0:00Okay, video three, perspective number three.
0:02What if I told you that the way you built your dashboard would fundamentally change
0:09the cost to run that dashboard.
0:11It sounds like a really crazy concept because from that you'll evaluate that probably what that will mean is people will build their charts differently to try and save on costs.
0:20And the minute you do that, you start to decouple this idea of viz best practice from actual best practice because best practice will suddenly turn into how do I make things cheaper.
0:31This transition is taking place and it's about to hit the Tableau ecosystem in a big way.
0:36And I think this is going to be felt the most through the way Tableau Next works.
0:40And so in this video, I want to explain this transition.
0:43Not to say that it's wrong, but to say that it's been happening for some time, at least as far as I know, the last five years.
0:50And I want to sort of slowly step you through how this has happened and why it's going to hit like a truck when it finally lands in the Tableau Next ecosystem.
0:58And it's not really Salesforce's fault in this particular instance.
1:02Although I'll explain that it's probably something they should be aware of and realize they need to soften the blow a little bit.
1:08So let's start first with this idea around pricing.
1:12Now the the heritage of Tableau pricing, I'm not a pricing expert, as I've said already.
1:16But if I just slowly sort of uh reveal parts of this diagram, you'll see that Tableau's always in the past works on this concept of seats and cores, right?
1:25So you bought
1:26um seats or licenses and you sl sent them out to users and for your server you bought CORES Compute Power and that was it.
1:33Once you did that you could shove as many people onto that resource as you wanted and get it across the organization as as easily as you wanted.
1:40Yes, it costs you effort and time to get that to work smoothly, but really you kind of had fixed costs once you'd bought that.
1:47You could renew your license at the end of the year.
1:50And know that other than adding licenses or i i inflation, it was basically gonna cost you the same in real terms to run the same solution.
1:58It doesn't matter how you know many queries you ran or how many dashboards were loaded on, as long as your server could handle the load, you were broadly fine
2:05And then the concept of cloud computing came in.
2:08AWS started to make cloud computing more ideal.
2:10And then you saw Tableau Online and then that got rebranded to Tableau Cloud.
2:14You even saw concepts such as web watering come into the fray.
2:17And so
2:18The crazy thing about this is that this was a very soft transition and it was really presented as a convenience thing.
2:23So uh allowing Tableau to manage your solutions for you and that would save you cost because
2:28They had all the staff and the people to do this well.
2:31You didn't.
2:32And it also enabled smaller organizations to have products like Tumba because they didn't need to have an IT back office.
2:38They just needed to know a URL and who the admin was.
2:40And that was it.
2:41They could get started with tablet in minutes.
2:43And to this day, this is still the fastest way to get started with tablet.
2:46It's a good thing.
2:48Then we're moving to this world now where I'll just delete this diagram where consumption has come into place.
2:55And this has been like a slow but quietly simmering
2:58thing inside of Tableau Cloud you've had the concept of consumption through something called prep conductor initially you used to have to get credits
3:06essentially to run this at varying levels and so you had a base number of credits available to you and then after a while those credits would run out and you'd need to sort of purchase
3:15additional credits and there have been various add-ons and components to Tableau that have broadly touched on this concept of consumption.
3:21So the more you use
3:23the more you pay.
3:23And in the Salesforce world, Einstein Analytics had this sort of consumption component already bolted onto it.
3:29And with the launch of AI tools everywhere, you've seen consumption be the thing that everything is priced on
3:35Because those queries to the AI models cost money.
3:38And so the idea is that every query you send to an AI tool is going to have a cost.
3:43And so this idea of consumption has come to the fore.
3:46And what that really pins itself on is this term that I'm going to call cost per query.
3:51Now remember that because we'll come back to it.
3:53Now when I
3:54had a think about this.
3:55I was thinking, where have I seen this before?
3:56And of course I have in databases.
3:58So if we go over to the world of the Snowflake, you'll know that they charge um for um data in a very sort of
4:07interesting way they charge you for compute and storage and everything else is essentially free.
4:12There is a little bit of a margin on on the cost to operate the platform but you kind of see that in the consumption.
4:18So
4:18I have uh a Snowflake account, I'm a Snowflake customer, I pay for that, I pay roughly $100 a month, and what I use that for is the demo data sets to do all the stuff that I do and I actually use it to store my YouTube data.
4:30So it's actually quite
4:31a nice thing now the great thing about it is that I mostly just pay for storage because I rarely use the data and when I do need to use the data
4:39It spins up a warehouse, that's what I get charged for.
4:42I get a small warehouse and then it runs the query.
4:44And so what you see here is essentially the benefit of doing that.
4:47You see, if I had to stand up a database
4:50I would essentially have a fixed cost and a fixed capacity.
