Drill to Detail - Drill to Detail Ep.116 ‘Spotify, Semantic Layers and Steep’s Metrics-First Approach to BI Tools’ with Special Guest Johan Baltzar

Episode Date: November 18, 2024

Mark is joined in this episode by Johan Baltzar, previously Product Analytics Manager at Spotify and now co-founder and CEO at Steep to talk about the role analytics played in Spotify’s growth story, the startup scene in Stockholm, Sweden and Steep’s metrics-first approach to user-centric business analytics.The Kry founders factory: Meet 15 employees-turned-foundersSteep homepageNew in Steep : Cube and dbt  Targets & BI - How hard can it be?

And I think that's a big change as well, you know, compared to a lot of the previous tools is that in Steep, the ones who are actually creating the content is not the data team, but everyone else. So, hello and welcome to Drill to Detail, the podcast series about the people, products, and new innovations coming out of the data and analytics industry. So, I'm very pleased to be joined all the way from Sweden, actually, in today's episode by Johan Balzár, co-founder and CEO at Steep. So, welcome to the show, Johan.zac, co-founder and CEO of Steep. So welcome to the show, Johan. Thank you, Mark. Great to be here. So Johan, we met actually for the first time in

person at a Cube event in London, but maybe just kind of to start off really, Johan, by just explaining who you are and what Steep is and maybe your journey to this point, really, in your career. Yeah, absolutely. So yeah, Johan Balzza, co-founder and CEO of Steep. I mean, basically I've been doing analytics, maybe not as long as you have, Mark, but 14 years at least. So I used to kind of, I built up product analytics back in the days

over at Spotify. And that's where I think I, you know, really got into data and analytics and kind of got stuck ever since. So I moved on to a couple of other tech companies based out of Stockholm. So iSettle or Settle, if you know that one, kind of card payments. Got acquired by PayPal for $2 billion. So great journey there. And then also i've been at the company called uh it's called livy in the uk it's a kind of telemedicine uh company that uh was uh you know you know i started there before covid and then covid hit so then that was quite the ride uh i

would say but basically doing doing analytics uh over again. So I started in product analytics and moved on to just like all kinds of analytics, love building teams. I think my last data team before we started Steep was at QI Livi. And then I hired like 30 people, built up a team that I was super proud of, kind of across, you know, data platform, data engineering, internal machine learning, and a lot of analytics, like a lot of, you know, analysts, I think I hired maybe 40 or 50 people over the years. I'm curious to understand, what did you do around analytics Spotify? And what kind of role did analytics play in Spotify's growth and product design and product evolution? I mean, I think it was really cool to see and be working with Daniel and Gustav, Daniel Ek and Gustav Söderström, who are still there, still running the show, and Alex as well on the growth side, I think they had this kind of clear idea that like, hey, data, analytics,

machine learning, which was, you know, as hot as AI is now, that was machine learning, you know, 10 years ago. I think they had this kind of clear understanding back then that, hey, if we want to become like a big global, you know, music platform platform the global music platform we need to invest a lot in doing things smarter than than the other competitors i mean now there aren't that many competitors but you know you know 10 years ago it was you know it was the wild west you know when it came to music there was a weezer and there was you know pandora and like youtube and youtube music and then so we were always like you know thinking about this as a global opportunity they always had this way of saying like you know go big or go home uh which we kind of lived by and always you know being slightly afraid of like

you know at some point google or apple is going to come in and, and just like, you know, totally squash us. So I think we always try to be, you know, using data, using analytics to, to work smarter. So, I mean, to be more specific, I guess I was working on specifically on building up the product analytics practice. So we were investing in like, how do we do A-B testing? How do you do experimentation at scale? How do we kind of instrument the app? You know, we had like logging at every touch point across the app, whatever kind of content you were being served or you were seeing or interacting with.

So it's like a, you know, tremendous amount of data and still is. I think it's still some awesome amounts being generated over there. And we're just trying to use that. Like, how do we help our product teams to make smarter decisions faster? That was it. So kind of building up teams across Stockholm.

