"In other words...it will inevitably collapse into the same singularity: Database connectors; a code-based query editor; a code-free query builder; lots of charts; dashboards; self-serve."
And to be clear, "lots of charts" will include "Map of New Jersey" as part of this singularity.
I prefer the Kimball Group DW/BI term. The DW (dimensional data warehouse using conformed Dimensions and Facts in a star schema) is the core, and will help any BI tool, even if it only copies the star schema directly.
Kimball Group recognized the front end BI client was important, but almost of all of their work focused on the data warehouse (dimensional) and the ETL required to shape the data, and more importantly, the planning work with business to ensure success. I need to reread the Kimball Data Warehouse Lifecycle Toolkit book, which focuses on the entire process, instead of just modeling. https://www.kimballgroup.com/data-warehouse-business-intelligence-resources/books/data-warehouse-dw-lifecycle-toolkit/
Gotcha, yeah, that's the part of Kimball I'm more familiar with. Though in some ways, I think that semi-standardization makes it even easier for everything that sits on top to become BI, because it makes BI seem almost within reach. Like, if you have somewhat standard models, there's predictable patterns to it, and all of that, it's really easy to think "eh, we could just build a couple charts, right?" Whereas if the data underneath is a mess, I think people would hesitate a bit more to get involved with it.
Kinda? VCs seem to have two minds about this stuff. They tend to like pitches that could be something different, but once they're invested, they become more risk averse and want to run the better understood playbooks. So I think raising money on a different idea isn't bad; staying different is tough. Having money doesn't make that easier (though it makes you seem more credible when you say you're gonna keep doing the uncomfortably different thing).
Yeah, absolutely. And that's why it's hard - it usually doesn't work. To paraphrase an email someone sent me:
“If you want to be different, it takes will, not wits.” Also, if you're that stubborn, your company will probably die. But if you want to build something different, that’s the chance you’ve got to take - if you want to build something that probably won’t die, you’re gonna build BI.
hey Ben, nice semi-ironic write-up, grazie) what do you think about Quadratics? They seem to offer similar to Mode use-cases and functionalities and they are open-source.
Word, so this was kind of cool. I like the core idea. It's actually somewhat similar to something I wrote about here. It's not quite the same thing, but the closest I've seen. https://benn.substack.com/p/the-next-billion-programmers
That said, it feels...confused? Like, is it a spreadsheet? A BI tool? An application for code-lite development? I think it's got a long way to go to be anywhere near competitive with spreadsheets, and even further to go to be a BI tool (if the only charting option is writing custom Python, you are completely dead on arrival as BI). So I think the interesting thing it could do is the third one, where it's more about trying to make basic scripting - which can be realllly powerful - much more accessible. That's a tough sell, but I like the general idea.
(And, the fact that it's a tough sell is sort of the whole point of this post. Like, they could say they're going to build that product, which is very different than other things out there. But that will seem risky, and what's likely to happen instead is that they slowly add easier charting, and some dashboards, and then they become a fairly standard BI tool.)
Agree, seems very reasonable. They write in their roadmap that they are going to add all the default Excel functionality (so, if they succeed, then, I trust, #1 will be addressed), but an overall premise of their tool is like "do not do Pandas analytics in Terminal, do it in our free to use tool". As you fairly mentioned, we'll see to which way they tend to go soon.
"everything is bi" is also a favourite trope of mine and a colleagues... So we made t-shirts! https://greyskull-analytics.myspreadshop.co.uk/everything+is+bi-A6377e7d613102c4d816e562e?productType=812&sellable=jwyGQ0n8kyIMeERONJV7-812-7&appearance=348
I need to find a "Kimball" tee. :)
Now I’m just imagining someone making shirts that say “BI is everything” and oh man that makes me want to die.
Great article! Resonate a lot with the startup I'm building!
"In other words...it will inevitably collapse into the same singularity: Database connectors; a code-based query editor; a code-free query builder; lots of charts; dashboards; self-serve."
And to be clear, "lots of charts" will include "Map of New Jersey" as part of this singularity.
That's what at the center of it all, really.
Visionaries are not guys who foresee different things, but those who stay different. Great articel, love the Mode's foundrasing story!