4:52And when I'm not using it, I have to keep that capacity running.
4:56I have to keep those resources running.
4:57And I also have to employ a team to run those resources.
5:01But the problem then comes is when I have a surge in demand, I I can't just easily increase the resources.
5:06So Snowflake came about and Databricks as well and introduced this idea of consumption.
5:11in a in a very interesting way in that they allowed you to scale up and scale out your resources.
5:16Scaling up is essentially making them faster.
5:19Scaling out is essentially adding more nodes and more compute capacity.
5:23So
5:24Scaling out is good if you want a lot of sort of simultaneous queries to be happening.
5:28And scaling up is a good way of increasing the throughput, increasing the speed of what's going on.
5:34Nonetheless.
5:35your saving kind of came in this uh you know orange area here where you weren't using so much of it and so you're not actually paying for that and that's essentially where your saving came in
5:44Obviously, you have these surges where you do pay a bit more than you'd be paying, and the marginal cost probably is a little bit higher compared to owning the resource.
5:51But generally speaking, you shouldn't have as many surges.
5:54if you size your resources correctly.
5:56So that's in itself a component of Snowflake and understanding how that works.
6:00And so this diagram kind of helps to explain that.
6:02And so it's broadly made sense.
6:04But here's the thing.
6:06You see
6:07If we think about this for one second, in the past, your your pricing has been
6:15n has been something that has lived in multiple places.
6:17So to to to just show you this, I just want to bring out like a new area here.
6:21So I'll just call this like I'll just bring out like a um a text box here and we'll just call this um uh table.
6:27Okay, so you've had this
6:29um you know tableau uh license imagine this uh text here is the tableau invoice so you've got this invoice that comes from tableau we'll we'll give this a nice little diagram
6:39Okay, so you've got your Tableau invoice.
6:41It's essentially what you pay to use the Tableau platform.
6:44Okay.
6:44And then before that, if you were connecting to Snowflake, you'd have a Snowflake invoice, right?
6:49Okay, so you have um Snowflake
6:52Okay, I should have made this ahead of time, but nonetheless, we can do this quickly.
6:56And um you might also have let's say uh maybe AI credits that you were using.
7:01So you may be using Claude or Anthropic, okay
7:05And on top of that, you might have also had some other sort of relevant costs like related to running the service.
7:11But let's just say this was your data stack, Snowflake.
7:14Let's also add
7:16dbt because you know if you're using semantic models you you'd you'd be you'd be running an instance of that as well so let's just get this out here and just type in dbt okay so
7:25You're using a bunch of analytics tools and services and at the moment What you can basically sort of surmise from this is that yeah your analytics team would see this invoice right and this this this makes a lot of sense
7:38Snowflake uh typically tends to sit within an IT team because databases tend to be managed centrally.
7:45So your IT team would see that invoice.
7:47um Claude would probably be in the IT budget maybe probably and then um a DBT would probably sit in sort of the analytics um budget here so
8:00Um this is sort of like an interesting landscape.
8:03Now I will add another component to this which is very pertinent for Tableau Next, which is obviously the Salesforce bit.
8:09Salesforce and the Salesforce budget probably doesn't sit in either of those, it probably sits in like what I would call the marketing budget because it's a CRM tool, so it's really about sort of activating
8:21and targeting your customers.
8:22So your probably your CMO would probably look after the Salesforce budget and they'd be paying for that capability because it enables them to sell and reach to their customers.
8:32So in this world, three different organizations were getting an invoice to run an analytics platform.
8:39And in the world that Tableau Next is proposing.
8:45All of these come from Salesforce.
8:48Right?
8:49And this is really the the game.
8:51The bit that's not on here is Data Cloud.
8:53Obviously this customer doesn't use DataCloud, but what you would do is you would if you were going to sort of convert this into the Tableau Next Paradigm, you'd drag these all down here
9:04And you would add a component of a data cloud right about here.
9:10So let's just drag this here
9:12And data uh I I copied it the wrong thing.
9:17Let's just do let's just do that and let's very quickly type this in.
9:21So data cloud, I'll just put data C.
9:23Okay.
9:24So there's data cloud and um I don't know where this would sit, but analytics would probably look after this because it's kind of related to Tableau.
9:33So in this new world, Snowflake is still going to be going to IT
9:41But Salesforce really want to be.
9:46If I sorry, I just knocked the mic, so sorry if you heard the feedback.
9:49Salesforce really wants to be um
9:55This person here.
9:56So Salesforce wants to be the one invoicing you for all of this, okay
10:02And that's that's a super interesting proposition because before you were just seeing the invoice from tablets.
10:07So you maybe had a a value and a net worth related to analytics.
