Skip to content

Personio - Cash Forecasting Discovery - 2024-11-26

Metadata

  • Date: 2024-11-26
  • Company: Personio
  • External Participants: Tom Thorn (Treasury)
  • Palm Participants: Emma Sjöström
  • Type: Lead Call
  • Domain Areas: Cash Forecasting, Variance Analysis, Investments & Debt, Scenario Planning
  • Recording: https://tldv.io/app/meetings/6745e2dd642efd0014be47fd/

Summary

Context

Deep discovery call with Tom Thorn from Personio Treasury exploring what makes a cash forecast trustworthy, usable, and valuable. Tom shares detailed insights on forecast granularity, confirmed vs estimated forecasts, machine learning use cases, and the relationship between forecast accuracy and buffer management.

Key Discussion Points

  • Purpose of forecasting: Efficiency - having visibility enables better cash deployment and creates feedback loops with the business
  • Core forecast data points: Value date, currency, account, amount are essential; categorization (type of payment) is highly valuable
  • Confirmed vs estimated forecasts: Approved invoices = confirmed; pending approval = estimated; customer inflows = estimated with lower certainty
  • Buffers are a symptom: Lack of forecast trust leads to larger buffers on accounts, reducing efficiency
  • Machine learning use cases: Best for payments outside business control (tax direct debits, predictable patterns) rather than business-driven payments
  • Value date vs accounting date: Treasury cares about value date (when it actually debits); accounting may use invoice date or credit terms
  • Current process: Budget-based, manual spreadsheets, averaging monthly inflows across days, limited ERP integration

Pain Points

  • Manual spreadsheet process - No system, everything is basic and manual; takes significant time
  • Budget-based forecasting only - No transactional-level visibility; forces large buffers
  • Lack of ERP integration - Would love supplier payment reports from ERP but hard to prioritize
  • Cash inflow variance hard to explain - Took significant effort to understand why collections were off vs budget
  • Forgotten tax payments - Payments that happen once a year get forgotten, causing panic
  • Stakeholder alignment challenges - Business teams don't see value in improving invoice approval timing

Feature Requests & Needs

  • Confirmed vs estimated split - Tag forecast lines as confirmed (high certainty) vs estimated (assumptions)
  • Rolled-forward unpaid items - When forecasted payment doesn't happen, roll it forward and flag it
  • Machine learning for direct debits - Predict actual debit dates within tax authority windows
  • Automated bank statement reconciliation - Match forecasted items to actual transactions automatically
  • Treasury policy scenario planning - Model how liquidating investments affects policy compliance
  • Granularity for accountability - Drill down to specific transactions to hold teams accountable

Jobs & Desired Outcomes

Job: Achieve just-in-time funding efficiency across bank accounts

Desired Outcomes: - Minimize the buffer balances required on operational accounts - Increase the amount of cash deployed to higher-yield investments - Reduce the frequency of early investment liquidations due to forecast misses

Job: Identify and explain forecast variances to create accountability

Desired Outcomes: - Minimize the time required to identify which payment types are causing variances - Increase the ability to provide specific feedback to accounting/business teams - Reduce the frequency of unexplained variances that erode forecast trust

Job: Manage Treasury policy compliance while meeting funding needs

Desired Outcomes: - Minimize the risk of policy breaches when liquidating investments for funding - Increase visibility into how funding decisions affect policy metrics (liquidity %, tenor mix) - Reduce manual effort tracking investment portfolio against policy limits

Domain Insights

  • Four core forecast data points: Value date, currency, account, amount - "without them it doesn't really help"
  • Categorization is bonus but valuable: Knowing if it's a lease payment vs general AP helps with variance investigation
  • Confirmed vs estimated framework: Payroll, tax = confirmed; customer inflows = estimated; approved invoices = confirmed, pending = estimated
  • Machine learning sweet spot: Outside business control (tax direct debits with date windows) vs business-driven (late invoice approvals)
  • Feedback loop philosophy: Granularity enables accountability; can show business the cost of late approvals
  • Treasury policy complexity: Managing liquidity across maturity buckets (30 days, 1 month, 3 months) while meeting daily funding needs
  • Budget vs transactional forecasting: Management cares about budget level; Treasury needs transactional level for efficiency

Action Items

  • [ ] Design confirmed vs estimated forecast tagging
  • [ ] Plan rolled-forward/unpaid item tracking
  • [ ] Consider Treasury policy scenario modeling for investments
  • [ ] Explore machine learning for direct debit date prediction

Notable Quotes

"Cash forecasting from a Treasury perspective is purely around efficiency. Having full visibility of flows in and out at a transactional level allows you to be fully efficient with your cash - that is then a value add for the business." - Tom Thorn

"If you don't trust the forecast because you have variances and can't identify where they're coming from easily, you end up having bigger buffers." - Tom Thorn

"Do you want the machine to learn that it should adjust, or actually should it just identify and notify that it was paid late? That gives you the opportunity to align with the business to say, don't do that again." - Tom Thorn

"If you're efficient with your cash, that's what I would describe when it comes to cash management. If the rest of the business isn't hearing anything, that's really a good thing." - Tom Thorn

"At a point of significant transformation, you need to set good boundaries... If you can get that level of granularity, you can actually put a number to the inefficiency." - Tom Thorn

Full Transcript

00:00 Tom Thorn: Make it work.

00:02 Emma Sjöström: reliable forecasts that that people would use, right?

00:06 Tom Thorn: Yeah. Yeah.

00:07 Emma Sjöström: so, that's kind of Well, I'm trying to explore a little bit like yeah, settings and boundaries for us to be creative and innovate within but to make the

00:16 Tom Thorn: Okay.

00:18 Emma Sjöström: direction is the right one. That's

00:19 Tom Thorn: Right. But yeah. So you're still keeping in line with the ultimate goal of the good forecast that also trying to do some, some interesting of Yeah I don't know. It makes sense.

00:28 Emma Sjöström: Yeah. Yeah. How can we? Yeah, exactly. So how can we, you know, keep it familiar so it's like make sense to treasurers, but how could

00:34 Tom Thorn: Yeah, okay. Nice. Interesting.

00:35 Emma Sjöström: all these new innovative technologies? And yeah, how can we Think about the next steps so to speak. Yeah. but so yeah, this is this is going to be very like, just Broad picture large, you know, streaks, but it's yeah. I'm actually going to start with a really broad. Like, just what do you think is the purpose of cash forecasting?

01:05 Tom Thorn: um, in a word I would say, well, it Cash forecasting. Okay, from a Treasury perspective. It is purely around efficiency. So understanding visibility and what your your funding requirements that having full visibility of all of that. So any flows going in and out of your bank accounts at a transactional level allows you, then to be fully efficient with your cash that is then a value, add for the business.

01:39 Tom Thorn: And if you go a little bit deeper down into that, visibility point, it allows you to stay better connected with your staples. So the accounting team or other teams in terms of Feedback. So, for example, if you just have very high level supplier payments or very, you just have it at a very, very high level.

02:20 Tom Thorn: You just have, okay, we have 1, million's worth of supply payments today, that's helpful for me, if it's always accurate, right? But it's never always gonna be accurate. So if you can have more detail within there, that gives you, okay, what type of supply payment or who entered the supplier, The more data points that you have, if that payment moves.

02:43 Tom Thorn: So, say 80% happens with the day that's great. But the other 20% you want to be able to have visibility of that so you can have like a bit of a feedback loop within across the team. Say Okay What was there an issue with this was Proved late by someone within was the invoice approved late or, you know, what was the cause of the delay and because that then helps you to feed back to the business.

