Taimur Abdaal - Turning Data into Action: The Missing Link in Forecasting

Taimur Abdaal - Turning Data into Action: The Missing Link in Forecasting

Episode 025: Where we learn from Taimur Abdaal, CEO and co-founder of Causal, a modeling platform that automates everything from financial models to headcount planning.

Listen in as Taimur shares his journey from being a data scientist to co-founding Causal in 2019, and how his experiences with spreadsheet challenges inspired him to create a better modeling tool.

Discover how finance teams can leverage technology to save time and make better decisions. Taimur shares his insights on when it's best to consider moving away from spreadsheets and towards a dedicated tool.

We also discuss the current generation of AI models and their potential to look at wider business contexts when it comes to forecasting. Learn about the importance of decision-making in the tech world, and how Taimur's mental model for decision-making ties into forecasting and decisions.

Additionally, we discuss how staff satisfaction can contribute to employee retention and how the younger generation of employees expects to work on more exciting tasks instead of manual Excel work.

Tune in to learn how to make the most of your time and achieve high-impact results in both your personal and professional lives.

Audio Podcast Links

Show Notes

Where to find Taimur:

Tools Mentioned

Other Mentions


(0:00:00) - Modeling Tools and Data Storytelling

(0:13:46) - AI and Machine Learning in Forecasting

(0:25:35) - Twitter, Decision Making, and Personal Blogging

(0:30:56) - Frequency in Decision Making and Development

(0:35:42) - Investing Time for High-Impact Results

(0:40:28) - Maximizing Time and Productivity With Tools

Chapter Summaries:

(0:00:00) - Modeling Tools and Data Storytelling (14 Minutes)

Taimur Abdaal, CEO and co-founder of Causal, chats about how finance teams can leverage technology to level up their lives. Taimur shares his journey which included working as a data scientist and co-founding Causal in 2019, which has now raised 25 million from investors. We then discuss the threshold for when finance teams should consider moving away from spreadsheets and towards a dedicated tool. Taimur explains that if they need to build something they need to reuse on a recurring basis, or if they need to involve machine learning or AI forecasting, that's when it's best to consider a dedicated tool.

(0:13:46) - AI and Machine Learning in Forecasting (12 Minutes)

We discuss how important seasonality is and the external factors that influence forecasting. We explore how the current generation of AI models has the potential to look at wider business context when it comes to forecasting, and the advantages of using LLMs for forecasting. We chat about the evolution of finance tools that are training models better, and the opportunities offered by LLM for making manual tasks easier.

(0:25:35) - Twitter, Decision Making, and Personal Blogging (5 Minutes)

I chat with Taimur Abdaal about the importance of decision making in the tech world. We explore how Twitter has changed over the years and the rise of the tech Twitter community. We discuss the probability of LNM interfaces playing sequel queries in the next two years and a personalized number plate on a Mazda 6. We also cover Taimur's blog post titled 'Measure: A Mental Model for Decision Making' and how it relates to forecasting and decisions. Finally, we discuss books Taimur has bought in 2019 but not read yet.

(0:30:56) - Frequency in Decision Making and Development (5 Minutes)

I chat with Taimur Abdaal about the concept of magnitude versus measure when making decisions. We explore how to think about the value of something, and how frequently that value is obtained over a period of time. We also discuss how to think about investments in things like kitchen objects, health, and the development of products. Finally, we look at how Twitter has changed over the years and the rise of AI models.

(0:35:42) - Investing Time for High-Impact Results (5 Minutes)

We explore how to prioritize time-based decisions to maximize high-impact tasks for the company and on a personal level. We discuss the importance of employee onboarding and why HR processes can often be skipped due to being busy. We look at the 80/20 principle and ask questions about what investments are generating the most results both in the short and long term. Lastly, we look at how ROI can be used to free up time and how to make sure we are doing the highest impact thing each week.

(0:40:28) - Maximizing Time and Productivity With Tools (10 Minutes)

We explore how teams can free up time and evaluate ROI from modelling tools. Time saving is often the most tangible ROI and can qualitatively change the way teams work with the rest of the business. Staff satisfaction can contribute to employee retention and the younger generation of employees expect to get on with the 'fun stuff' instead of manual Excel work.


Transcript generated by Podium.page

0:00:00 - Taimur

It turned out that most finance teams didn't really end up using this style of machine learning or AI forecasting because it doesn't really give you anything actionable. It might tell you, hey, revenue is going to be some certain number next quarter or next month, or sales might be this, but it doesn't actually tell you, ok, what do you need to do in the business in order to get that?

0:00:24 - Adam

Hello, and welcome to Tech for Finance where we help finance professionals leverage technology to level up their lives.

I’m your host, Adam Shilton, and in this episode we’re going to be chatting with Taimur Abdaal, CEO & Co-Founder at Causal, a modelling platform that helps you automate everything from financial model to headcount planning.

Taimur has a degree in Mathematics and Statistics from Oxford, and after a short period working as a Data Scientist, he co-founded Causal in 2019 which has now raised $25M from investors like Accel, Coatue, Passion Capital, Naval Ravikant, and others…

In his spare time, Taimur enjoys playing the piano, playing badminton, and exploring the London food scene.

If you like what you hear today, make sure to subscribe to this podcast on your favourite platform, or on YouTube.

It's great to have you on. Yeah, great to be here. Thanks so much. Not a problem.

So we'll kick things off and I'll ask you the question that everybody's probably asked you already. So, data science modeling. You just want to very quickly talk us through what that transition looked like and how you've ended up where you are today.

0:01:25 - Taimur

Yeah, absolutely So. I think it kind of started when I was a teenager. I was pretty into computers and just discovering the internet in general, Started off more on the design side of things, so messing around in Photoshop and making graphics and graphic design and logos and things like that And then eventually learning some basic coding like web stuff, like HTML and JavaScript and that side of things. And I think from a pretty young age I always wanted to run my own business, start my own company, And so I was always trying to think of ideas and building projects and apps and websites And so, when I was at university, continued doing that stuff. That's where I met Lucas, my co-founder. We worked on lots of site projects and we'd always planned on starting a company together. We didn't have any sort of particular ideas We were really interested in by the time we graduated, and so we both ended up working for a year or two.