And we had a big New York office and still has. When you think about, I suppose, the inbuilt advantage that Apple have got with the the way that you know with the um with the kind of the 30 that they could take from spotify for for you know you to think that spotify is still there and still still kind of like heavily using you know great engagement and so on it's testament really to the product design and the team behind it really yeah i think so i mean i think it's the it's still the same people right who's kind of uh still daniel it's gustav it's alex and the team and they've had this kind of amazing focus and i think vision over the years and just being better and better at what they do so i uh yeah really impressed by by where they got to okay so so what's your tell us about your role at steep and give us steep's elevator pitch really okay yeah so i

mean the background to steep was basically me basically me kind of building these data teams over and over again and being quite frustrated with what I call like the last mile of analytics. So if you think about like the first big part of analytics and data is just like the whole modern data stack, right? 10 years ago at Spotify, this was a mess, right? We had to kind of build our own internal tools and there wasn't no DBTs. We had to build our own internally at Spotify.

It was kind of Hadoop and MapReduce and a lot of Python. And now it's like so much better. You know, we're all kind of on the same modern data stack. It's all cloud. It scales nicely and it's all SQL mostly, which is good, I think. So, but what I was frustrated with is kind of working in these kind of consumer tech companies, which had a lot of ambition when it comes to using data.

So we want to kind of use data everywhere across the company to back every decision if we can. And my frustration was just like, how do we, we have a lot of amazing data in my data warehouse, like it's modeled, we can do a lot of cool things with it. I have a team of really smart people. But how do we make this data? How do we get it out to all the people that actually need to use it? Right? So all the business people, the actual end users are like marketing and operations and product and tech and finance all of those people that are you know specifically non-data people right so how how do we how do we do that right

so we were kind of i mean i've been using i think all the different tools uh out there the bi tools to to get this job done uh and i was very much kind of frustrated with the whole paradigm of how this works. So it's basically like, I need to have higher, really smart people and then I need them to be kind of internal customer support. They need to kind of answer questions from anyone across the business. I hear teams, they have like an Ask Analytics channel on Slack

and it's kind of getting flooded with questions all day. And then we're trying to kind of build content for them right so it's like oh marketing team you want to follow your campaign performance here's a dashboard for you guys and then if you have more questions it's like okay we'll need to kind of iterate on the dashboard to build more dashboards you know and you you kind of just repeat that over and over again. And after a year or so, you end up with like 400 different dashboards and, you know, oftentimes a bit of a mess, right? So we were kind of like, how can we solve this problem in a better way, right? So if like the business users are the ones that really need to get their hands on the data,

how can we kind of like, is there a way of working with data that makes this kind of immediately useful to all those end users, right? So I think we got really excited about some of these ideas that have been bubbling in the data landscape. So like semantic layer, and specifically for us, like the idea of metrics first BI. So could we use kind of metrics instead of dashboards

as kind of the core component, the thing that we're defining as a data team, the thing that we're owning, the thing that we're shipping to the organization? And could we like radically go all the way to kind of build

a product that empowers them to use metrics and dimensions freely to solve all their questions themselves? So they don't have to go and ask and we don't have to build content for them. So I think that's in essence is what Steep is. We've gone all in on this. I've teamed up with Nino Höglund, who was like the chief design officer over at iSettle. He was part of the whole journey there for 12 years. And we kind of teamed up because we both saw the problem and we saw the potential. And so we're trying to build a, you know, Steep as a very different kind of BI tool.

As it's kind of, sometimes we say like it's the first BI tool that's actually built to be used by everyone. So the data team becomes, you know, kind of works with the semantic layer, sets things up, defines things, and then they're shipping metrics to everyone. And then everyone else is just like exploring, analyzing, and actually creating their own content. Sorry, that was a bit of a rant there. Okay. Okay. So if I was in the audience now, I'd be shouting, but surely this has been solved by many, many sort of BI tools. So the definition of a BI tool, I suppose, is that it does make data available to the average kind of person, really. And you've got, you know, you very kindly referred to my years in the industry,

but I remember i was back working with business objects and and cognos and and through all those tools there and they all they all feature a semantic layer um they all uh user-friendly self-service so so what's the i suppose in your mind what is the what's the what was the what was the problem with those existing tools and why were they not in your view as self-service or as business orientated as as kind of you wanted really yeah i think we've been i mean sometimes i i try to avoid the the word self-service because we've been using it to your point right for the last 15 years or so uh so it's i think it's a bit watered down in a sense i mean i think the fact

that we're still talking about it is because we haven't solved the problem really um i think what we're seeing is that the the bia tools at its core are built for us like the data people they allow us to do a lot of cool things we can create whatever visualization we need if it's a good tool and then we can kind of we can build content but then everyone else just they get what we give them, right? They give this rather flat view of the world, you know? So they ask for something, we try to give them something that solves their problem. And usually immediately they come back with more questions, right? They're like, oh, but now I want this, or now I want to look at it slightly differently.