02:54 Tom Thorn: This is the cost of, of this kind of activity. Obviously, it's not always going to be a hundred percent, but it just helps having that extra level of visibility and detail. I guess helps you then to feed back to the business and get them to understand like Approving. A payment late from the accounting side has a knock on impact.

02:58 Tom Thorn: It means we're inefficient with our cash. We we may have liquidated and investment early and we needed to which means that we lose out on interest it's bigger payments so there's it helps to have that that level of visibility and granularity and for a forecast really coming back to that efficiency point of view, it helps the whole business, be more efficient with, with their cash.

00:00 :

03:16 Emma Sjöström: That makes a lot of sense. I you

03:18 Tom Thorn: Good.

03:19 Emma Sjöström: little bit like on the next question, which is around more about. Like what do you think is needed to make a forecast and like usable

03:27 Tom Thorn: Yeah.

03:29 Emma Sjöström: In a first place. But then also trustworthy would be the second one.

03:31 Tom Thorn: Yeah.

03:31 Emma Sjöström: and you already start, you know, you already started talking. So you would like that granularity, like some explainability about everything that goes into the forecast.

03:43 Tom Thorn: Absolutely. Yeah, absolutely. That I would say there's, there's like a few really, really key Core Data points to any forecast without them. It doesn't really help. So, if you have, if you have a very complex Treasury function, with multiple bank accounts, multiple entities, multiple currencies, all making payments, it doesn't help anyone just to say We have five payment today, you have no idea where that's five payments going from.

04:10 Tom Thorn: So you need information like which bank account is it going to be paid from? What is the value, particularly What is the value date? That's often one that gets very tricky to manage across team. So the accounting team look at accounting date, or they might look at an invoice state or they might face on credit terms.

04:33 Tom Thorn: But actually, when it flows through to our banking system based, or maybe the transactional time lines, set the payment versus an urgent, why payment versus across currency transfer. What these minor different things will have an on impact actually, when it debits our account, That's what treasure reasons. Just did is the value date versus some of the dates.

05:11 Tom Thorn: The accounting team may look out, and then you start to get into Interesting ones, which help from a Treasury side. So you have those three three main ones value, maybe Fort Wayne ones, value value, date currency and account. Those are like really needed, so we can specifically identify. This is the debt that's gonna hit that account.

05:19 Tom Thorn: Everything on top of that is a bonus, but there are quite there are some quite helpful ones. So, for example, if you have within your accounting system with you, have within your forecasting system, a way of identifying, the type of transfer it is Can be really, really helpful, right? So, if you just have those four first trans like those those four core ones, you just see a debit on your bank account, but you don't know if that's a tax payment or if it's a supply payment or a payroll payment if you start to be able to add more data points onto that and then maybe you have some subcategorization within that.

05:31 Tom Thorn: So you say supplier payment but it's actually at least payment or this one is a genuine AP payment. You then start to be able to have more granularity there and that really, really helps with forecasting versus versus actuals. So you say a week ago, we were forecasting to make this at the top level was forecasting, me and millions worth of supplier payments this week.

00:00 :

05:54 Emma Sjöström: Yep.

05:55 Tom Thorn: If you just have that and then you have like a really big difference that we only maybe 500 left the account. It's pretty difficult, then to identify just at that top level. But if you have that additional level of granularity to say, least payment, AP payment, whatever you can actually, then reach out to the individual stakeholders across the business and start to question it.

06:12 Tom Thorn: So you, you have that extra level of granularity, in terms of what makes it the full value for the weekly transfer to dedicated that then again, is that visibility point, that means a lot. So I would say there's some four really cool ones that extra add-on bit enough to be the challenge because the business isn't necessarily working for the Treasury Department, if that makes sense.

07:01 Tom Thorn: So you're you're having to influence the business a little bit to work in your favor and that's when that feedback loop comes into play to say, we need that level of visibility that granularity. Or we're going to be, we're gonna have to be, might, it might need us to be inefficient with my cash or We just don't have an understanding of where these variances are coming from which often leads to unfortunately having buffers on your bank account.

07:00 Tom Thorn: So you don't know, you're not again, you don't miss any trust fully the forecast because you have variances and you can't necessarily identify where they're coming from easily and so you you end up having bigger buffers. Yeah, good. Good. You can

07:07 Emma Sjöström: Right.

07:08 Tom Thorn: tell you a little bit about that one from me from our time at Uber, some of the challenges have to leave office on my accounts.

07:14 Emma Sjöström: I bet just to help me understand, right, so in my, in my mind, I'm

07:18 Tom Thorn: So sorry.

07:19 Emma Sjöström: picturing two different things, like one would be kind of the level of detail that goes into the forward, looking like the forward looking, it of the forecast,

07:28 Tom Thorn: Yep.

07:30 Emma Sjöström: and the other bit would be like the explainability provided for variance analysis which would take place off the you have your actuals already

07:37 Tom Thorn: Yep.

07:39 Emma Sjöström: there, right? So you can do some par.

07:40 Tom Thorn: Yeah. Yeah.

07:41 Emma Sjöström: comparison with the forecast versus what actually happened.

07:46 Tom Thorn: Yes, yeah, yeah fish yeah. What you

07:47 Emma Sjöström: Right.

07:48 Tom Thorn: would hope is that or what you want to be able to build out is some form of reference across both. So you have your forward looking forecast, that

07:54 Emma Sjöström: oh,

07:56 Tom Thorn: contains information around like that, that subcategorization that granularity on the transfers. But if there's a way to then reconcile, that directly with your bank statement, if it's an automated way that's like perfect, right? The bank statement comes back and it's able to identify Exact chance. The rate the exact flow that it was

08:16 Emma Sjöström: And what didn't happen? You

08:16 Tom Thorn: Exactly. Yeah. Exactly. The exact flow that it was based on and kind of recon saw this, then you get a lot of value. That essentially automates, a huge amount of what the Treasury team is doing that. That's very difficult

08:29 Emma Sjöström: it's

08:31 Tom Thorn: to do because yeah, what you then need to build out in a way, is it across the accounting team? If they're making these payments, if they're having these flows you need them to include reference information. Most likely include reference information in the payment detail to then be able to take it with in some way to say This is a lease payment or This is a you know and that's difficult to build out.

09:10 Tom Thorn: I mean that takes a lot of It's quite impactful for the accounting team and that you know again they don't necessarily always see the benefit of it. So you you end up from a traditional ideal world, that's what it would look like. But I'd say most of the time, well, with the experience I have is that the Treasury team is trying to build out ways of identifying themselves on a Best effort basis.

09:12 Tom Thorn: So when the

09:13 Emma Sjöström: Yeah.

09:14 Tom Thorn: statement comes back in you trying to, you know, some some of some payments are very obvious. The tax payment is pretty obvious when you're making it but other ones and you're trying to type them in a way to say, Okay, that looks like at least payment that looks like a general supply payment and you just manage it as best as possible based on your knowledge of what is actually set up to go through that bank account.

09:34 Tom Thorn: Save, one bank account only has 10 different types of payments in it, that's fine. But if a pain is used for, it's like an account is used for multiple different purposes, which often happens. then it then against

09:45 Emma Sjöström: Gotcha gotcha. So but they remember from last time we spoke you, you really like the kind of just in time mindset in terms of like running your

09:53 Tom Thorn: Yeah, that's that's me how I I yeah.