Thanks, glad you liked it!
This somewhat matches my recent sentiment around building a company, so much of that is about execution and less about the idea.
Awesome read Ben!
Thanks, I appreciate that!
I prefer the Kimball Group DW/BI term. The DW (dimensional data warehouse using conformed Dimensions and Facts in a star schema) is the core, and will help any BI tool, even if it only copies the star schema directly.
I know the Kimball stuff for data warehouses, but how does that get used to define BI too?
Kimball Group recognized the front end BI client was important, but almost of all of their work focused on the data warehouse (dimensional) and the ETL required to shape the data, and more importantly, the planning work with business to ensure success. I need to reread the Kimball Data Warehouse Lifecycle Toolkit book, which focuses on the entire process, instead of just modeling. https://www.kimballgroup.com/data-warehouse-business-intelligence-resources/books/data-warehouse-dw-lifecycle-toolkit/
Gotcha, yeah, that's the part of Kimball I'm more familiar with. Though in some ways, I think that semi-standardization makes it even easier for everything that sits on top to become BI, because it makes BI seem almost within reach. Like, if you have somewhat standard models, there's predictable patterns to it, and all of that, it's really easy to think "eh, we could just build a couple charts, right?" Whereas if the data underneath is a mess, I think people would hesitate a bit more to get involved with it.
Such an amazing read. Thanks for writing this. I have this bookmarked now!
I imagine the different thing is hard when the funding (and investors expectations) are the same.
Having money from previous ventures seems to be a helpful thing when wanting to do something truly different.
Kinda? VCs seem to have two minds about this stuff. They tend to like pitches that could be something different, but once they're invested, they become more risk averse and want to run the better understood playbooks. So I think raising money on a different idea isn't bad; staying different is tough. Having money doesn't make that easier (though it makes you seem more credible when you say you're gonna keep doing the uncomfortably different thing).
Ya - if having the money gives you more influence (and runway?) to stay different - I guess that’s a thing.
But lots of people are also different and worse. Big assumption that we are talking different and better.
Yeah, absolutely. And that's why it's hard - it usually doesn't work. To paraphrase an email someone sent me:
“If you want to be different, it takes will, not wits.” Also, if you're that stubborn, your company will probably die. But if you want to build something different, that’s the chance you’ve got to take - if you want to build something that probably won’t die, you’re gonna build BI.
Ya - makes sense. We also all have the bias of knowing the stories of “better and different” and not knowing many of the failure stories.
hey Ben, nice semi-ironic write-up, grazie) what do you think about Quadratics? They seem to offer similar to Mode use-cases and functionalities and they are open-source.
Do you have a link? This is like Census and is impossible to google (“census data” is some tough SEO)
I believe they are referencing https://www.quadratichq.com/
yep, indeed - https://www.quadratichq.com/
Word, so this was kind of cool. I like the core idea. It's actually somewhat similar to something I wrote about here. It's not quite the same thing, but the closest I've seen. https://benn.substack.com/p/the-next-billion-programmers
That said, it feels...confused? Like, is it a spreadsheet? A BI tool? An application for code-lite development? I think it's got a long way to go to be anywhere near competitive with spreadsheets, and even further to go to be a BI tool (if the only charting option is writing custom Python, you are completely dead on arrival as BI). So I think the interesting thing it could do is the third one, where it's more about trying to make basic scripting - which can be realllly powerful - much more accessible. That's a tough sell, but I like the general idea.
(And, the fact that it's a tough sell is sort of the whole point of this post. Like, they could say they're going to build that product, which is very different than other things out there. But that will seem risky, and what's likely to happen instead is that they slowly add easier charting, and some dashboards, and then they become a fairly standard BI tool.)
Agree, seems very reasonable. They write in their roadmap that they are going to add all the default Excel functionality (so, if they succeed, then, I trust, #1 will be addressed), but an overall premise of their tool is like "do not do Pandas analytics in Terminal, do it in our free to use tool". As you fairly mentioned, we'll see to which way they tend to go soon.
That makes sense, though also, "Excel functionality, coming in Q3" is...ambitious.