As you mentioned, I was a data scientist, had a couple of different startups, And that was really my first exposure to the kinds of stuff that companies do in spreadsheets, And so I was working at this company Nested. It was a startup in London. There were about 80 people at the time They had a finance team and obviously, like most finance teams, they were doing a lot of stuff in Excel and in Google Sheets And they had a lot of the sort of typical challenges that people have with spreadsheets. Right, You know, spreadsheets are amazing and that they're so flexible, you can do pretty much anything in them. But there's a lot of tricky pieces from getting data in from different systems to the actual formulas and errors and syntax and that kind of stuff, To then kind of presenting stuff with people communicating, sharing, And so you know, from seeing some of those problems that the finance team at Nested had with their spreadsheet financial models, that's kind of what got me and Lucas, my co-founder, thinking about hey, what would a better modeling tool look like?

Why hasn't someone you know tried to build a better modeling tool in a spreadsheet? Yeah, spreadsheet's been around a long time, And so that's what we're really trying to do with Cozol Trying to build a very general tool for modeling, you know, calculations, data visualization, things like that A bit like a spreadsheet in that sense. But we're not interested in a lot of the other things you might use a spreadsheet for, For example, managing some business process or tracking your grocery lists and all the other kind of stuff you might do.

0:03:45 - Adam

Very good, thanks for that. And what do you say? the threshold is then So, and I wholeheartedly agree. You know finance, especially. You know we love spreadsheets, right, yeah, it's the go-to. Somebody asked you something, right? Well, i need to do my calculations. I'll open up the spreadsheet, right, and we get to the point now where and again, it's not in the same league as what you guys can do but spreadsheet add-ons, i think there's even a forecast button, excel now as well. So I guess, if we were looking at this in terms of levels, you know, entry level is, you know, i guess, a basic model. It doesn't really have many variables in it right, maybe build on spreadsheet, then maybe next level is, you know, I've got a plug-in that's going to enable me to do X1 Edge. In terms of your thinking, what do you think is that sort of crossover between sort of entry level and the point where they might have more of a dedicated tool, if that makes sense, yeah, yeah.

0:04:45 - Taimur

Yeah, i think that that's a really good way to think about it. I think a couple of the sort of points at which we typically see people considering moving away from spreadsheets, or where I think it's worth considering, is, firstly, if you need to build something that you need to reuse on a recurring basis, so you know, you might need to whip together a quick ad hoc model for you know some business thing that you're trying to do. You might just need that model tomorrow and then you're not going to look at it again. If you're pretty adept at Excel, google Sheets, you'll find to do that in Google Sheets or Excel. It'll take you a couple of hours, it'll be done. You'll move on.

If, however, you need to build something that you're going to be coming back to on a regular basis, maybe every month you want to update some numbers from your data source, or maybe once a quarter you want to run scenarios, or maybe you want to share outputs every single week as numbers are updating. I think that's when the fact that spreadsheets are so manual it really starts to add up And you know that sort of one or two days a month of pulling data from different places, you know, really starts to become a time sink, and that's where it's worth considering Okay, are there tools that can automate some of this manual stuff so that, on a recurring basis, i'm not having to spend a bunch of time maintaining this spreadsheet? Does that make sense?

0:06:02 - Adam

Yeah, and I guess there's two ways of looking at it, isn't there? So apologies to kids keeping you up all night and you're voice and, oh good, i'm so sorry about that. There's two ways of looking at it. So, obviously, the point that you addressed is the time saving piece, right, you know, i don't really want to have to build from scratch every time I'm working through a new scenario. We'll have to do that, i guess, as the other side and it's kind of linked to some of the stuff that you've been posting about recently when we talk about data storytelling, right, because I think we're now moving to the point where decisions need to be made quicker And I don't know whether you're saying this is as well, but some of the previous conversations we've had, you know some of the advantages of having stuff that's a little bit more live actually is quite useful.

So, instead of going away and thinking, right, well, you know, let's have a thing that's, you know, go to a few more spreadsheets or go through a few models, that could take us days or weeks, you know, in the room. So I was wondering whether you could speak a little bit more around what you consider to be, i guess, good data storytelling And the second part of that is for me, and I don't know whether I'm speaking for others. I really struggle not so much with the story piece and the presentation piece but on the what's the focus, because it's sometimes difficult And of course you've got background in mathematics and statistics and this sort of stuff. You know how do you find the data that tells the story. You know how do you prioritize what is good data versus bad data. Speak around that a bit.

0:07:35 - Taimur

Yeah, i think, yeah, data storytelling is kind of super interesting And I guess it's not really positioned that way, but it's a lot of what finance teams are doing. It's telling stories with the different data from around the business, whether that's financial data or more operational data. I think some of the I think probably the most important thing that we see when working with our customers and even internally when trying to make decisions using our own data, is just having really deep domain expertise in whatever it is that you're looking at, if you're on the finance team of a company just really understanding the company and a lot of detail of exactly how it works, what the different teams are doing, how data is being collected into different systems. Is it being manually filled in somewhere and then dumped into something? is this coming from the CRM?

I think the thing that's indispensable when working with data and then having to present it or tell stories off the back of it is actually understanding that data really deeply. I think once you have that understanding, you then know, oh okay, that tool or that system might be a good place to start going to see. What do we have in there? what does that data look like? Or you might know that, oh, actually, that stuff is not very reliable right now. We had some issue with that, like with the CRM last month, so I can't actually go there for that data, and so I think domain knowledge very, very hard to overstate just how important it is to understand exactly how the business works, understand how that data was collected, before you even get to the point of then starting to explore it.