09:54 Emma Sjöström: accounts. Yeah.

09:58 Tom Thorn: that's me how it should be in an accurate forecast is Like a key component or the key component of that if you don't have. Confidence in, in the forecast. And Yeah, then you end up having to build buffers or It's not just, you don't have just in time payments anymore, because confidence is two things, right? You have confidence in the values or you have confidence in the valley days and if the value days away of say, a payment can be brought forward by a few days.

10:46 Tom Thorn: That means that you're often having to just leave balances on bank accounts either balances, essentially because you just not not confident that's all in. What you see, usually that, I mean, I wouldn't say often famous get brought forward, but it can't happen. And or there's just something random that doesn't, you know, doesn't pay everything gets improved, very late, and doesn't appear in the system anyway, anywhere until the day of the actual payment, Okay.

10:48 Tom Thorn: That what can you do if that's happening quite often? Then, Yeah, you have to leave either balances and try and buffers either. Balances you try and manage it in that way, but again, it comes back to that point of efficiency, it's just not efficient to do that.

11:00 Emma Sjöström: Right.

11:02 Tom Thorn: Of which, yeah.

11:02 Emma Sjöström: I'm trying to figure out the ways in which a system could imagine you. Yeah, you do manage a bunch of bank accounts and then they A system with like what would the outcome you're looking for? Is a way to basically identify, which accounts you having inefficient process or like related in efficiencies in

11:21 Tom Thorn: Yeah.

11:22 Emma Sjöström: of maybe accounting not like approving payments on time or like making it hot and style due to like, I don't know. The references of the transactions, not being good enough or like you see trying to dig deeper into like what is the

11:37 Tom Thorn: I would say the yeah, that I would.

11:38 Emma Sjöström: Didn't.

11:39 Tom Thorn: the challenge doesn't necessarily happen like in an account level it usually happens that a payment type level or so you. So for example you

11:45 Emma Sjöström: Okay.

11:48 Tom Thorn: If things are relatively centralized within a company you'll have one team who's managing your AP payments. You have one team who's managing payroll? You'll have one team who's Manag?

11:55 Emma Sjöström: Right.

11:56 Tom Thorn: you know, personal, whatever it might be, you have that one team managing it. They're then often. Reaching out and collaborating closely with. That they're in charge of make they're in the they're like the operational side of things, right? So then making those payroll payments but then not the cost center owners.

12:14 Tom Thorn: They're all the business owners who are actually owning those supplier contracts. So you will have people right through, like down to the business manager level, who don't have any interaction with finance, it's not their job to do that. It's not a job really. Their job is like building out customers and, you know, whatever it could be multiple different.

12:30 Tom Thorn: But they're the ones who actually are in charge of approving an invoice to be paid.

12:35 Emma Sjöström: Yeah.

12:36 Tom Thorn: And for whatever reason, it's just not a priority for them. It doesn't get approved. That has a big knock on impact through the business, the accounting team. It means it just goes through to the next payroll run. But from a forecast, point of view, it could be by the following me, Okay? We're not gonna pay this week.

13:09 Tom Thorn: We'll pay next Friday. Instead, the, the issue for us is that that's an additional week, where we funded for that payment and it has an impact on our, on our, on our interest, on a yeah, cash management and the balances that we hold. So, it's, I would say, It's not necessarily on the account itself is generally like, say you just have an issue with the real estate team.

13:30 Tom Thorn: Like they just process invites invoices whenever they want to, they don't load them into the system correctly. They don't prove them on time. They, you know, they're just difficult to deal with because they don't see the value in improving them you know. Managing them as we like to, that would have an impact over all accounts pay real estate all accounts playing easy.

13:30 Tom Thorn: So me just have an issue with at least payments and then have. So going back to that point I

13:33 Emma Sjöström: Yeah.

13:34 Tom Thorn: was saying earlier, if you have that granularity, when you say, you have So you categorize it that you have supplier payments, and then you have general AP transfers, you have, then you have leases being one element of that like one like subcategory for example. And you see like every week there's just these wild variances from your forecasts, when you see

13:51 Emma Sjöström: Yeah.

13:52 Tom Thorn: you see the payments go through, you can identify that feed their back to the accounting team. Hopefully the accounting team, then picks that up with the real estate team. So what the hell's going on? Yeah, so that granularity helps with that, that feedback.

14:05 Emma Sjöström: That makes a lot of sense. okay, so so get, let's say that we was possible to provide that level of sort of like, Variance analysis or like easy access to why. But before that even the forecast itself like again like just really doubling in on that like what? Like what what will make you trust a forecast that is not necessarily coming 100% directly from you? Do you see what I'm trying to get like what

14:33 Tom Thorn: Yep. Yeah, yeah.

14:35 Emma Sjöström: and Relaxed in the first place, right?

14:36 Tom Thorn: I, Yeah, so I I the way I would usually frame it is there's certain payments that have high certainty, right? I'm pretty confident that the payroll team is gonna hit their their value date. Because if people don't get paid on time, they're gonna be pissed. Right? That. Yeah. All that.

00:00 :

14:54 Emma Sjöström: Now, that's

14:56 Tom Thorn: There's other types of tax payments, again, they're usually quite quite accurate. I mean, the values are very high, generally for the business that like, in terms of them over all that they're pretty high. And so you. And and they generally, you know, if they're direct direct debited, they're generally going to fall within quite a tight window.

15:30 Tom Thorn: So there's certain ones that hats and forecast types of payment types that have a relatively high degree of certainty. There's other ones that then you would apply, maybe a slightly different view on. So you take customer inflows, the actual influences you have very little unless it's a direct debit or something like that, you have They less certainty over those.

15:31 Tom Thorn: So

15:33 Emma Sjöström: Yep.

15:33 Tom Thorn: you probably put them at a slightly. Like it. I mean you have a threshold, I guess you in an ideal world, you put, maybe a percentage against it. So Actually systems can reverse the reference one I have. They actually had quite an interesting one, where they would have. You could separate out between what would be called concerned? Forecasted can't confirmed forecasts, which would essentially tag that line to say.

16:00 Tom Thorn: We're pretty sure that these are gonna happen on this day or within that like one or two days and you would have estimated forecast which is more based on assumptions. We think that's the pain. We're going to go through the other One. I would say is Yeah, invoices If you're able to so that, that's so that.

16:14 Tom Thorn: What if I covered that customer inflows probably that's more of an

16:18 Emma Sjöström: so,

16:18 Tom Thorn: estimate payroll or tax? Probably more confirmed, forecasted confirm, force, casting when you come to something like supplier payments, if you can base those off invoices through your, ER, for example, invoices that are in there. And maybe you have invoices that are already approved. Right? That should be for, I would say that's confirmed forecast that we know that they're going to go out within the next payment run because they're already ticked off all.

16:42 Tom Thorn: They need to be done, is loaded up through that, that payment batch, and they're gonna go through. If you then have another line, which

16:48 Emma Sjöström: Yeah.

16:48 Tom Thorn: I've never seen, but it would have been really helpful to have is invoices that are due to be paid. Like, based on the payment terms, they're due to be paid on that day, but they haven't yet been approved. I would say those are estimated.

17:02 Emma Sjöström: Right.

17:02 Tom Thorn: Does that make sense? So you have

17:03 Emma Sjöström: Yeah.

17:03 Tom Thorn: like everything that you get to a layer maybe across all of those you have like A split of say 75, 70% of those payments of what we would consider confirmed forecast, like payroll's. Gonna go out on the day tax, get around, they approved. Invoices are gonna go out through that payment run, we can go.

