The Desktop App - with a backend that you speak of already exists.....it's called Dodeca by a small company known as Applied Olap. It works with relational databases as well as Oracle's Essbase. An awesome tool but one built around Excel. I do not work for Applied Olap, but deploy it at my corporation and our Finance team absolutely loves it.
Oh word. This seems like the right idea. But it seems a little tricky that it's so tied to Oracle (or requires a database at all). That's why I kind of like the DuckDB idea. With DuckDB, you could make something like this, and essentially have a database live inside of it too. So you could use it without any external sources, or connect it to one if you wanted to.
For us in the corporate world the big use case for Excel is to pull data out of databases and dump it into Excel to analyze. (Including Oracle apps and DBs) - So that is why Dodeca works so well - it does that for the user. When you get multiple users going after the same set of data (we support 400 users globally) this can take alot of time and effort. When we can set this up in Dodeca for them we reduce manual effort (time) and mistakes. Its been a huge win. At the end of the process the user gets what they want, the data they need in a format they want to look at it. (Excel) - One user described it as "Reporting Nirvana!!"
What's kinda funny about that is I've never heard anyone ever describe any of the fancy startup data tools as anything close to reporting nirvana. They're all "annoying but fine, I guess."
The problem with having database at all is that it's always somehow structured. And the clue of Excel is that data is unstructured but you can treat is as structured if you only want.
Yeah, very much agree with this, and think it's probably the biggest mismatch between excel and BI tools, honestly. So much of what happens in Excel is stylized tables and P&L type stuff that isn't a structured database table. And databases - and therefore, BI tools - really struggle with that.
I've always wondered how that balance is going. It seems like lots of people use Google sheets almost exclusively for personal use (budgeting and planning and stuff), and a lot mostly-Mac tech companies use it for work. Because those are the worlds I'm in, I personally don't actually see Excel all that much. But it also still seems like the world runs on it. Is that because every Fortune 500 company (and especially finance, etc) is in it? Or do I just live in a weird, Mac-using, cloud-zealot bubble and Excel still has 95% of the spreadsheet market share?
It usually depend on bundle contracts with either MS or Google. But even in a Google-bundled companies there is still Excel in use, at least limited to analysts.
Also migration between Excel and Sheets is in practice impossible as both solutions have very different logic of work.
Excel is desktop centric (online version is a meh). Sheets are cloud centric. I found that people build their solutions in very different way there. In Sheets it's often a bunch of interconnected sources, often with live data flow. Big focus on cooperation and group working. While Excel worksheets tend to be more complex (VBA, PowerPivot etc) but at the same time more isolated from each other.
Wild take to call Taylor Swift Gen Z pop music. xD Here's the question that haunts me: Out of the main pop girlies right now (Taylor Swift, Chappel Roan, Sabrina Carpenter, Charli XCX) who is most likely to be a closet Excel fangirl? I feel like Charli probably keeps some really bizarre spreadsheets like ranking raves she's played at, things like that.
ok, these are conversations i'm really here for (and not the, uh, entanglements about BI nonsense). My take:
- Taylor Swift: Has an army of people with Excel sheets behind her, reviews "state of the tour" decks every week, carefully triangulates everything. But she doesn't use Excel; that's what her minions are for. And they spend most of their time making sure the formatting is exactly how Taylor likes it.
- Chappell Roan: Uses Google sheets. Does very boring, normal stuff with it. Keeps a budget. One is shared with her friends, and it is a splitwise.
- Sabrina Carpenter: Doesn't own a computer. Not because she's a luddite, but because she wants to brand herself as a fun girlie who's here for a good time, and fun girlies who are here for a good time don't spend their time on computers.
- Charli XCX: Has a computer covered in kind of alarming stickers. Her keyboard is dirty, and a little sticky. She has 12 Excel files open, 1 is some rough attempt to do some sort of math, the other 11 are completely empty.
No part of me will ever believe that taylor swift doesn't do something like that. And being so rich and famous that you can successfully live without a computer really is the dream.
We're building version control for documents to solve many of the issues you outline here. We currently support Word files, but will support Excel in time.
Ah, nice. The wrapper around Word as opposed to a new doc editor entirely (assuming I read that right) seems like a useful way to do it, since as you said, Word is eternal.