0:09:17 - Adam

And I think as well, it's also known what questions to ask. Right, because, as much as finance, they get kind of the honor reposition of being responsible for all company data, right, right, yeah, so other people do have different responsibilities and it's not purely dance finance to manage the data. Right, so, asking good questions and I'm seeing a bit of a trend with this at the moment, because, again, we can speak about it a bit later if you like but we say AI is the hot topic right now. Right, Yeah, of course, yeah, and people are getting a bit panicky and a bit overwhelmed. I personally was very overwhelmed earlier on this, trying to make sense of everything as though it was basically the word, just because of the pace of it. Right, but people are now thinking Rob, what skills do I build? You know, if there's not a lot of point in me developing those low level hard skills anymore, where am I? So we've seen more of a shift to business partnering, whereby finance are kind of breaking out of finance and that doesn't necessarily mean that like it works somewhere else or work through a different department. It just means that they become better partners.

I guess Tainty point there about finding the data, making sure that you've got that domain expertise. The way that you're going to get that is by asking questions from your peers. Yeah, yeah, absolutely. Some businesses do it well, some people don't. But if you're not doing it, then my recommendation is just go out and wait to ask. Yeah, you agree with that on it.

0:10:44 - Taimur

Yeah, no, absolutely I think. Yeah, anytime you are visualizing data or, yeah, anytime you're working with data, you're kind of looking at it through the lens of whatever questions you're asking, and so getting those questions right is really important, and I think that's where finance can actually add a lot of value, where I think the other team of sales teams, marketing teams, the other teams that finance works with, they're often pretty stuck in the weeds. They're kind of very deep into what they're doing. They might not have context around what's going on in the rest of the business, and so finance has that kind of full picture of all the different teams in the business and they can take a bit more of a step back than probably the people who are doing those jobs day to day, and by asking those right questions they can kind of guide each team to kind of making decisions based on all of that extra context.

0:11:35 - Adam

And then, i guess, expanding that a little bit. So we've got data, whether it be from an HR system, a CRM system and finance, so we can combine that all together right, whether it's in very well presented spreadsheets or a dedicated tool. We then need to start thinking about well, if we are doing modeling and we are doing that forecasting and looking at that future vision, how are we going to go about that? So there's a couple of things and you might tell me off because you've probably got more experience in this than I do But there's a couple of things that I've been playing around with personally, because I try and provide insight where I can, at sort of the lower level to begin with, sort of finding their feet. So, google Sheets, plugins there's a load available. The guy that TensorFlow did a plugin called Simple ML, for sure, and basically it's dead simple. You've got your date range in one column and then you've got your values against those date ranges. So when the dates continue, and then you click a button and say predict what the dates are going to be Simple, and you've got some basic machine learning algorithms in there There's maybe a couple And then you can apply seasonality if it works.

Doesn't always work. Yeah, yeah. So that's one example, and then another one, and we'll get into this conversation in a second, because it's something else that I really want to ask you about is the likes of Notable, which is essentially kind of a user friendly Jupyter notebook that will help you build visualizations from your data connected to chat, tpt right now. So here's data and again being a spreadsheet that you upload and say, look, can you give me some visualizations? But you can also tell it to build in those additional machine learning parameters, to say it's bothered about analyzing the data as it is, can you give me at least a basic representation of what that looks like over time? That's you.

Sorry, it's a long way, isn't it? Assuming we've got good data, be it from financial, different departments? What's the next level? So is it just a case of being able to say, right, well, i'm going to overlay machine learning algorithms. How important is seasonality? Are there any other external factors? whether it's? I mean, we've had examples of weather data previously, that sort of stuff. So what are you seeing there? What are you seeing in terms of variables and the things that make forecasts better?

0:14:00 - Taimur

Yeah, Yeah, i think the sort of machine learning and AI approaches to forecasting are pretty interesting because I think so far I think until now, until sort of LLMs and the current generation of I guess we're calling AI now, whereas previously we called it machine learning because it was at a different level I think that the previous generation of machine learning models that people used they were kind of regression models, where you have this kind of structured data sets. The machine learning algorithm figures out some patterns. It can maybe incorporate other pieces of things that you include in that data set, like seasonality and so on, and then it spits out a number. I think it turned out that most finance teams didn't really end up using this style of kind of machine learning or AI forecasting because it doesn't really tell you, it doesn't really give you anything actionable. It might tell you, hey, revenue is going to be some certain number next quarter or next month, or sales might be this, but it doesn't actually tell you okay, what do you need to do in the business in order to get there? It wouldn't be able to tell you, hey, revenue is going to be this much because you need to go and hire like three more salespeople and you need to spend X amount on marketing across these channels.

I think the old generation of machine learning algorithms, they were very sort of narrow in terms of the data that they can look at and interpret, i think.

And so, yeah, basically we don't really see many finance teams using them for forecasting.

Forecasts are kind of much more you call it manual, but kind of built by the finance team based on understanding of how the business operates And so to get to some revenue, that happens because something changes in the business, whether it's hiring, whether it's sales and marketing spend, et cetera.

I think we haven't really seen it's very early to see what will come out of LLMs, but I think the new generation of AI models does actually have the sort of opportunity to be able to go have context on the much wider business where you can kind of feed in maybe like a whole 20 tab spreadsheet that includes hiring plans, it includes models for each team, et cetera, and so it could actually in theory be able to tell you hey, if you hire these people and if you do X, y and Z, then revenue will be this amount and that's one strategy or it might come up with another strategy. So I think we haven't really seen what that looks like yet. I think probably in the next couple of years we'll start to see this kind of stuff built into spreadsheets, tools like Causal and other planning tools, and so I think I'm pretty bullish on AI now being applied to forecasting and modeling, where the previous generation of algorithms just kind of wasn't the right fit for it.

0:16:44 - Adam

And the trick is in the way that you train the model though. So, again as a refresher for people that haven't been part conversations before, llm is a large language model. It's what ChatDPT is built on vast amount of data. But when you interact with ChatDPT on a basic level, openai have trained that large language model to interact well with you through a chat interface. Yeah, that's the way it works.