17:36 Tom Thorn: Okay. Those are, those are pretty good and then the rest of the 30% were a little bit unsure about, but we have some degree of certainty that, you know, that some of those big, maybe some of those bigger invoices gonna pay them that day, or maybe you even start chasing up through the business.

17:51 Tom Thorn: To say, We have a surprise payment. It's 500 K, it's not approved yet. Is that confirmed because they're gonna be pushed, You know, then that you actually have some interaction. If you have again, it's the same wording over again, you have that visibility and that brand new latency to spit it out in that way.

17:48 Tom Thorn: And you just have more ability to to align with business and engage, you know, ask those types of questions.

17:54 Emma Sjöström: So would you say that it would be helpful if you're managing multiple accounts to get some sort of notification of like Hey these are unpaid like do invoices that were and paid or like

18:08 Tom Thorn: Yeah. No, that would be it. Yeah. So

18:08 Emma Sjöström: Yeah.

18:10 Tom Thorn: yeah. For sure if you had something that, well, I guess what I would hope to see is if you had like a forecast,

18:16 Emma Sjöström: Anything that you confirmed in your forecast, I didn't happen.

18:19 Tom Thorn: Yeah, that would be a good one for sure.

18:22 Emma Sjöström: And like of, yeah.

18:22 Tom Thorn: Yeah. Yeah, so if you have that forecast hospital, so you would hope that it would just roll forward as well, right? So if you had a full so you don't want to forget about it. If it's not being paid, it's probably still gonna get paid at some point so you don't want to forget about it.

18:36 Tom Thorn: So you still want it to appear within your forecast, but if it was able to go Say be just started breaking out that that list of. If there was like another one underneath that that said, I don't know, I don't know what the title would be with like previously forecasted or just unpaid or something forecasted unpaid and

18:55 Emma Sjöström: Yeah.

18:55 Tom Thorn: you'd have this other category of things that are rolling forward, but then you can start to see those. That would be really helpful. Yeah.

19:02 Emma Sjöström: No. I mean that makes sense. And how do you feel about like incorporating any form of machine learning or AI in this?

19:12 Tom Thorn: Yeah.

19:14 Emma Sjöström: Like what? like just your wireless dreams, what

19:15 Tom Thorn: Oh yeah.

19:17 Emma Sjöström: would that look like?

19:18 Tom Thorn: Yeah that that no food 100%. That would make a lot of sense. The big one I see it working on is like is validates and a good example is like direct debits. So tax payments direct debits. You often. If they give you like a window in terms of when they're going to debit.

19:36 Tom Thorn: So it could be one to three days. We want to five days you'd ever quite know. But if there's some machine learning that takes a bit of the manual effort out of your hand, it's like not the biggest deal in the world. If you know you're gonna fund for that first day on the assumption that it could happen over the next week.

19:58 Tom Thorn: But say for example there's some machine learning that would say in February they usually debit on this date in March, they usually debit on this day in, you know, forward across the year. There could actually maybe give you more of an act more accuracy, more certainty of the actual debit dates and based on his thought, Yeah, it makes some machine learning.

20:20 Tom Thorn: I I, that is where I see, right, you know, some help coming in. So the invoicing everything that's within And the businesses has, it's kind of like a manual, a push from the business. I'd be that. That is, I I think maybe Machine learning can come into that, but I think it's more of a, an alignment with, with stakeholders say, If an invoice, is saying this date in the forecast and doesn't get paid on that date.

20:59 Tom Thorn: It's almost like, Do you want the machine to learn? That it should adjust it or actually from my perspective. You it should just identify it notify that it was paid late. And it gives you the opportunity to align with the business to say, Don't do that again. But when it, when it's anything that's outside of the business's hands, Then, yeah, I would say machine learning can definitely come into it.

21:00 Tom Thorn: And maybe, maybe identify like. Sometimes you just have payments types of payments that people forget about and it happens quite often, unfortunately, but you can have certain like a tax payment that gets paid once a year for a, for a company that has very little else going on on through it and everyone forgets about it and then it gets to that and it gets that point in the year and everyone starts panicking.

21:39 Tom Thorn: No, we got this big payment, there's no cash in this account. There's no into company loan in place or whatever and it's, it happens more often than not if you have a lot of different things to look at, it can happen it again. It was a good example who's happening all the time, and if you can have some machine learning in there that maybe build out into like, Three months prior to that happening.

21:45 Tom Thorn: We just have like this estimated forecast to say, Maybe I'm not again, going down that list of lines like confirmed forecast. Estimated forecast rolled forward from non-paid and you'd have, like, machine learning based on last year to say, Tom, in March, last year, there was this payment. Don't forget, it's not in the system.

22:03 Tom Thorn: Yet, you know, something like that would be would definitely be helpful for sure. I think I I think the forecast I mean like a combination of all of those is very tricky to do but that would be like in an ideal world that that's how it would be.

22:19 Emma Sjöström: So, thank you. Would you mind walking me through kind of your typical cash focusing process as well?

22:26 Tom Thorn: Yeah. Episode. Yeah, I can I can try. It's not great. I'll be honest.

22:33 Emma Sjöström: like,

22:35 Tom Thorn: So yeah, we We have a few operations, we don't have. It the the cash management. The liquidity structure here is actually less is not the most complicated, so it's relatively straightforward. And but essentially what I do is I will

22:49 Emma Sjöström: Yep.

22:52 Tom Thorn: take budget Forecast cash, inflow. So that's not a transactional level. That's just budget for the month and then based on. So for instead of having Going back to what we discussed before, you have those transactional level forecasts. You need account. Account information at the account number basically or some some account, reference in your currency.

23:15 Tom Thorn: All this kind of value, all this kind of stuff. I don't necessarily have that at the moment, so I just take the budget cash, inflow for the month, and then I split that up per day and based on kind of a Best estimate basis of when we tend to receive more cash going to when we tend to receive less cash.

23:32 Tom Thorn: So for example in the middle months with it, in middle weeks within a month we tend to receive more cash as well.

23:38 Emma Sjöström: Right.

23:38 Tom Thorn: I'll kind of adjust it a little bit to that but I generally average it out over a cash inflow across the month. Just say Okay, on a regular month, this is how much cash we're going to receive over the like each day based

23:51 Emma Sjöström: Right.

23:51 Tom Thorn: on what has been budgeted by FN 18. So the FN 18 is, is building out their own engines and their own forecasting based on all multitude of different things. So, I'm happy to take my numbers from from them and then just kind of split them out in my own way, in a man in a manual spreadsheet.

24:12 Tom Thorn: I also that so no system may notice system base. And then doing something similar for US, payments. and payroll, we have a dedicated file where the payroll team will input their information that I aligned on line with them earlier in the, in the earlier, in the year, essentially, for each different entity, each different So each different market each different entity, there in putting information to say, This is the amount that we're going to pay for salary.

24:45 Tom Thorn: This is tax. This is pension, this is miscellaneous stuff. And then within that, I built out like some rationale in terms of validates, when they usually pay that there's some of those salaries always pay take 25th of the month, but other payments, particularly tax can happen across the month.

25:05 Tom Thorn: So, getting a bit of a bit more granularity on that is, is quite helpful, but that's essentially it is Some of the forecasts. Is built on a bit more of an alignment with the internal teams. So payroll for example, others are take basically taking our budget numbers, and then trying to Assign those across to the bank accounts somewhat based on just just averages somewhat based on historical data and what what tends to go through those bank accounts.