And we have to kind of painfully try to make it more flexible for them again and again. We're embedding logic inside of these visualizations, right? We're kind of like all the way to the right when it comes to embedding and building logic, and we're kind of endlessly repeating ourselves. So I think what's amazing about this way of working is that A, of course, like you're centralizing your logic, you're having this semantic layer as a thin layer of definitions on top of your data warehouse. And the really amazing thing we tried is like, hey, could we use metrics and dimensions as the core abstractions instead of tables and columns and graphs?

We're talking about metrics and dimensions. And then we built the entire tool to be designed to be used by everyone. So it feels more like a Figma or a Notion or Slack, that kind of UX that is very approachable, very intuitive, extremely fast, extremely nice to use, if I may say so myself. You should try for yourself uh and the cool thing about it is that it solves that unsolved problem i think that we've had for so long it's just like can you just give this product to any kind of person in the company like a marketing manager a product manager uh you know someone in finance, and just go for it.

Just like, there's a lot of metrics and dimensions here. Go. And it actually works. They can just pick this up. They will start exploring metrics, analyzing, kind of composing views out of metrics. And I'm happy to explain what that means. And then actually creating their own reports.

And I think that's a big change as well, compared to a lot of the previous tools, is that in Steep, the ones who are actually creating the content is not the data team, but everyone else. Okay. I mean, one way, I mean, one of the things that,

arguably, what we've been saying there is if you hadn't seen Steep, you would be thinking, well, I've heard this quite a few times before, you know, and I think it's, for me, the reason I wanted to get on this kind of recording was actually when I actually used Steep, it was immediately obvious to me how it was different to what i'd seen before and it sort of reminded me a little bit of the olap tools of my my kind of youth really so so i think to sort of maybe to walk through the kind of the the the i suppose the user workflow i mean it's it's when you log in you see typically you see you see a list of metrics you know you don't see um like say with looker you might see an explorer with a whole list and list of kind of various items which could be dimensions could be attributes there could be measures and so on you know steep has that very kind of dare I say sort of swedish you know sort of design approach you know or certainly very minimalistic approach where you go in and you get a list of metrics and you start from there yeah maybe just talk talk us through the thinking behind that and the rest of the workflow around that yeah no I

think the metrics catalog as we call it is is kind of at the heart of this way of working i think that's usually where people kind of get like aha like that's uh now i get it like this is different you know because previously like you would have like oh you know you can go into your tables or columns or it's or it's like an you know uh looker type semantic layer where you're using the explores and the measures and dimensions it's still a lot of work like you need to find the right stuff you need to combine them in the right way you know and it's if you ask a business user no it's not very user friendly no it's not approachable like you need to learn a lot of stuff to really use it and for us in the field like we've done this for many years so to us it's like it's so obvious you know it's easy. I can do this. And then like, yeah, but not everyone else can do it, which is the whole point, right? So with

metrics, you get this nice, to your point, the metrics catalog with nice names, nice documentation, you can create this kind of structure for the entire company. And then the cool thing is that this becomes like a go-to place. So anyone in your company, any end user, they can just kind of find what they need, you know? So my example, like you work in marketing, you want to see how the campaign, you know, had an impact on regs or, you know, if that also impacted activation and then you can just find what you need and just, oh, here's regs, click on it. Cool. Now I'm getting a time series and I can start exploring changing time grains periods i can break it down by any dimension and then the cool thing is that you can actually then compose

stuff right so you can overlay any metric on top of another metric and we try to just automatically align them and you can filter and drill down across metrics if there's an like an aligned dimension if there's a country in both metrics it will just work and we try to make everything intuitive streamlined super fast using a lot of caching and just make it work for for all users okay so you said something you said composable analysis there you said a couple of times now so just to be clear what do you mean by composable analysis i think that's one of the cool things which might not be immediately obvious with this metric based way working is that we're like we're when you're using a metric you're