What we're now seeing is the evolution of finance tools that are training models better. So you ask them a question, it's more likely to come back with a finance-related response than it is just a generic response. So we've had this whole sort of proliferation of advanced prompts and prompt recommendations and all that sort of going around the internet. So down to the fact that if you ask a rubbish prompt to ChatDPT out of the box, it's going to give any number of variables out of it. Yeah, i think there are already companies that are creating, like FinChat and all of these sorts of narrower applications of that large language model. But I think you're right, it's going to be very interesting from a forecasting perspective where they get to, because it's not just about how do we train the large language model. So it is good at creating forecasting recommendations. It's also how do we train it. So it's good at forecasting, but then also good at understanding the way it's built on. Yeah, so there's a lot of work. It's exciting, of course. It's exciting But tricky. I don't envy your job.

0:18:17 - Taimur

Yeah, we're definitely.

We started thinking about how we might be able to embed some LLM magic into causal and kind of what layer of the product we want to try and insert there.

And I think there's a lot of kind of low hanging fruits of just making very manual stuff very easy. So things like pricing, sql queries or, you know, if you have some big model, just being able to ask a question like you know what was revenue in this month for this product, you know, filtered by whatever that might otherwise require like a bunch of different clicks or it might require knowing SQL and things like that. So I think, and the low hanging fruits will kind of become a commodity where I think in a few years time you'll kind of expect in most products that you won't have to do manual things like write SQL queries and you won't have to do a ton of configuration to make a chart. You can just describe the chart and it'll create a base on the data. Yeah, i think I think there's more kind of ambitious, you know, reasoning and forecasting based on like actual reasoning and actual logic. I haven't seen what that'll look like yet, but it's definitely something we're thinking about.

0:19:20 - Adam

Yeah, because it kind of ties to the external dates piece, doesn't it? Because you know there are some modeling tools on it And again, i'm not a causal pro, so I don't know whether you guys are doing it yet You know, has the ability to add a variable that says you know what, if you know there was a flood in six months time, right Yeah, recession, or you know another pandemic, or something like that.

You know, data that sits outside just called a two. Is that that, as I say, contextual data there, and I think it is going to be a while. And again, i'm happy to be proven wrong, because if technology has said anything in the past six months has been pretty in a lot of people right, but yeah, they're very interesting. Chat with the guy called Chris rightly if you put podcasts ago, and he does a lot around advising in terms of financial modeling and and we had a similar conversation and I posed to him right, well, you know what does? what does a human based activity look like versus a machine based activity? Words were essentially you know he doesn't think that huge that machines are going to replace human gut feel And you know that that experience anytime soon.

Yeah, because, yeah, in theory, like when you stop the stock markets and financial and all that sorts of, then there's a theory to say you know, you build an algorithm that's going to predict trends and all that sort of stuff. Again, not that I suppose in theory there's. There's an argument that says the model could be trained on any number of, you know, weather variables or any external, but I think in the short term it's still going to require a human to say, right, well, we've done the best that we absolutely can with the technology in the AI. Yeah, we now need to make the decision of what we actually think is going to happen, right.

0:20:54 - Taimur

Yeah, yeah, yeah, i think. ultimately, i think the thing that is missing with AI is just any kind of accountability. Right, you know, if you're, if you're the finance team, it won't be good enough to say, well, we did that because the AI sets right, like, ultimately, you are accountable And so, yeah, i think that's going to be the challenge and AI always has to be the tool there, but a human is ultimately the person who is making the final call, making the decision, and is accountable for whatever happens after that.

0:21:22 - Adam

Yeah, i have been using AI to run some scenarios though, oh yeah, and it's the prompt's too long and share it with you the other day. But it's kind of decision fatigue is a real thing, right, you know, everybody's got too many decisions to make. you know, at the end of the day, when you absolutely exhaust and you think, am I really going to have to decide on something else, you can use AI as a sounding board, you know. you can use people as well, right, you know. but but there's an argument that says that AI might be more objective than they may be. There's a bias that I don't. we don't need to get into that discussion.

But I just ran through. I said, look, you know, here's the current staff. you know it's their skill set, here's their responsibilities. Yeah, we're thinking about doing this. You know we're thinking about hiring people with these skills. Yeah, you know, these are the people that are leaving. These are the challenges that we've got. And it was like an A4 page prompt And I looked at this and I thought I'm going to get garbage out of the back.

0:22:21 - Taimur

Yeah, there's no way it's going to sound the same It digs, it, digs.

0:22:24 - Adam

It's obviously the T4, right, you know, the most intelligent that we've got so far. Right, and it actually came back with some pretty credible scenarios Because I said to it give me five scenarios. It wasn't just a blank kit What would you do in this situation? That's not going to be useful, right? But based on the information above, give me five different scenarios for how you would look to hire and solve these challenges. Okay, and it gave five and we ended up going with one of them.

Yeah, it was kind of the decision that we'd already made. You see what I mean. So that human element was still there, but it was good to have that validation to say, look, you know, you've got five different scenarios here based on logic rather than emotion, because that's what it's doing, right, it's working on emotion. So, to have a logic based sounding board as a, you know, because, as you know, you know you'll have some, some people in your business that are endlessly enthusiastic and positive. Oh yeah, you should do that, Don't worry about it again, it will be fine, right? And you've got the polar opposite, which are the people saying don't do that, it's too high risk. You know, we're not interested in that, whereas an AI is going to try and do balance, for it Can Yeah.

0:23:34 - Taimur

That's awesome, that's. yeah, that's super cool, it's it's. it's cool that that's. that's actually useful in that way. Yeah, i think what it, what has been useful for me for, is just kind of you know, like you said, you kind of already had a sense of, hey, we're thinking of going with this decision.