25:56 Tom Thorn: So for example, if I have 10 million for the month, in terms of supply payments, I can take a look at what we paid through our GBP and USD bank accounts in the previous month and that probably gives me a relatively good estimate of what's going to be paid for this month.

25:59 Tom Thorn: So I deduct that from the Euro budget value. I then split out the Europe, budget value. Oh, like we pay, we have like twice daily payroll runs for our euros, but then once a weekly GBP in USD. So based on some of that rationalized spit it out and basically then in that same spreadsheet, I take that information all the all the various numbers validation so forth and plug those into various different.

26:24 Tom Thorn: The tables to bring me through to a daily cash flow forecast, where I expect my balances are gonna be. So I have that. That's the fork up the fourth forecasting bit. And then I also need to to your point layer in. There's two parts that the opening balances from those bank accounts and that allows me then to, to have a bit of you at the beginning of each month and then throughout the month of And where we're going to be bounce-wise these bank accounts.

26:37 Tom Thorn: but,

26:38 Emma Sjöström: Right.

26:38 Tom Thorn: The issue with that. It's very, very basic. It's based on all of estimated numbers, so it's the budget numbers versus like confirmed forecast. It's all budget numbers. And meaning that I often, I'm often leaving big buffers in a bank account. so, That's where the lack of a system and the lack of like really.

26:58 Tom Thorn: So having like a feed for example, from my ERP system, to say, supplier payments and this guy to have that directly coming in to a system or even into a spread if you had a report, something like that and would really give me much more granularity in terms. So at the moment say, I'm always buffering or is assuming Three million a week is going to, in terms of supply things, he's going to go through our Euro operations account.

27:24 Tom Thorn: But if I have a report to say in week one, it's 500k in week two, it's three, it's three, it's three point five million in a week, you know, that would help me quite a bit. I could be probably a bit more intricate with my

27:36 Emma Sjöström: And started to make clear that report would be based off of historical.

27:40 Tom Thorn: Invoice. No. I I yeah it

27:41 Emma Sjöström: Or. so, like, Some like the actual heart? Yeah.

27:48 Tom Thorn: Yeah. Say sorry. I don't know but it was like if we have that and if that

27:52 Emma Sjöström: just,

27:52 Tom Thorn: report, if that report could be based off our ERP system and that if going back to what we discussing before is you had that approved, like invoices that are definitely approved and then you have the invoices that are still pending approval. If you had a report like that, that would help me again quite a bit.

28:25 Tom Thorn: But at the moment, everything is at such a relatively kind of basic level. I would say that. It's always like where to start in terms of improving that. So suppliers is something the supplier in flow bit is something that we could, we could probably improve on. But it takes in terms of priorities with well, from my side, but then also with the accounting team and, and essentially, the projects that they have going on, it's often one of those things.

28:40 Tom Thorn: But okay, from my side, love to be able to build something up else out like that. But there's this, there's this like, balance to say, be great to have that, but maybe in six months time or so, we're actually implementing assistance that's gonna do this for us. So how much time, how much time do we invest now to, to

28:56 Emma Sjöström: Hmm. No.

28:56 Tom Thorn: improve it, probably what all this other stuff to do, you know?

28:59 Emma Sjöström: That's very, I'm just so. I think the buffers are very interesting. I think the granularity is interesting it, but also like to just ask you about so that Probably certain types of cash flows. That are more predictable than others. Right? And if you know those

29:16 Tom Thorn: Yep.

29:19 Emma Sjöström: could for example rely now I'm just I said because about Palm has today currently, right? We have a Phone calls based off of historical transactional data. Do you see that there any categories or any types of capitals that you would find that useful?

29:38 Tom Thorn: Yeah. Yeah, absolutely yeah, payroll payroll intact payroll and tax defin.

29:41 Emma Sjöström: Yeah.

29:43 Tom Thorn: definitely. Probably customer inflows as well. I'd say all those three. Payroll and tax because they're very repeated you know, month to month. They're not going to shift a lot. Unless there's this huge recruitment drive or, you know, something there's something big happening within the business, they shouldn't they, they shouldn't change much.

30:23 Tom Thorn: The one that I I would say is a little bit. In terms of machine. Like, in terms of back, we're looking in statement is the supplier side because a lot can be changing on that we month to month. So if Palm, for example, is able to get right down to the supplier line to say, we're going to play Amazon Web services, this amount every month.

30:32 Tom Thorn: That's great. If you would if you were able to have that but if there's new supplies that there's new big contracts coming on the swings within supplier. Payments can actually be pretty significant and if it's always based on statements, maybe you get majority of it. But maybe this potentially miss out on some pretty big new invoices just because there's there's no precedent for it.

30:54 Tom Thorn: There's nothing to learn off. If that makes sense.

30:56 Emma Sjöström: No, that makes 100% sense. And super cool. And so, would you say Just I'm just curious. Like, now I'm just going away bit from my script. I'm just curious about your boppers, like, I'm whatever assume that they are not.

31:14 Tom Thorn: Crack.

31:14 Emma Sjöström: on, like, supplier, like the account.

31:15 Tom Thorn: Yep.

31:17 Emma Sjöström: and they are for payroll or acts or

31:18 Tom Thorn: Yep.

31:20 Emma Sjöström: Yeah, these. So, I'm just trying to

31:21 Tom Thorn: Yes.

31:22 Emma Sjöström: get that, like, where the value would sit with you to use an external forecast, like it would be to drive down Us. And it is on the more complic.

31:30 Tom Thorn: yeah, no that I absolutely so that

31:33 Emma Sjöström: types of cash flows. Like Right, trying to make but it's

31:39 Tom Thorn: there's

31:40 Emma Sjöström: You would appreciate sorry, I'm just trying to

31:42 Tom Thorn: Better to be sorry. Go go.

31:44 Emma Sjöström: You would appreciate like you could use the machine learning based forecast for the payroll or tax.

31:50 Tom Thorn: Mm-hmm.

31:50 Emma Sjöström: Or a customer. Maybe customer inflows is a bit special but still like you would appreciate it. But would it be like, The thing that would make you buy the product, you know what I mean?

32:03 Tom Thorn: would help if you could have

32:04 Emma Sjöström: Yeah.

32:07 Tom Thorn: Taking out any kind of human potential for error. So, the I just described to you when I say spreadsheet, it is very, very basic very basic and it's and it points So you take that element out

32:19 Emma Sjöström: so,

32:20 Tom Thorn: element out of it. That's already like a de-risking for sure. The machine learning and having everything projected essentially is I'm doing at the moment but just at a very, very basic level and without the Insight the that that really key learning. So I'm just basing it on what I see through the accounts.

33:09 Tom Thorn: It's like a general. Okay, that looks about right? Whereas if you have them you get a look you get a lot more confidence in what you're looking at. So to your point, all of that allows you to be a bit more confident to bring down those buffers. Say for example, you're leaving five million buffer which not too discipline doing the moment.

33:06 Tom Thorn: A big buffer on our operational bank account and any given day if you're able to half that over the course of the year that's huge in terms of interesting, interest earning The second part of that. So that that's just the, that that pays a lot for itself. So you get the system, you get more confident in the forecast as saves a lot of your time or I'm doing all that is not.