clicking metric you're on that kind of explore view for that metric and when we're running the query the semantic layer is kind of compiling that query with the current parameters from the ui and then getting those numbers back from your data warehouse. And then we do a lot of caching on the way. But that's like, we're pulling that data for that metric in isolation. And then you can just add another metric on top of it. And then we're running a separate query just in the same way. And then we're aligning them across time. So they all have a primary time dimension, and then we can figure out how we align them across time. And you can do that with also targets and budgets. You can overlay comparisons and built-in window functions. And it's just like having different layers in Photoshop. You can

just click and add them on top of each other, which allows for a lot of cool analysis. So you can compare how metrics move together. You can kind of drill down in the same metrics across the line dimensions. And you never have to kind of go back and prepare the data for this. Kind of, it just works. Okay.

Okay. I've heard you mentioned, I've heard you comparing sort of Steep to Figma in the past and talking about how figma was you know by definition it was it was collaborative it was it was multi-user and so on is that something you're aiming for with with uh with steep as well yeah i think so i mean the the cool perspective is is working with my co-founder nino there who's who's not a data person but a design person and he was kind of comparing this kind of looking at the the journey that the design field

has done from you know way back you know they were using tools like photoshop to design ux you know which was super powerful you know and if you're a good designer you you knew the ins and outs of that you can do whatever you wanted to do but of course like no one else could kind of like you know go in and collaborate right so you're kind of like, you know, go in and collaborate, right? So you kind of like, you know, don't disturb me. I'll perfect this design and then I'll throw a file over the wall. And I think we've been super inspired by Figma coming in, not just like a modern tool, but like it actually changes how you collaborate and how you work so it makes it so much easier for more people to kind of join the fun join the design process product managers and business folks and and developers i think especially and it's just like you know yeah it changes how you work with design inside of a company and we were kind of

looking at you know bi and analytics and we were saying like hey it's totally the same thing like maybe even more so for bi that it can't just be for us data specialists like it needs to be for everyone you know if we want to have you know we see a lot of tech companies consumer companies e-commerce companies big b2b companies they're all super data intense generating tons of data and like every employee across the company potentially wants to look at data you know if not weekly maybe daily right so it's like how do we how do we serve everyone like at scale and then we can't do that with the model of them going and asking a centralized team for help and then just like building you know bespoke content again and again you know it's like that doesn't like it

doesn't make any sense like why are we doing this well it's the way we've always been doing it so we're just like well we need the same kind of change you know we think it's possible to make this easier and more streamlined and then we can actually change how we collaborate between the data and the business side yeah it's about to say because i mean one of the i suppose one of the downsides of a figma for example is if you look to our figma instance it's like a sort of dumpster fire of of kind of images and and disorganized things in there um how do you avoid i suppose the more democratized you make access to things how do you avoid it you know becoming like this kind of mess really you know what's the way in which you i suppose put guardrails on that collaboration

and make sure there's still some governance there yeah and i think that i mean that's the the beauty of the semantic layer and the the metrics catalog it kind of like it it's a forcing function to you know kind of force you to work in a very systematic way right so it says to your your data people it's like stop being the bi specialist and start being the analytics engineers, right? So you need to like more systematically, you know, sit down, talk to your business folks, like decide once and for all, what are you going to call that metric? Like what's revenue? Is that one metric or multiple variants of that? What do we call it? How is it defined? Which table in your data warehouse is kind of the canonical data for that metric, right? And then once you've done it, it immediately becomes available and useful to everyone in your company.

And then maybe you need to go in and tweak it once in a while. But it's an extremely structured, very nice way of working that gives you, you know, usually I talk about like, you know, built-in consistency. It gives you this kind of common language and consistent numbers because we're all using the same metrics and we're calling it the same thing. And it will always give everyone the same numbers regardless of who's using it. Okay. So we met at a Kube event in London. And it was the integration with Kube that I guess got you invited there as well. So how does this work in terms of definition of a semantic model? Is it your own definition? Do

you pick up the one from Kube? Where does dbt come into this? What's the kind of integration story with Steep? Yeah, I'll tell you how we're kind of thinking about it. And it's an ongoing story, I think, in the data field. There's a lot of interesting things happening with semantic layers, and you have Snowflake as well investing in this. I think when we started out, of course, we started with our own native built-in semantic layer with Steep. And as we're seeing our now partners also investing in semantic layers, we were saying like, hey, it makes more sense that it shouldn't be like our semantic layer versus your semantic layer.