This would be cool with Excel, though man, it seems hard. Like, you have to build all these complications to go from plaintext files (easy with git) to word files (plaintext but with fancy formatting, more or less). The amount of extra things you'd have to add to make that work with spreadsheets seems like...a lot.
I will say, in this job (https://benn.substack.com/p/no-better-room) I did almost all of my work in Google Sheets, and there was something relaxing about it, knowing the exact territory you had to work in.
I think you have it on the nose. I always find myself returning to Excel, because unless the data is so huge I need SQL to run calculations, nothing beats being right there in the weeds and knowing precisely what I am creating without anything blurring it
What do you think about products like Anaplan? Does an evolutionary Anaplan that figures out how to better support scratch padding and less structured analysis sound compelling to you?
That would, if it could really pull of the scratch pad part. My take on those tools is that they're obviously valuable for system-of-record type stuff (they're all basically CRMs), but they put so much abstraction between the data in the CRM and what you can do with it, people inevitably want to pull out the raw tables and numbers and do their own thing. But if you had anaplan with a sandbox that just gave you a bunch of bare metal numbers that you can play with, I could see that being pretty clever.
Excel needs branches IMO. I think that would solve a lot of the most immortal issues you call out here, like people overwriting numbers for their own scenarios and the dreaded "v1 v2 final FINAL" (unless those become the branch names)
Branches would help, though I can't even comprehend how you'd merge a branch back into Excel. On one hand, it'd be cool if someone made that actually work. On the other hand, my god, imagine an Excel merge conflict.
Very true, so maybe that's the real million dollar problem to solve. It sounds more like a UI/UX issue to me because I could imagine this working decently well with tools that support envs like duckdb and sqlMesh
Yeah, I suspect that would be the hardest part. But given how hard it is to deal with merges of just code, merging Excel notebooks seems almost impossible.
Dope, I'm into it. That's the holiday dream - a fireplace, a hot drink with whipped, some gentle Michael Buble in the background, and trying out the newest desktop based spreadsheet software applications.
So this kind of exists? But it's like, Python cells, which essentially just gives you more flexibility with Excel formulas. I want something that's more like a sheet tab that's Python, where you can:
- Hideously mash stuff together in Excel, like you normally do
- Take those results and do some python nonsense to it
- Put the results back in Excel, for more hideous mashups
Getting that experience right seems really hard. But it would be pretty cool.
Great post! I think Google Sheets has some of the cloud capability you talk about. by some sources, it has about a 20% market share today, and is growing. The precursors to excel were lotus 123 and visicalc. VisiCalc pretty much died when 123 came out, but 123 I understand kept being sold by IBM until 2013. So it takes a long time for these tools to die.
Yeah, it seems like if Excel really does go away, that'll be the boring version of the story. Microsoft fell behind Google in the cloud office suite world; when companies slowly move off of Outlook and desktop Word, they choose Gmail, docs, etc over O365; sheets is easier (and cheaper) to pick up in that world than Excel. Still, it seems like Sheets will be kind of capped if it's just a saas product that doesn't really work without an internet connection (offline mode exists, but you gotta remember to enable it). How will the consultants do their number crunching on airplanes??
It got a link in this sentence: "So far, most attempts to make a polyglot data tool stick notebooks in BI tools or stick spreadsheets in notebooks." Though, yeah, I would call it an enhancement to their notebooks by mimicking a few things about spreadsheets, and not an enhancement to Excel.
The Desktop App - with a backend that you speak of already exists.....it's called Dodeca by a small company known as Applied Olap. It works with relational databases as well as Oracle's Essbase. An awesome tool but one built around Excel. I do not work for Applied Olap, but deploy it at my corporation and our Finance team absolutely loves it.
Oh word. This seems like the right idea. But it seems a little tricky that it's so tied to Oracle (or requires a database at all). That's why I kind of like the DuckDB idea. With DuckDB, you could make something like this, and essentially have a database live inside of it too. So you could use it without any external sources, or connect it to one if you wanted to.