33:36 Tom Thorn: None of the best way you get more insight and then you get the benefit hopefully through those two things and and through the that will automated. And and it's basically taking the time away from data, validation and manual effort into analysis and you're actually then able to make better decisions and more informed decisions in terms of your investments, that pays for itself into should pay for itself in terms of interest.

33:40 Tom Thorn: But the

33:42 Emma Sjöström: What?

33:43 Tom Thorn: other bit is the risk bit. So if you're again I'll use the same example. But if there's a tax payment happened in December last year, and I'm not aware of it and we miss it because I, I didn't fund an account that was supposed to be funded. That's also a big part of the machine learning.

34:31 Tom Thorn: So just having that that, you know, that depth, that that like reliance on something, that's systemized versus all different types of tribal, knowledge and manual effort, and manual bits of slack alignment to say, Oh, we got this payment going out next week. It's not just the Treasury team. It's all different types of teams impacted in terms of Effort and time and and distraction whereas if you have a system that builds it out and has everything laid say deemed a 70% of that, it's a huge huge time saying that and it should, you know, theoretically should give everyone a lot more confidence in terms of their investments.

34:35 Tom Thorn: So, yeah, it's a risk and, and an efficiency.

34:39 Emma Sjöström: Yeah. Now that makes a lot of sense. Thank you. and I'm just gonna, Do. So yeah, I mean we've covered it partly, but like which parts of the Cash 4 casting process. You find more critical and why?

34:57 Tom Thorn: Yeah, okay. I would say the Which was the cash forecasting process. I would see it's the In terms of my process, I see it's beginning. If the month process, where I refresh all of my sheets, new data goes in there, new, but so I have to check and recheck that make sure I'm comfortable with the buffers that I'm going to leave on the bank accounts.

35:25 Tom Thorn: It's getting a bit of an understanding of when I'm going to need to take action to funds or defund certain accounts during the month, I then throughout the month, I've recheck recheck that and see how things are going but that's probably the most critical is a refresh of data and don't back we discussing before.

35:45 Tom Thorn: If that was automated that that takes it. Yeah, that's a big time to, right? So that's where a lot of the the company comes in crystal bit is just Rechecking work as well, is to make sure if I'm assuming a payment is going to go through a certain bank account.

35:58 Tom Thorn: I'll have to check that number of different times. Another critical bit. We haven't necessarily touched on. I know we're discussing forecasts at the moment but what is sits with me at the moment is all the cash flow reporting so it's like the other part of that is the the data coming back in and us doing a bit event.

36:12 Tom Thorn: Sorry. Quite a detailed exercise looking at what was forecast, what was and the actuals

36:18 Emma Sjöström: Yep.

36:18 Tom Thorn: having it, you know there's a way If the system had that learning within there, if there was visibility on on historical information as well and that was able to be split up in a similar way. That's also super, super critical part. Actually, when I was looking at the budgeting for next year, hopefully, with the system involved in that, that's one of the critical parts I I discussed is that we're doing all of that.

37:09 Tom Thorn: It's not necessarily a reconciliation, but it's a budget versus of like it's a budget versus an actual piece and it's the way that we're doing at the moment is very, very manual. It's really, really trying. It's very very challenging to take all of that statement information from from our ERP system and essentially map that across the way that like our FP and a team want to see the information.

37:35 Tom Thorn: So you have Two and a half thousand, three thousand transactions, during a month, we're then in taking that and we're having to sort filter through it map it and get it into a format that looks, right? And that that's a real challenge that that's where if the system is essentially able to automate that piece for us as I assume it would be based on that machine learning thing, that is a huge part for me.

37:28 Tom Thorn: So the, there's the refresh at the beginning of the month, the forward looking piece, but that actually then the back we're looking piece that they enables us to. And essentially get it and understanding of how we landed, that's the budget.

37:41 Emma Sjöström: Yeah, and why you learn?

37:42 Tom Thorn: Yeah, yeah, exactly.

37:45 Emma Sjöström: No, that's That's difficult to hear. We're actually doing a lot of like yeah, automated variance analysis. Also in the app. Which is I like it a lot.

37:55 Tom Thorn: Nice. Yeah, I know. It's good to hear. No, for sure. That that's a Yeah, it's it's a pain point. It's a definite pain point for us at the moment it takes up a lot of my time. It has done throughout the year. It's and we've been trying to a good example.

38:17 Tom Thorn: So good example. This is like one of the challenges we've had is on cash inflows. We had variances to our budget cash inflows actually supposed to budget for months and there was really, really challenging to get good insight into what was quite causing that. So obviously all types of things from the Treasury side to say, timing of transactions, impacted various different things that AR team were doing that kind of thing.

38:43 Tom Thorn: And I feel with machine learning that probably, we would have been able to get an answer with that relatively quickly. Okay, I don't know how how much time and effort to get to get to, you know, relatively basic commentary with some machine learning. That's good. Say Certain transactions role forward because of, you know, month and dates or weekends or public holidays or all this kind of stuff that they can happen.

39:04 Tom Thorn: You know, it's just general. Yeah, it's not the most tricky stuff, but to understand the actually cash impacted that was was very challenging. I'm not sure if that's actually available within Palm, but that's the kind of thing I think with some machine learning may be, we may be able to get there a little bit quicker than we did in the end.

39:23 Tom Thorn: Yeah.

39:24 Emma Sjöström: Yeah, just not yet. Not yet in that

39:25 Tom Thorn: Anyway.

39:27 Emma Sjöström: sense but we're doing now. Is we really want to make sure like the foundational like the forecast

39:32 Tom Thorn: Yeah, of course. Yeah. No, no. I'm

39:33 Emma Sjöström: Is the best it can be. Like the felt we have like an in my opinion really great kind of blueprint around like cash positioning workflow. now we're really doubling down like on the certain parts of it especially forecasting just to make sure like okay You trust his forecast, he used his forecast, it will actually help you, right? That's kind of

39:53 Tom Thorn: Yeah.

39:55 Emma Sjöström: Where we're getting out, even though it is actually surprisingly good already. In terms of just like this.

40:01 Tom Thorn: Yeah.

40:01 Emma Sjöström: Then there's always going to be certain account. Certain categories that of course, looking only at transactional data, like the history historical data, like you mentioned all it's not going to be, it's not going to be enough for every every single use case, So I mean it feels like we've already touched but it's like nice to really, really make sure.

00:00 :

40:27 Tom Thorn: Good.

40:27 Emma Sjöström: That your point of girls, right? So but how would you define a great cash forecast? What would it consist of? And what impact would it have? Like, what would be like the wider impact of a great forecast as well? So it is

40:38 Tom Thorn: Okay.

40:39 Emma Sjöström: and how it kind of, yes.

40:40 Tom Thorn: Yeah and yeah okay that's good question. It's like a bit more philosophical.

40:47 Emma Sjöström: What?

40:48 Tom Thorn: I'll say. The forecast like a high degree of accuracy, but then, on with the inaccurate, but understanding where the inaccuracies may come from and being comfortable with us. So, I accuracy in forecasting is, You know, we want it to be a hundred percent. But having understanding, if, maybe it said, 80/20 and within that 20 being comfortable, why, why it's like that? And having, yeah, enough granularity that you can and visibility across all those various different practices that we discussed before.

41:41 Tom Thorn: And And in terms of the process, everything should be repeatable. You don't want to be having to change things around all the time, right? So, if you can get to a standardized process, a standardized format that everyone's comfortable with, it's intuitive and easy to visualize. It'd be surprising like how across these companies.