If possible, it would be better if we can align. It's better for all of us in the ecosystem if things connect and work together. And if we're not diverging in terms of standards, but if possible, try to work together. And if we're not kind of like, you know, diverging in terms of standards, but if possible, try to kind of come together. So I think that's been our philosophy with Steep is that we, as much as possible, we try to work with dbt, dbt cloud, and with cube and partners like that to make sure that, hey, you can use our built-in semantic layer in Steep. You might as well just connect it to dbt cloud and define your semantic layer there.

You can plug it into cube, work with that. And we're trying to give folks option and trying to make all these technologies work well together, I guess. Okay. Okay. So specifically, how does the cube integration work? I mean, you've got you can you import the kind of the semantic how does it use it live i mean what what's the how does the integration work and

and um and uh what um i suppose limitations are there and what benefits you get from that approach now we love working with the cube team i think they're really really smart folks um and practically the way it works is you you kind of up, you enter your cube credentials into Steep. We're using the REST API. We're getting all of their metadata from the semantic layer you defined in cube. And then we're importing that into your Steep workspace. And there's a convention there for how we're kind of um interpreting i guess the concept of a metric so there's a called a you have cubes in cube and views and that's what you're using to

define metrics and then you get that as a native concept into into steep and then basically we're we're just kind of when you're using metrics that are coming from cube, we're just calling their API to ask their semantic layer to generate the SQL and actually executing that query for us. So instead of kind of using our native semantic layer, we're just kind of switching in the back end and using there. And for the end user, it's totally seamless. They don't even have to care about which semantic layer is kind of like producing producing their metric it's just metrics to them uh which is great so you can even kind of use these things uh just next to each other right you can have native metrics

and cube metrics and dbt metrics okay what about dbt i mean obviously there's the dbt um semantic layer do you integrate with that and uh any other integrations with dbt? Yeah, so we integrate with dbt cloud and the dbt semantic layer. Same story there. I mean, they acquired, you know, transform and the product metric flow, which we think was great. So super happy to see that. And, you know, as immediately we started investing in a very nice integration with dbt cloud as well.

So it works just like the cube integration. I think they have a slightly better concept of a metric in metric flow, which makes sense. So it works really nicely with Steep. And what might have interest. I mean, in Kube, it's a bit more general. It tries to be like a universal semantic layer. The views are used for different things.

I'm getting into a lot of details here, but there's a convention for how do we translate measures, how do you export measures using views in Kube into metrics in Steep? It's quite straightforward. With dbt Cloud, there's the native concept of a metric, of course, and there is a semantic layer. So then it's really kind of a one-to-one mapping there, if that makes sense.

Okay, so you can also define you can also define your metrics in cube directly so is there anything are there any kind of i suppose things you can only do doing it directly or is it sort of parity across all the different sort of semantic layers i would say we're currently at like i don't know 95 percent kind of feature parity across them so it's like really high it works really well and we have like you know big customers that are you know going going all in on these different setups so it's really like about giving giving folks the options right the choice um i think more and more like now we're trying to kind of push the boundaries of like you know how many more advanced bi problems can we solve

using semantic layer and metrics so now we're kind of getting into uh like a new feature we're launching this month uh is called entities so it's basically like how do you do row level capabilities using a semantic layer so from a metric can you kind of mark a point and just say like, show me the customers or show me the users or the products that were kind of like driving this metric at this point in time. So then we're kind of extending, like that's not a metric.

That's what we call an entity. And there are kind of similar, you know, corresponding concepts, I guess, in both Kube and dbt cloud. So there's like an ongoing journey here for us. It's like, we're trying to really kind of similar, you know, corresponding concepts, I guess, in both Qube and dbt cloud. So there's like an ongoing journey here for us. It's like we're trying to really kind of push the boundaries. And then we're trying to make that work really well for all the semantic layers that we're supporting. Do you think you'll ever support LookML as a semantic layer?