10:12But when you see the invoice from Salesforce and you look at wow, this is what it's gonna cost to run a dashboard, then it becomes a slightly different proposition.
10:20The numbers suddenly looks bigger.
10:22and you're having to foot the entire cost and maybe your organization isn't set up for that, it's not ready for that transition, it's not ready to see that invoice in that way.
10:31And actually to get this invoice in this way, a bunch of other people need to be involved.
10:35Maybe IT needs to be involved in the Tableau discussion for the first time because of Data Cloud, because of the way it touches AI.
10:41And so it's a really interesting
10:44shift that I think is about to happen.
10:46And then the second shift, the second really important twist here, is that because of Salesforce owns this whole entire model
10:54Because they also want to put a consumption pricing model on top of it.
10:59It makes the way you work with dashboards very, very different.
11:02And so if I go to the right here, what I've done is I've just taken a very simple explanation.
11:07This is actually very similar to how Salesforce themselves talk about dashboards in the Tableau Next World
11:13I've got a dashboard with four charts and ten users want to go and use it and they typically run five interactions every time they're on it.
11:21So they'll go and change a filter, they'll click on something, they'll change a filter, they'll click on something else
11:26And then they'll maybe go click on another tab, they'll review it, and then they're done.
11:30That's essentially five interactions, okay?
11:32But in the consumption world
11:34Each activity is a chargeable event and the four charts themselves are a multiplier because each four of those charts generates a slightly different query.
11:43And in the Tableau Next world, everything connects to Data Cloud in real time.
11:47There is no concept of an extract in Data Cloud just yet, and Tableau Next definitely doesn't have that concept.
11:53So it's connecting to everything in real time all the time
11:56And um there there is a concept of caching, but I'm not sure it's the same concept of caching that we used to in the Tableau desktop and Snowflake World.
12:04So
12:06If we just fast forward this a little bit, user loads of dashboard, user clicks, you get tooltips, you do all of these things here in the query pipeline, and then it comes all the way back.
12:17All of these things are happening in the background constantly.
12:19They're just
12:20Lots and lots of activities.
12:21It works out as about 500 individual events just for that one dashboard for 10 users, five interactions in one day.
12:30500 activities and then on top of that you gotta add alerts and subscriptions and there's a ton more there and so I'm just not sure
12:38the community or Tableau users in general or people who buy Tableau are ready for this shift.
12:45If you've come from the Salesforce world
12:47You're already familiar with this because this is how consumption has already been priced into some parts of Salesforce.
12:52If you're coming from Einstein Analytics, this isn't new.
12:55But we go back to that idea of what is a dashboard worth.
12:59To be able to sit there and tell your manager that actually your dashboard could cost more this month because it's Christmas versus last month because it's not so busy, it's it's a very sort of strange thing.
13:12And I'm not sure we've really reckoned with the impact that has on how you will build a dashboard because that will lead to behavior.
13:20around choosing more optimal simpler charts so you don't generate as many queries and at that point you decouple
13:29this best practice from the capability of the tool.
13:33It becomes a priority to build
13:37more efficient, cheaper dashboards when cost is an imperative thing.
13:42And you might say, well, if companies value the dashboards and the outputs, then they'll spend the money.
13:48And you say that, but we all live in sort of really tough economic times globally and uh analytics teams will do what they need to do to survive, to justify, you know, the budgets they have and defend
14:01uh their purpose within an organization and if sometimes cutting costs is the way to do that then that is what they will do.
14:06It's why companies look at Power BI, it's why companies look at other platforms because they're trying to save on the bottom line.
14:13And so
14:14I'm not here to put forward an answer, but I think I think as as users, we need to start answering this question.
14:22What does this query
14:24cost and we need to do that with our dashboards today so that when we see Tableau Next, when we see the price and when we start to do some modeling, we can actually do a comparison and say, hey, this is this is the right price for this dashboard.
14:39Or even better, be able to compare the value that is derived from a dashboard against the cost.
14:44That's really what we're trying to do, to be able to say, hey, this is worth doing because
14:49even though it's costing us fifty sixty thousand pounds a year to run these dashboards, it's worth it because they generate millions or tens of thousands worth more of better revenue.
15:00I'm not sure a dashboard is worth that much
15:02But maybe I'm wrong.
15:03You let me know in the comments below.
15:06Right, we've covered a range of topics
15:10The next one will be on Tableau next.
15:11I'll do this over the weekend.
15:12I won't do it tomorrow because it's the weekend.
15:14So I'll do it over the weekend.
15:15I'll have a think about it and we'll put it up in the next couple of days.
15:18Thanks for watching.
15:18I'll see you in the next one.
Tableau | Industry trends | Tool strategy | Data visualisation