For us in the corporate world the big use case for Excel is to pull data out of databases and dump it into Excel to analyze. (Including Oracle apps and DBs) - So that is why Dodeca works so well - it does that for the user. When you get multiple users going after the same set of data (we support 400 users globally) this can take alot of time and effort. When we can set this up in Dodeca for them we reduce manual effort (time) and mistakes. Its been a huge win. At the end of the process the user gets what they want, the data they need in a format they want to look at it. (Excel) - One user described it as "Reporting Nirvana!!"
What's kinda funny about that is I've never heard anyone ever describe any of the fancy startup data tools as anything close to reporting nirvana. They're all "annoying but fine, I guess."
The problem with having database at all is that it's always somehow structured. And the clue of Excel is that data is unstructured but you can treat is as structured if you only want.
Yeah, very much agree with this, and think it's probably the biggest mismatch between excel and BI tools, honestly. So much of what happens in Excel is stylized tables and P&L type stuff that isn't a structured database table. And databases - and therefore, BI tools - really struggle with that.
As an ex-Googler who worked in Google Workspace sales, I can definitely affirm that Excel was always the biggest stone in our way.
I've always wondered how that balance is going. It seems like lots of people use Google sheets almost exclusively for personal use (budgeting and planning and stuff), and a lot mostly-Mac tech companies use it for work. Because those are the worlds I'm in, I personally don't actually see Excel all that much. But it also still seems like the world runs on it. Is that because every Fortune 500 company (and especially finance, etc) is in it? Or do I just live in a weird, Mac-using, cloud-zealot bubble and Excel still has 95% of the spreadsheet market share?
It usually depend on bundle contracts with either MS or Google. But even in a Google-bundled companies there is still Excel in use, at least limited to analysts.
Also migration between Excel and Sheets is in practice impossible as both solutions have very different logic of work.
What do you mean by different logics of work?
Excel is desktop centric (online version is a meh). Sheets are cloud centric. I found that people build their solutions in very different way there. In Sheets it's often a bunch of interconnected sources, often with live data flow. Big focus on cooperation and group working. While Excel worksheets tend to be more complex (VBA, PowerPivot etc) but at the same time more isolated from each other.
Wild take to call Taylor Swift Gen Z pop music. xD Here's the question that haunts me: Out of the main pop girlies right now (Taylor Swift, Chappel Roan, Sabrina Carpenter, Charli XCX) who is most likely to be a closet Excel fangirl? I feel like Charli probably keeps some really bizarre spreadsheets like ranking raves she's played at, things like that.
ok, these are conversations i'm really here for (and not the, uh, entanglements about BI nonsense). My take:
- Taylor Swift: Has an army of people with Excel sheets behind her, reviews "state of the tour" decks every week, carefully triangulates everything. But she doesn't use Excel; that's what her minions are for. And they spend most of their time making sure the formatting is exactly how Taylor likes it.
- Chappell Roan: Uses Google sheets. Does very boring, normal stuff with it. Keeps a budget. One is shared with her friends, and it is a splitwise.
- Sabrina Carpenter: Doesn't own a computer. Not because she's a luddite, but because she wants to brand herself as a fun girlie who's here for a good time, and fun girlies who are here for a good time don't spend their time on computers.
- Charli XCX: Has a computer covered in kind of alarming stickers. Her keyboard is dirty, and a little sticky. She has 12 Excel files open, 1 is some rough attempt to do some sort of math, the other 11 are completely empty.
-Taylor Swift running a WTR (weekly tour review) is an incredible mental image.
-Chappell understands that no stupid ass SaaS budgeting tool will top a GSheet budget
-Sabrina is real for this and we should all get rid of our computers
-I don't know what alarming stickers Charli has but she is brat for ALL of them xD
No part of me will ever believe that taylor swift doesn't do something like that. And being so rich and famous that you can successfully live without a computer really is the dream.
Great essay! I make several similar points about Microsoft Word's immortality in a piece I recently wrote regarding building Git for lawyers https://jordanbryan.substack.com/p/on-building-git-for-lawyers.
We're building version control for documents to solve many of the issues you outline here. We currently support Word files, but will support Excel in time.
Ah, nice. The wrapper around Word as opposed to a new doc editor entirely (assuming I read that right) seems like a useful way to do it, since as you said, Word is eternal.
This would be cool with Excel, though man, it seems hard. Like, you have to build all these complications to go from plaintext files (easy with git) to word files (plaintext but with fancy formatting, more or less). The amount of extra things you'd have to add to make that work with spreadsheets seems like...a lot.