I think, before making decisions, the thing that always runs through my mind is like, okay, this seems pretty good, but are we missing anything? Are we? have we just like missed something really big And we're about to like go in this one direction? And so if you can just ask the AI like Hey, you know, this is the situation. like what are the, what are like the 10 things we need to be aware of, if you look at that list, you're like, yeah, we've actually thought about all of those already. then you at least have some peace of mind of like okay, we're not missing anything major, like let's go for this more confidently, whereas otherwise, yeah, i think, i think that's what always might worry, particularly with decisions that are hard to reverse. So like, okay, how we can we consider, like the whole space of options here, you know?

0:24:23 - Adam

yeah, yeah, no, that's that's it. You know, i mean, you run a successful business. So, yeah, you're, you're level of stress and you know pressure when it comes to those decisions must be a lot bigger than mine. But yeah, no, i totally understand what you're saying. Yeah, i do totally appreciate that. Cool, right. Well, this kind of leads me on to the next point. You've touched on the, the, the last language model model replacing SQL queries, right, and I think it was something that you posted about recently. right, and we've just been through that. But if I've understood you correctly from what you've just said, we're essentially saying that we're cutting out that manual piece, right?

I think, so, yeah, i think that's something I would like to talk about with it instead. Yeah, Yeah. Okay, cool. So Let's move on a little bit to something that's not necessarily directly financial related, but kind of piqued my interest. So I've got to talk about your Twitter. Do you got like what is it coming out to? 22,000 followers or something like that?

0:25:22 - Taimur

Yeah, i don't really keep count, but yeah, so we're like, oh, i don't keep it, yeah, which is good.

0:25:29 - Adam

Is that just in the time that you've been running Crosal? Is that since uni?

0:25:32 - Taimur

Yeah, I mean I've had Twitter since, like secondary school. I remember I think I first signed up for Twitter back when the cool thing about Twitter was that like a bunch of celebrities were on there And like you could like talk to celebrities or something. And so I had this TV show I really liked with Modern Family, and I was like, oh cool, I can like follow the Modern Family actors on Twitter, amazing. And so you know I got it when I was like 16 or 17. I didn't really tweet anything, i didn't use it much. I think I properly started using it probably maybe like five years ago now, maybe like a year before we started the company.

Yeah, i think just for started off with just kind of tech related stuff. And then I think the kind of tech Twitter community has just kind of built up over the last sort of five, six years quite a lot And yeah, it's just like it's a great place to hang out if you work in tech. I will say that I think it used to be a lot more fun. I don't know if it's like algorithm changes since Elon took over or whatever, but I do remember Twitter being like a lot more funny, a lot more entertaining, because now there's just like a bunch of people arguing on my feet all the time And, yeah, i think it feels like it's going downhill a little bit, but I'm still a bit kind of twisted.

0:26:41 - Adam

It's great And from everybody that's listening needs to follow Taimur on Twitter. It's really good stuff. You got really good balance between the entertaining stuff and the more I kind of I won't say boring, but the more you know, the more sort of business type stuff, and that's why I saw that. So the question was you know what's the probability that LNM interfaces will play sequel queries in the next two years? And you've got like a 30% run response on less than 25% or something like that 92 votes. So I think it's the way it's going, but we'll have to see. Right, who knows? Yeah, we'll see in the next couple of years. Also on your Twitter, is that a Mazda 6 you've got with a personalized number plate on it?

0:27:25 - Taimur

Yeah, someone send me that picture. It's not my own car. That would be pretty good. Someone send me that picture. I was like, okay, that's pretty funny, i'll send that as a. Yeah, the number plate says causal. Yeah, i thought that would be a good comment.

0:27:38 - Adam

I have to tag life goals. I was expecting it. My siren will send me like that. So when I saw the back, i was a bit taken aback. Yeah, one day, one day. The one that I'm looking at was the. This is what the perfect and in in inverted commas man looks like according to AI, and I'll put the link in the show next to people people that want to check it out. But that is hilarious. I appreciate that, yeah.

0:28:02 - Taimur

Yeah, something like like it didn't do as well as I was expecting. I thought, yeah, sam was actually in London a couple of weeks ago. I think he's doing like a world tour at the moment, so he was doing some tour at UCL And so I went to see him in person, which was quite interesting. And, yeah, i think a few people. There was some posts around like you see, all these kind of like kind of crappy articles from weird publications about AI is predicting the as a, and there was a similar thing around like oh, this is what the perfect man looks like. I had so I thought you know, much funny to do a parody of that.

0:28:34 - Adam

Yeah, Spoiler spoiler. It's probably not what the I mean he's okay, but Yeah, no offense to Sam.

Yeah, yeah, no, no, it's all good. So no cool. So the last bit I wanted to talk about, I guess, is um, it ties ties back to that decision making piece. And then I'll ask you the last question that I always ask on these, these podcasts, about your favorite sort of tech. But I went on to see your personal blog and, again, i don't know how long ago it is that you've written on there Um, have you stopped writing on?

0:29:11 - Taimur

there now. So I think, yeah, i think since starting the company I just haven't yeah, haven't really been making the times right regularly, but yeah, for maybe a year or two I was doing it like once a week or a week or two weeks or something.

0:29:22 - Adam

Cause there's like a graveyard of the top of it says the books that I bought in 2019 that didn't read Yeah yeah, yeah, Yeah.

0:29:29 - Taimur

I think a lot of people post like lists of like oh, this is what I did. Reason I find that I buy a lot of books and then don't end up reading most of them.

0:29:37 - Adam

So on, there you've got atomic. Did you actually end up reading the atomic?

0:29:40 - Taimur

habits. I think I've skimmed. atomic habits, yeah, yeah.

0:29:45 - Adam

Cause everybody says, like, if there's going to be one sort of like productivity book that we're going to recommend.

0:29:49 - Taimur

Yeah, i think that is a couple of people.

0:29:53 - Adam

I'm gonna say I've not read it myself, but I've got the summaries from many guys.