Or is that maybe a step too far? You know, we've been thinking about it, actually. And we had another consultancy out of Germany who were kind of asking us to consider it. I don't know. I think it depends if customers want it, I think. We always try to work closely with customers and partners to see what do people actually want to use and so far i haven't heard a lot of folks asking for like hey we're we're just using look ml as like a standalone semantic layer and we're looking for tools that work well with it i think i haven't heard that yet so i suppose the problem the problem

with supporting the camel is you've then got the look you've got the local license cost to pay as well exactly i mean with that so so it's not exactly the most sort of you know i suppose cost effective way of doing things really no exactly i think that's that's the thing like if you're if you're investing in looker i mean you're already paid so much money ai right so everybody's got to have an ai um so either co-pilot or or conversation analytics um featuring their product are you gonna fall for that as well? Or are you going to do something sort of innovative, really? I think hopefully yes and yes. Yes, of course, like AI is, you know, it makes so much sense to combine AI and semantic layer. And we're, of course, not the only one to figure this out. I mean, that's, I guess, that's a big point of like Cortex and Cube and their AI features as well.

It's just like, because you have, I mean, you've heard about this before, like the AI being this, you know, hallucinating thing that is like on one hand magic, on the other hand, you can't really trust it. It's a bit of a black box. So if you're asking it to just generate SQL freely, then like that's cool as a personal use case but it doesn't solve the company use case

of we want to trust the numbers and look at the same numbers together so but the semantic layer does solve that right so if you're at if you're combining you're asking the ai models to actually use the same components as the humans are the same metrics and dimensions then suddenly like you have this great combo so so yes we have ai features in beta we're thinking a lot about like how do we do something again and we always try to have that as like the the principle for everything that we do is that it should be intuitive and useful for everyone right so it should like help to engage and make life easier for all the end users across your company. And then of course,

make your life as a data team easier as well. So we don't want to introduce any more headaches, like we want to solve your problems. So back to Stockholm and Sweden. So I spent three enjoyable six months working for Rebtel actually years ago in Stockholm helping them put in DBT and Looker and I really enjoyed working in Stockholm I noticed there was a quite a sort of startup scene the tech scene there so let me just paint a picture

of Stockholm as a kind of as a good place to start to I suppose incubate a startup and how does I suppose the Stockholm and Swedish maybe mindset influence how you a startup and and how does i suppose the stockholm and swedish maybe mindset um influence how you do things really and how you approach this kind of problem space yeah i think for me i mean i think we've been super lucky to have a couple of big successes right so we've had like spotify of course that we talked about arna is another uh really impressive company there's like big gaming with king King and a couple of other kind of like big fintech companies as well. So I think we've, you know, we've, I think the good thing about being in this smaller, pretty tech savvy market, it's a great place to start building your product so you have like this smaller domestic market where you can kind of you know you can work with design partners you can kind of perfect what

you're doing there's like a nice local scene of early adopters and enthusiasts and then the thing about you know building a startup out of you know northern europe or scandinavia is that you know your local market will never be you know where you end up right you're always kind of like immediately thinking about what is the global opportunity from day one so i think if you're if you're out of germany or france like or or maybe the uk you're like it's a pretty nice you know domestic market and maybe maybe that will be enough, right? And you're kind of focusing there for a long time. I think for startups here, I think that's a lesson I learned from Spotify.

It's like kind of thinking about the big global opportunity from day one. So I think now we're talking, it's like amazing talking to a lot of data teams across the world. And everyone likes doing kind of video meetings remote everyone's buying remote these days and it's so cool to see that everyone there's a lot of teams on the same data stack globally like we have customers in in japan in the u.s you know talking to folks in in in over in israel across europe and there, you know, you can talk to someone. It's like, yeah, it totally recognized your situation and your stack and what your challenges are, you know.