Once an analyst realizes that all roads lead to Excel, they find peace within themselves 🧘
I will say, in this job (https://benn.substack.com/p/no-better-room) I did almost all of my work in Google Sheets, and there was something relaxing about it, knowing the exact territory you had to work in.
I think you have it on the nose. I always find myself returning to Excel, because unless the data is so huge I need SQL to run calculations, nothing beats being right there in the weeds and knowing precisely what I am creating without anything blurring it
What do you think about products like Anaplan? Does an evolutionary Anaplan that figures out how to better support scratch padding and less structured analysis sound compelling to you?
That would, if it could really pull of the scratch pad part. My take on those tools is that they're obviously valuable for system-of-record type stuff (they're all basically CRMs), but they put so much abstraction between the data in the CRM and what you can do with it, people inevitably want to pull out the raw tables and numbers and do their own thing. But if you had anaplan with a sandbox that just gave you a bunch of bare metal numbers that you can play with, I could see that being pretty clever.
Excel needs branches IMO. I think that would solve a lot of the most immortal issues you call out here, like people overwriting numbers for their own scenarios and the dreaded "v1 v2 final FINAL" (unless those become the branch names)
Branches would help, though I can't even comprehend how you'd merge a branch back into Excel. On one hand, it'd be cool if someone made that actually work. On the other hand, my god, imagine an Excel merge conflict.
Very true, so maybe that's the real million dollar problem to solve. It sounds more like a UI/UX issue to me because I could imagine this working decently well with tools that support envs like duckdb and sqlMesh
Yeah, I suspect that would be the hardest part. But given how hard it is to deal with merges of just code, merging Excel notebooks seems almost impossible.
I'm building this! We support Word files for now but will support Excel in due time.
https://jordanbryan.substack.com/p/on-building-git-for-lawyers
I will send you a demo till the end of the year of a system that enables what you envision:
- opens locally
- optional remote compute to offload large data
in case you don't receive an email from by mid december, write me samuel (dot) stroschein (at) opral.com
Dope, I'm into it. That's the holiday dream - a fireplace, a hot drink with whipped, some gentle Michael Buble in the background, and trying out the newest desktop based spreadsheet software applications.
email sent ✅
I'm just here for the Decemberists (and Eschaton) references.
Good, because let's be honest, that's why I write these things.
Indeed! ✊🏻
Native Python in Excel will unlock God mode. Excel, possibly the best application ever created.
So this kind of exists? But it's like, Python cells, which essentially just gives you more flexibility with Excel formulas. I want something that's more like a sheet tab that's Python, where you can:
- Hideously mash stuff together in Excel, like you normally do
- Take those results and do some python nonsense to it
- Put the results back in Excel, for more hideous mashups
Getting that experience right seems really hard. But it would be pretty cool.
https://support.microsoft.com/en-us/office/get-started-with-python-in-excel-a33fbcbe-065b-41d3-82cf-23d05397f53d
Whay you've described is PowerPivot but with python.
Or even kinda just macros, honestly. Which are really popular, despite being whatever VBA is.
Great post! I think Google Sheets has some of the cloud capability you talk about. by some sources, it has about a 20% market share today, and is growing. The precursors to excel were lotus 123 and visicalc. VisiCalc pretty much died when 123 came out, but 123 I understand kept being sold by IBM until 2013. So it takes a long time for these tools to die.
Yeah, it seems like if Excel really does go away, that'll be the boring version of the story. Microsoft fell behind Google in the cloud office suite world; when companies slowly move off of Outlook and desktop Word, they choose Gmail, docs, etc over O365; sheets is easier (and cheaper) to pick up in that world than Excel. Still, it seems like Sheets will be kind of capped if it's just a saas product that doesn't really work without an internet connection (offline mode exists, but you gotta remember to enable it). How will the consultants do their number crunching on airplanes??
I'm shocked he didn't mention HEX and their enhancements to Excel.
It got a link in this sentence: "So far, most attempts to make a polyglot data tool stick notebooks in BI tools or stick spreadsheets in notebooks." Though, yeah, I would call it an enhancement to their notebooks by mimicking a few things about spreadsheets, and not an enhancement to Excel.