0:29:58 - Taimur

Yeah, yeah, that's what you need.

0:30:01 - Adam

So and this does relate to our conversation because, you know, when we talk about modeling, we talk about forecasting that sort of stuff it does. It does relate to decisions And I apologize as far if I'm absolutely pulling this to pieces. You can interject, but the name of the blog post is called measure a mental model for decision making, and headline. You know, should you spend 30 pounds on a pair of jeans or a hundred pounds? You know, should you get answer of coaching on your glasses?

And I think a bit further down there was the example of when you were buying a bin Right, yeah, a trash pen with your mum, yeah. And you know, instinctively say, right, well, i don't want to spend much on a bin. Yeah, it's a bin. Yeah, logically it says, right, well, actually it's in the middle of the room. You know it's got a last years, otherwise I'm just going to end up replacing bins and it's got to look half decent because you know people are going to see it all the time. Right, yeah, could you just walk us through Because I was thinking a really interesting point on. You know, magnitude versus measure, the inspiration behind that, and is that the sort of thing that's kind of informed your decision making now that you're the CEO of a company. I thought it'd be something that a little bit different and interesting to pitch brains on.

0:31:09 - Taimur

Yeah, absolutely, and thanks a lot for taking through the archives of my. You know that's a little bit of a little bit of a. You know I appreciate it, yeah. So I think, yeah, just to kind of explain the concept, i think you know, when we're making decisions, there's kind of two things that we need to keep in mind. The first is, like you know how, how valuable is, like the thing that I'm getting, if it points something, for example? And then the second thing is how frequently am I actually going to be getting that value from it? So on one extreme, you've got something like you know going out for like a really fancy you know Michelin star meal, right, it's going to be very expensive but it's going to last, you know, a couple of hours max. You know you're going to get some value for maybe a couple of hours And you might spend, you know, 100 pounds on that, 200 pounds. On the other end of the spectrum, you'll have things where you almost don't even notice the value that you're getting from them, but you're using them really, really frequently. So something like this might be the quality of your mattress or your pillow or you know toothbrush and things like that that you're using really really frequently, and I think when, typically when we make decisions, it's very easy to, it's very easy to think about and kind of simulate the magnitude of the value that you're going to, you can think, oh, it's going to be an amazing meal, it's going to taste so good. You can, all you know, kind of feel it. But that's only kind of one of these components that we need to be thinking about, and so the point of blog post is basically to say, hey, i think we sort of systematically underestimate this other component, which is the kind of frequency at which you get value from things. And actually, if we were really thinking about this, you know, logically, you actually have, you want to put it I think we'd be investing a lot more in sort of the, the boring things that give us a small amount of value, very frequently over a long period of time.

And so, you know, most people, if they were going out to, you know, let's say, like a sort of a really special occasion maybe it's like an anniversary or a birthday meal or something you know, you might spend like a hundred pounds, maybe more on on this meal And you think, oh yeah, this is a special occasion Like this is. This is great, it's worth it, it's a great meal. But I think a lot of people, if you know they were thinking about spending a hundred pounds on a kitchen bin or something kind of boring like that, it would seem like a lot of money. Like a hundred pounds for a toaster Really get toasters for 20 quid, you know, whatever, right, but actually, like you know, if you have, if you have toast every morning, you're going to be using that toaster every single day is probably going to last five or 10 years. If it's a nice experience, if it makes it toast better, it's probably a lot better to spend a hundred pounds on that toaster than it is to spend a hundred pounds on that anniversary meal And so, yeah, the post we're just talking about.

You know, what are these things that we might be undervaluing And I think you're boring? kitchen objects like bins and toasters is one of them, i think. I think investing in ourselves is another one. So things like you know it might seem like a lot of money to spend like a hundred pounds a month on, you know, getting a fancy gym subscription or something, or maybe like 200 pounds a month on a personal trainer, right, it feels like a lot of money to spend on those things. But if you think about how important your health is, you're not just getting the value of that personal trainer in that half an hour session, one hour session once a week. You're getting the value in the rest of your life where you're actually healthier, your body feels better, works better, and so I think there's lots of things like that where we're probably underestimating, you know, the sort of less interesting but more frequent, more kind of long term bits of value.

And then to your second question.

I think within the context of running a company, i think when we're developing the product, this is definitely something that we think about around like hey, is this, is this part of the product, is this feature something that's really high measure Like? is it something users kind of using you know all the time when they're using the product? or is it, you know, tucked away somewhere where you know it's still like an important thing but they're not going to be interacting with it on a daily basis And just making sure that we really get the details right on those very high measure things. So you know things like the initial onboarding flow of the product right might have like 50, 100 people signing up for the products every day, every single day. You know that onboarding flow is being used by all of these different people And maybe only like one person gets to the very end where you have some like more advanced feature. Yeah, i think we definitely try and invest a lot in the high measure parts of the products just to make sure those are really good experience.

0:35:29 - Adam

So really interesting financially. So what can we invest in that essentially has the greatest impact over? Yeah, yeah, exactly, and we can. I'm going to add to my set of questions. I think we're still within time. Yeah, that's right, absolutely So, because I want to move on to talking about ROI, because I speak with a lot of people that know that they need to change but they really struggle to generate a business case.

So maybe you can move on to that in a second. But it's got me thinking because people are saying the great resignation is over and all that sort of stuff. Yeah, i get that, but there's still a huge turnover of people, just in general. Right, you know, it is still very tricky retaining good people. So if we're thinking in terms of high-measure tasks I mean you talking about software onboarding make me immediately thinking about employee onboarding. Yeah, exactly, is the quality of your employee onboarding directly related to whether they make it past probation, staying with you for however many years? Yeah, is their personal development plan something that you're investing enough time in for people to fill out their part of the culture? Yeah, at one point again, every company is different, but when everybody's so busy and pulled from pillars of post, sometimes people will skimp on the HR bits because it's just more admin. Do you see what I mean? Yeah, yeah, exactly. Is there a way that we can flip that? to think of that more logically?