41:47 Tom Thorn: I've been at how differently forecasting is managed how differently it looks like. If it's a spreadsheet if it's a system and it's build out very differently but if you can have something you just plug into and is recognizable straight away helps a lot and then in terms of the value to the business It should then be.

42:18 Tom Thorn: Yeah, it should be the efficiency points. It's actually should be almost like a value add piece. So and yeah, if you're efficient with your cash if they're essentially, That's what describe it when it comes to like cash management and forecasting and funding and investments and so forth. If the rest of the business isn't hearing anything, that's really a good thing.

42:20 Tom Thorn: You know, it only really gets to like the management team of the financing of something's going wrong. So essentially you want nothing nothing within that. If you mentioned it before, if there's like a reporting that automated reporting, that you can say, You know, this is also where a good system comes into play.

43:07 Tom Thorn: To say, we were out within our forecasts, from our forecast by this amount. All we unfortunately, behind to keep a higher buff from this account, because we have less certainty of the of the cash flow is going through it. If you can get to that level of reporting, that gives a lot of vis.

42:54 Tom Thorn: Tea and oversight to the management team, where some of the inefficiencies across the business might be. And so, if there's something that, you know, Your accurate and able to rely on, but understanding those variables and being able to report on those effectively. And and everyone having full over side of that, I think that that's also a big one.

43:14 Tom Thorn: So value added efficiency, just making sure where we do with our caches as a business. And we're investing where we should be based on that good forecast, but where the forecast isn't working. We're also able to report that and we have insight on exactly what is driving it.

43:30 Emma Sjöström: Yeah, that's awesome. Thank you.

43:34 Tom Thorn: it's

43:35 Emma Sjöström: So do you I'm staying here a little bit like I You say that any specific metrics or feedback that you would use or could use to evaluate whatever forecast like your ideal forecast would meet expectations.

43:50 Tom Thorn: Yep. Yeah, yeah. And I deal World. I'm actually the best that I've seen it and a couple of companies ago, but

43:56 Emma Sjöström: Huh.

43:57 Tom Thorn: they they were able to drill down to a really, really detailed level. So you had almost like they're either in the county team are actually some people in the Treasury team. They would be almost budget owners for a specific entity, something like that. And you'd be able based on the actuals and the forecast accuracy because it was such a high level.

44:16 Tom Thorn: They were able to say that you're Accounts payable was off by 20%. Of the last month. Why was that? And maybe then even bring through a couple of specific transaction examples to say, You know, this is a big transaction. Why, why is it late or whatever? It's got it, I mean, it's it sounds pretty harsh, but it's almost whole thing.

44:51 Tom Thorn: Some of those teams a bit more accountable for their processes because they're knock on impact is is pretty big for for the Treasury team. So if you're able to get that, that level where you can again, I've mentioned it. All right, let's keep repeating myself Sorry. But those that that level of granularity, where you can pick it out and then you can you can pick out it forecast item and then it like See the actual come through or not.

45:45 Tom Thorn: That then really helps you to feed back into the business and hope. Yeah, hold some of those teams accountable, particularly when At a point of significant transformation which that that company was going through. At the time, there's all types of changes in terms of processes and new people doing new things or whatever.

46:16 Tom Thorn: A lot of upheaval, you need to set some good boundaries in terms of this, you know ways of working and and setting out go post in terms of what people should be aiming for you. Set that out early on to say, from a KPI perspective, hit your type like This is what we're expecting from a Treasury department.

46:21 Tom Thorn: If you don't hit those targets, the impact in terms of cost of capital is X, if you can get that level of granularity then that's, that's a yeah, huge huge. You can actually put a number to the in,

46:34 Emma Sjöström: Right.

46:36 Tom Thorn: At a point of significant transformation which that that company was going through. At the time, there's all types of changes in

46:37 Emma Sjöström: Thank you.

46:38 Tom Thorn: See, I get a bit like Big Brotherish. Okay, it can go like a little bit.

46:39 Emma Sjöström: so, I'm trying to

46:39 Tom Thorn: far for sure, but It's certain point maybe at some point you relax back off of that when everything's working fine. But if you can get to that level of understanding that those kind of keep forms indicators so, okay, like wait, wait, it cost to capital and then the percentages like accuracy of certain forecast lines, you know, we're going

46:47 Emma Sjöström: little bit.

46:47 Tom Thorn: hope payroll. The salary was great, but the payroll taxes off, or with, you know, whatever it might be, pensions way off the first, this certain entity. If you can get that level of granularity feed that back. Yeah, that's that's a good one. Yeah.

46:52 Emma Sjöström: I guess, I mean, I feel like we

46:52 Tom Thorn: Yeah.

46:54 Emma Sjöström: actually a lot like if there anything that you feel we haven't yet covered like any challenges or inefficiencies just like

47:04 Tom Thorn: I guess maybe one thing I, I don't

47:05 Emma Sjöström: um,

47:12 Tom Thorn: know if we've covered it yet, but we obviously focused a lot on the transactional side of things. What

47:18 Emma Sjöström: Yes.

47:18 Tom Thorn: also quite helpful. For me from a Treasury side of things is you take all of that information. We then still need to make the decision on the funding site. So actually what we're gonna do, so we have balances, we have deposits. We have other types of liquidity movements that are going to impact that forecasting, deposit maturities or overnight account, balances or money, market funds, all these various different Treasury acting.

47:42 Tom Thorn: So actually layering those back on top of the forecast, to say, if we need to fund this account, Where should we be funding it from? Because the the other part to that and this is maybe going. I don't know if it's going a little bit further in the forecast itself but it's like you might have a Treasury policy.

48:01 Tom Thorn: Right, this is one of the things I

48:01 Emma Sjöström: All right.

48:02 Tom Thorn: struggle with At the moment is I have a Treasury policy to say we should have x amount of funds available at any one time and then the then you have say 30% then set like 60% to mature over one month period and then the rest over at three month, period.

48:21 Tom Thorn: So I'm managing my balance across our investments. So we're cash-rich from from the last seed round and whatever. So we have a lot of cash that I'm trying to manage within this Treasury policy. But I'm also managing those liquidations and those maturities of those of those funds of those investments in line with our Needs within our transaction bank account.

48:43 Tom Thorn: So, if there's a way to layer that answers like in terms of debate, essentially, what I'm getting us to have some like scenario planning on it. Like, if there's a way that you can say, Oh, what if I take XM out from this this investment and then X like why am out from this investment? How does that then filter across to my Treasury policy? At the end of the month? I'm going to be lying in my mind with it.

49:05 Tom Thorn: That's a. Challenge that that I I don't know if there would be probably it's not something you considered at the moment. I don't know if that's even a problem for other Treasury departments but you have the forecast bit, but if you have a choice, If you're then moving around your Treasury, balances your investments and your your other cash sources to meet those track, those those funding requirements.

49:30 Tom Thorn: If I do it the way I'm planning to do it. Where do I end up at the end of the month? In terms of the rest? All of my investments, what we like, where what stack going to look like and actually am I compliant? Am I gonna be I've got too many long-term investments.

49:45 Tom Thorn: I've got too much in GB like Sterling and USD versus Euro. Those kind of questions.

49:54 Emma Sjöström: Now, that's interesting actually. Yeah. Yeah no just there's definitely been. I know talks we're not focusing on it currently.

50:10 Tom Thorn: No, no, that's this complicated one.

50:13 Emma Sjöström: How could a policy be used also in some in the context of an AI and how could it be applied on like sorts of things? It's really interesting stuff.

50:26 Tom Thorn: Hmm.