There's some minor differences, but mostly it's like, yeah, it's the same problem globally. So I think that's how we're trying to approach things. We're trying to build a product that does this thing that we're doing, you know, best in the world if we can and just working trying to work with the right kind of customers that are you know pushing us in the right direction i guess do you think the way that swedish people work i know it's obviously a massive generalization and stereotypical stereotype whatever do you think the way in which you work

as a team and people work um together in companies in in sweden is a bit different as well and and maybe kind of led to some of the successes there i mean just as an example i think when i worked obviously over there before it was everybody sort of had a voice and everybody was very kind of um there was no real i suppose hierarchy about suggesting things and so on is that is that an influence is that really one experience at one company or is that common to a lot of sort of swedish i suppose companies yeah i think that's really ingrained in the swedish work culture that anyone can speak up and do speak up right and i mean it's not always it's not always best but i think it at its when it works well i think it allows us to have form very tight-knit teams where like

everyone can kind of contribute from their perspective uh and then i think we've you know you also of course need to learn to be like okay someone needs to put the foot down and set the direction and everyone just follows but i think like we can have really with the right kind of teams and the right kind of people you can have really strong execution when it comes to kind of thinking about the problems from a lot of different perspectives and doing that in a very collaborative way and i think that we're i mean so i mean i love i love product development and it's been amazing kind of you know having working i worked kind of with a lot of product teams during the years, but it's great to be part of a product team

and actually hands-on crafting a product. And it's really a team effort, right? So it's all the developers bringing their expertise, also their product experience. It's the designers, it's product leaders, it's data and analytics as well. So we need a lot of eyes on this. It's not an easy product to build. I mean, BI tools are famously, you know, big products, you know, and we're trying to build a very powerful and big product, but we're trying to make it like super always kind of keep that simplicity and making it intuitive for everyone so it's uh it's a daily challenge i would say but uh

i don't know love love the team that were interestingly i think one of the things that again struck me was steep when i used it was was again you could be sort of uh it could be seeing what you want to see but it sort of had that scandinavian sort of minimalist um i suppose design which which again i could just be sort of like just projecting onto that, you know, what I think of Scandinavia. But it's quite sort of, I suppose, the design is quite sort of, is not so sparse, but it's kind of, it's not cluttered and it's got a simplicity to it. First of all, was that intentional? I guess it was. But secondly, how do you avoid the long-term doom of all BI tools that they become this kind of, you know, all bells and whistles and they lose that sort of like clarity of

design? You know, what's the design thinking behind it? And what's the roadmap for the product to avoid it becoming just yet another all bells and whistles BI tool? No, I think that's our main challenge, really, because our ambition is to make this like we're doing BI in a very different way. It's pretty radical saying like, hey, data team, stop building dashboards.

And now we want to enable everyone to do this using metrics. So it's a pretty radical proposition, but it really works. And when we see the customers going all in on this, it's like a transformative experience for the entire company. We just changed how we work. So I don't know. I guess that's been, you know, we're not kind of designing things to be pretty or minimalistic for its own sake. We're always doing it to get the right kind of outcome, right?

And like, can we empower more people? But we always want to empower them to do a lot of cool things, have a lot of flexibility. So we're always adding more nice ways to do more analysis, more visualizations, or create even more customized report. So yeah, I don't know. It's an ongoing big design challenge in how do you add more stuff and make it more useful without kind of making it a mess? But we're kind of looking, it's interesting to see. I mean, for us, we're always looking at tools like Notion or Figma that I think with a pretty slimmed down design tool, and then they added design systems and prototyping and community templates and all sorts of things. lot of different modules a lot of different functionality uh and still you know keeping the core experience pretty easy to kind of pick up i mean more and more you can see them struggling with it but but it's a lot to a lot of extent i think they've been successful yeah so so how else do people found it so how else will somebody find out more about this then so uh what's the way in which they can get a trial or they can get sort of a you know a free account or whatever to see how steep works yeah we try to be

very flexible and easy to work with you can head over to steep.app and then you can you can just immediately you know sign up with your google account you get a workspace you can get started with demo data or just hook up your own data warehouse and get started and it's it's actually free for up to three users so you can totally just play around with this within your data team and then just get in contact with us we'll help you set up like uh if you want to test this out i usually recommend to actually like you should test this out with your business folks right it's fun to play around with with the data team you should learn the tech um but the big magic moment is when you start inviting your your stakeholders and your business folks because they will get something that they never

had before so then we we just you know talk to us we're we're we're friendly friendly people and we'll help you get sorted with uh with the, Johan, it's been great to speak to you. Thank you very much for coming on the show and best of luck in the future with the product. Thank you. Great to be here.