But I suppose you could then also apply that principle to systems and process. It comes back to 80 to 20, i guess What's the part of our process that generates the most results, not just in the short term, but over time? How do we double down on that, right, yeah, i think again, it just comes back to better questions. People can ask themselves about what are we investing in and how does that reflect on our time. But a lot of people also see these investment decisions as kind of financial based. We've got big capex investment. We need to build it into the forecast. How is that going to reflect? But I think people often forget about the time element because they are so immersed in their day to day. So again, another question for you before we talk about ROI is obviously time is a commodity. How do you approach that in terms of your time based decisions and what to prioritize over other stuff?

0:38:02 - Taimur

Yeah, yeah, i'd say we're actually at a stage of the company where most of the investment we're making are time investments rather than dollar investments. Obviously, even when you're hiring people, you kind of think about okay, how much will we be paying this person over the course of a year? What value do we think they'd bring in? There is a bit of that, but I think the vast majority of the decisions we make are like, hey, should we spend time on X or Y? And we have limited people, limited resources and we have to try and make sure we do that really well. I don't know if we have any specific frameworks or tricks that we use. I think we just try and make sure that whatever we're doing right now should be the highest impact thing we could be doing, and I think we're reasonably relaxed about cutting projects that we now feel like, okay, this doesn't really make sense or there's actually something better we could be working on, whereas I think it can be a bit of a trap to get stuck in. Two months ago we said we'd like do this thing or like, oh yeah, i said I'd like meet with this person because we were working on that thing, but actually now we're not working on it.

Yeah, I think, just constantly evaluating and reassessing that we're doing this thing. Reassessing, hey, what do I think is the highest impact thing I could be doing across all the teams, across all the people in the company? That's just something we try and do. Anything on a personal level? Yeah, i try and sort of. I think there's definitely better weeks and worse weeks for this. But at the start of the week, look at my calendar and see, okay, what's the most useful thing I can do this week for the company, and are all the meetings in my calendar actually aligned towards that thing? Or am I doing a bunch of stuff that isn't that And actually I can postpone it or actually just cancel it and things like that.

0:39:49 - Adam

And again, i'm guilty of not doing that. Turn up on a Monday morning, immediately get swept into the to-do list. Yeah, yeah, it's so easy to yeah, absolutely. you know, nobody's perfect, so we can't beat ourselves up too much, i guess. Yeah sure

But that, i guess that's when it comes into and this relates to what we can talk about with ROI, because I posted about it a few weeks ago because it only takes a small step to free up a bit of time. That then enables you to free up more time, and you know, it's the whole principle of compound interest, right, you know, you need to invest a little bit and continue doing that to then reap the big rewards in the end. So when I speak with people with a laser focus, i say look, you know what's the goal and what's stopping you from doing that? Where's most of the time that's not useful happening? right, you know, and you only need to start with freeing up an hour or two a week, yeah, yeah, and then you're not going to be able to purpose that time with a view of then freeing up another hour or two. Do you see what I mean? Yeah, yeah, that's what some people sometimes miss when they're building a business case or deciding whether or not to invest in tools, because the immediate thought is you know, i can't take on another system. Right, i'm overloaded, as it is. Right, yeah, i can't think about another process.

But first, what process can you eliminate? And you alluded to it there. You know, are there any projects that we just need to scrap completely? Yeah, what projects do we need to scrap to free up time for the things that are then going to allow us to free up more time? Yeah, exactly, is that the way that you look at it? And for people that are looking to maybe look at the ROI from modelling tools and similar, is it time saved? Or I mean, i come up to I grew up in the RP background, so you know there's cost reduction percentages, There's all sorts of things in terms of saving number of hours. I'm to employ more people. How do you tend to think about that? not just from causal's perspective, but also from your own internals? How do we justify that return from something?

0:41:49 - Taimur

Yeah, i think time saving is often kind of the most tangible thing, where you can basically put like a number on it And, you know, depending on how you value your time, you can put like a dollar value on hey, this thing is worth this much to us in time savings. I think the thing that that often misses you know related to what you're saying is that the extra time or you know like in the context of something like causal or you know like a modelling or forecasting tool, you know we have a bunch of features that might save teams like a couple of days every month from not having to manually pull data from different systems etc. And making it quicker to run scenarios and answer questions. So there's a very tangible time savings. And I think the thing that gets missed from trying to be super quantitative about the ROI is that you can, you know, if it's say 10 times faster to answer some financial question or to run some scenario, it can actually sort of qualitatively change the way you work with the rest of the business.

You know, if people know that, okay, the finance team is going to say like three days to get anything back to me if I ask them something, they will only go to the finance team when they have, you know, at least three days before they need to make a decision And when it's something you know that they perceive as like, hey, this is like really big, we need to like plan ahead and go to the finance team like a week in advance, whereas if people know, oh cool, i can actually go to the finance team with random stuff whenever and they can give answers, you know, basically in real time for most things And if it's something deeper than they'll, they'll go into deeper.

It'll actually kind of change the way that they work with the finance team, will change that relationship completely. And I think it's hard to put a dollar value on that, but I think, intuitively it's like okay that that seems like a much better place to be in the business, where different business partners, different stakeholders, can come to us and we can provide real time answers most of the time, and then for bigger things we have to go away and, you know, spend a few days and do some analysis.

0:43:41 - Adam

It doesn't come back to a point there. It doesn't give you much time for the relationship piece. If all you're doing when you sit in a meeting is analyzing history, it doesn't make sense. Yeah, i think there's a. There's a big argument to be said that you know, as we spoke about earlier, get people in a room and get everybody involved in those decisions, right, right, and improve the way that the business works in general, right, but, as you say, it's difficult to quantify that sort of stuff.