50:26 Emma Sjöström: Thank you for bringing that up. so, I'm just gonna keep probing you because this is nice. Sure, we're getting out. What's so if you could change only one part of the cash for casting process, you have today, what would it be?

50:44 Tom Thorn: To automate it. That's a very yeah, very honest. But it's that, that would be the best. Like, I mean, that that takes that's probably multiple different parts but as a headline, it would be to automate it. Yeah. And if that would be to automatically, bring through some reports from our ERP system, And have everything automatically populate it doesn't okay.

51:06 Tom Thorn: Not even in Elizabeth, it have things refreshed and automated way. That would be great but the best thing would be. Yeah. When I say automation that systemization, for sure will be a big help.

51:18 Emma Sjöström: yeah, that makes a lot of sense and that the biggest impact of that would be I'm you I'm not gonna say, so you can tell me just to make sure

51:27 Tom Thorn: Yeah. Yeah, it would be visibility of all of our transaction castle, more certainty, with the flows, and how they get, how they get tracked that brings efficiency, and should reduce risk. Yeah.

51:40 Emma Sjöström: Yeah. Very nice. Yeah. I feel like I've missed something probably, but yeah, I guess it we could dig slightly a bit more into like maybe or stakeholders if you go into like your management team or your CFO or whomever is like at the receiving end of then, like when it comes to reporting time or like, if you

52:14 Tom Thorn: Yep.

52:15 Emma Sjöström: discuss your forecast continues, like everyone has a different process, right? But like

52:19 Tom Thorn: Yeah.

52:22 Emma Sjöström: Yeah, how could we help you? Trust in the forecast that you present to them, do you know what I mean? Like I'm sure that it a lot but

52:29 Tom Thorn: Yeah. So well, at least the persona with the moment they the management team tend to focus more on the budget forecast, so that got more on that

52:39 Emma Sjöström: Right.

52:40 Tom Thorn: monthly level. There's all types of if like detail going into that. But they tend to focus on that as a From an accuracy one from a long-term use so that's every year there's that their forecasting out for a full year and actually beyond I think that this year was like five year forecast or something like that.

53:21 Tom Thorn: So they're relying more on that and then updates, you know, on a periodic basis in terms of, in terms of their, in terms of performance, in terms of how we're tracking, again, plan, all that kind of thing. When it comes to the transactional level forecasting, I tend to feel that that's less.

53:17 Tom Thorn: Report. So, In terms of the forecast accuracy. I would say that. Yeah, it's an interesting one. When it comes to the transaction level forecast, that the treasury-based forecast, the one that I'm interested in with the granularity of the split of various different accounts, all this kind of thing, the management.

53:33 Tom Thorn: My finances are my financial direct. She's not gonna care about that. Like, in

53:38 Emma Sjöström: oh,

53:38 Tom Thorn: of like, she's not gonna necessarily gonna care of the transaction level stuff.

53:43 Emma Sjöström: No.

53:43 Tom Thorn: But she would care on the impact that it has on cash and how we performed in terms of interest over the course of the month and where our balances are. So it's like a knock on impact,

53:54 Emma Sjöström: Yeah.

53:55 Tom Thorn: So I would almost see like, the transaction level forecast and our accuracy against that. The KPIs is more of like a sideways conversation with the accounting team. Whereas the there's there's two different types of reporting. There's like the variance in KPIs on that side but then actually when we look at our cash investments over the course of the month, I would probably have it in there as a comment to say our transaction level forecast was crap this month for export, for whatever reason based on our conversations with our accounting team or whoever which led us holding excess balances that we weren't able to invest, so it almost be like a common tree, a commentary piece to the financial.

00:00 :

54:32 Emma Sjöström: but,

54:32 Tom Thorn: director, just to let her know that we're struggling with our transactional level forecasting, inputs are meeting expectations and but the ownership is on asking the accounting team to improve that.

54:44 Emma Sjöström: Right? So it would be just around like really providing that added content and helping to quickly.

54:50 Tom Thorn: Ha yeah, yeah.

54:53 Emma Sjöström: get to the bottom of like if there was any major events going, all

54:57 Tom Thorn: Yeah.

54:57 Emma Sjöström: anything like that? Yeah.

54:58 Tom Thorn: Yeah. she because the at the moment, at least a lot of the more detailed stuff to say if we had When we come to month, then she'll be looking at things at a category level to say Personal costs supplied costs. You know, this kind of said she would look at it a very like customer inflows at a very very high level and she will just ask.

55:33 Tom Thorn: Why will we down on our cash inflows? Why we offer on supplied? So you she's looking at like one or two bits like high level commentary on why that was missed at the budget level when it comes to what I'm having challenges in terms of our caching investments. Which is that bit that I own it's more to do with that transactional.

55:36 Tom Thorn: Forward letting very granular forecast which is having an impact on how much interest I'm generating or if we miss it or if we went into an overdraft position on one of our bank accounts, that's when she's going to start to get interested. I would say. So it

55:49 Emma Sjöström: Right.

55:50 Tom Thorn: depends. Yeah. I don't know if I've answered. That's so clearly but it depends. It's probably And company dependent, but at the moment, that's at least how how things are working here.

56:02 Emma Sjöström: Very cool, cool.

56:02 Tom Thorn: No.

56:04 Emma Sjöström: All right. I think I'm very happy. I've gotten a lot more stuff to work with and insights and so amazing. Have you have you seen a product by

56:16 Tom Thorn: Yeah, no I have yeah I bumped it with

56:18 Emma Sjöström: the way?

56:21 Tom Thorn: Gidget a couple of times but I've been speaking with him like on and off throughout the year because I

56:26 Emma Sjöström: Yeah.

56:27 Tom Thorn: would like to I'd love to to look at it and, and go through like a trial run or something like that, but it's just difficult to. Yeah, I mentioned before, I prioritization across the business and where things are at, it's like a look, it's a, it's an upward strug.

00:00 :

56:41 Emma Sjöström: Yeah.

56:42 Tom Thorn: to get people to See. Yeah, I mean to sit down if I was able to sit down and have this conversation with say my and include like my Say for example, my manager up my financial director, but to ask these questions, I would hope that they would understand a little bit more of that, but it's difficult to get it up the list where they have a thousand different items that they're looking at from their southerners.

57:05 Tom Thorn: So I'm hoping for next year at some point, but we'll see. Yeah.

57:11 Emma Sjöström: No, no pressure. It's just like

57:14 Tom Thorn: I I have seen this system though, I'm really it looks good and yeah.

57:17 Emma Sjöström: What it would be cool to get your

57:18 Tom Thorn: You've got.

57:19 Emma Sjöström: feedback like on what actually

57:20 Tom Thorn: Yeah.

57:20 Emma Sjöström: building at some point. Then, all

57:22 Tom Thorn: No, for sure.

57:23 Emma Sjöström: at some point. I'm sure. I'm hope we can get

57:25 Tom Thorn: Yeah, and

57:26 Emma Sjöström: Yeah.

57:27 Tom Thorn: I I happy happy to do that even if it's not with a view to kind of implementation, I hope it is but if not that I still be happy to look at how how things are developing for sure.

57:36 Emma Sjöström: Super cool. Thank you so much your

57:37 Tom Thorn: Yeah. Oh, no problem. Thanks Emma.

57:39 Emma Sjöström: time. Have a great night.

57:42 Tom Thorn: You too you too. Hey.

57:43 Emma Sjöström: Thanks. Bye.