60:50Yeah, absolutely.

60:51Right.

60:52Andy, thank you so much.

60:53It's been great to talk.

60:54Obviously we talk all the time, but this this is a great conversation I think to have in the open um because I think you have so much to share on the topic

61:01And you come from a place of authority with your with your experiences in evangelism and the books as well, so.

61:06Um I was keen to hit record and just

61:09have the conversations we've had anyway for a long time.

61:12Yeah yeah that's great.

61:14Yeah.

61:17We had a conversation like this in the past and I said we should have hit record

61:20Um so yeah.

61:21Yeah.

61:22Uh thank you so much.

61:23Uh yeah, and hopefully we'll get you back very soon.

61:26All right, that'd be great.

61:27Thanks everybody.

61:27Thanks, Tim.