Yeah, yeah, and what I've seen in our business and with with some of the companies that I work with as well, as related to that side of things, it's kind of staff satisfaction that's built into that as well. You know, when you look at younger generation of employees, right, you know, and it's the expectation nowadays, especially working remotely, yeah, and I'm not, you know, doing a disservice to anybody who spent a life, you know, building financial models in Excel, because you know Excel still exists. You know, and there's a, there's a business case for that. But I think there's a younger generation of people saying, well, i don't want to build a spreadsheet, i just wanted, i want to get on with the fun stuff. Sure, yeah, and so from an employee satisfaction point of view, that retention piece might go up as well. You don't know you can't say yeah, yeah, exactly.

And, as I say from my experience with the systems that we use, now we onboard new people and the feedback is we love the system. Yeah, just works so well. Yeah, yeah. Yeah, it's the stuff that is harder to quantify, but you know, it's still there.

0:45:11 - Taimur

Yeah, absolutely. I think there's always these sort of second and third order effects that can be kind of much greater than the really obvious ROI and health system.

0:45:20 - Adam

It's the same forecasting, i guess, isn't it? Yeah, you've got the second and third order. Yeah, yeah, we had another John. John Coley built a tool called Wifi. another model in part, oh cool.

0:45:37 - Taimur

Yeah, I think I've actually checked that out. Yeah.

0:45:39 - Adam

Yeah, yeah. So so he's good. I mean, he's got background in visual effects and he started it because he wanted to get better with his own personal finance. Yeah, by just saying, yeah, that's all from there. But his view is, you know, you can get rid of a single forecast or a single budget or ever, and then, yeah, based on multiple scenarios, right, you know, and this is where we're moving to. So, but no good, right? So the question that I always ask and again we've had some selection of answers to this, but the podcast will type of finance, right?

Yeah, i think I've been myself professed product and product productivity slash, app nerds, you know, whatever you want to call me. So I'm forever down on the apps on my phone. I'm forever trialing like bits of software on, on, on, you know wherever I can find it, yeah, so if you could pick a tool, it doesn't have to be a software, it could be like a gadget. You just can't say a smartphone because that's a call. Yeah, yeah, and is there like a gadget, piece of software or tools that you just? is it irreplaceable, that you couldn't level out?

0:46:42 - Taimur

I think for me, probably the biggest one is my.

I mean, aside from obvious things like smartphone Airpods, stuff like that, i think that the thing I'm most happy with that I haven't so changed for the last like six, seven years now, is the, the razor that I use for shaving High-smile It's, it's it's called, it's called the Phillips one blade And basically I mean it's just like an electric razor.

It's quite nice on the skin but you can also adjust like the closeness of the shave, and so it has this like attachment where, like, if you just want to trim rather than kind of completely get rid of everything, you can like set the set, the length and do that, and then if you want to kind of do the edges or whatever I mean I didn't really have a bit or anything, i mostly just like get rid of all of it. But yeah, it's just such a great experience. I started using it in like 2016 or 15 or something. So still using it, yeah, i love it. So, aside from all this stuff, like my phone and my iPods, my Kindle, i think, like this is I was using it like yes, i was thinking, man, i actually haven't changed this product for so long. I get newer versions of things or upgrade things or project. I haven't changed it in so long And I didn't think it will.

0:47:52 - Adam

That's an amazing answer. I'll let you into a guilty secret I have not shaved myself, and for At least four months, really wow, it looks pretty tidy for not having shaved. So my beard doesn't grow quickly. Barber just does it. Okay, two weeks and that that's it, you know, and by the unruly, i'm just back in the barbers again. Yeah, yeah, well, but not to to use it in, you know? yeah, and I was a fantastic answer. I did say gadget. So What are you reading on your Kindle at the moment?

0:48:32 - Taimur

Rating. I always have a bunch of different books and depending on my mood, i make sure of like business C type things. There's a sales book called Just effing, demo it or something. It's about sales demos, and then some like more philosophy type stuff But called after virtues, worked for at the moment on ethics Sorry for that effect. So there's always like a mixture and depending on my mood when I'm going to sleep whether I want something Like re cerebral, the way I have to like think about every paragraph, or whether I want easier, i'll kind of just based on that.

0:49:04 - Adam

Wow, okay, i did philosophy a level. Oh, it's cool. Yeah, i didn't get good mark. I love, i love nature and I loved and I loved a lot of the studies that we did, but I just completely rubbish my dissertation, like you know, i think like exams and yeah, okay, mark's this kind of different ball game.

Yeah, absolutely So. Yeah, so I did it on plate as Republic, particularly, something I was particularly fired up by, and sure very good marks, but no, i'm still into it. I mean, yeah, it's all good. So now that's been been fabulous time. I really appreciate your time. So, before we leave, i mean, where can people find out more about you? I'll put your link to your Twitter and and your blog in the show notes. Yeah, i think Twitter's the main one, and then, obviously, if, if, any folks here.

0:49:50 - Taimur

You spread sheets when you do any financial modeling or planning. Do check out causal or Zoolot app. You can actually type in casual dot app easy misspelling, But it still redirects the right place. So maybe the easiest thing is to go on casual dot app So you don't have to worry about the spelling. That's amazing And there's I mean there's there's a free version of it, right.

Yeah, yeah, anyone can just sign up and start using it, and then if you want to use more advanced stuff like data integrations, then you have to pay a bit money.

0:50:19 - Adam

You get like five, what is it? five models and the free one, i think, yeah, yeah, there's a couple of limitations right now. Yeah, and you're on LinkedIn too, right? Yeah, linkedin, absolutely, i've definitely dropped me a connection request.

0:50:28 - Taimur

If you're interested in saying in touch.

0:50:31 - Adam

Perfect, well, absolute pleasure. Really appreciate you coming off.

0:50:34 - Taimur

Cool. Thanks so much and all the best.

Transcribed by https://podium.page


tech for finance

©2022 by Adam Shilton. Privacy Policy - Terms of Use

©2022 by Adam Shilton. Privacy Policy - Terms of Use