Skip to content

On - Categorization Outcomes Review - 2024-10-31

Metadata

  • Date: 2024-10-31
  • Company: On
  • External Participants: Amanda Mitt (Treasury Analyst), Lucía Galán Cáceres (Team Lead), Sofia Govoni (Treasury), Rodrigo Cabrera (Treasury)
  • Palm Participants: Emma Sjöström, Gurjit Pannu, Christian Sobkowski
  • Type: Customer Call
  • Domain Areas: Categorization, Cash Forecasting, Cash Visibility, IC Activity
  • Recording: https://tldv.io/app/meetings/6723a9a5666f3d00137bd590/

Summary

Context

Customer call with On's treasury team to review categorization outcomes on their Swiss accounts. Palm team presented a "categorization handbook" showing LLM-based categorization results on historical transaction data, with the goal of getting sign-off to go live before end of November.

Key Discussion Points

  • Product update: New cash positioning feature now live, reporting improvements coming
  • LLM-based categorization approach vs rule-based (easier to maintain, works across languages/geographies)
  • Walked through categorization breakdown for account ending in 1901 (most active account)
  • Intercompany makes up 75%+ of volume (absolute amounts, in+out combined)
  • Key distinction needed: Intercompany (between entities) vs Intra-company (ZBA/book transfers within same entity)
  • Opex categories (inventory, admin expense, payables) have most room for improvement
  • Collections (DTC, DTCR, wholesale) relatively well-categorized
  • Unknown bucket has only ~3 recurring descriptions - easy to resolve

Pain Points

  • Distinguishing intercompany from intra-company cash pooling (ZBA transfers) - different reporting needs
  • Opex categories can overlap without additional business context
  • Wholesale transactions can look like refunds
  • Rule-based categorization in Kyriba requires extensive manual setup and maintenance

Feature Requests & Needs

  • Access to Palm tool to start learning in parallel with data review
  • Forward all bank statements once Kyriba integration is live (next week)
  • Entity-level and group-level cash visibility across all accounts

Jobs & Desired Outcomes

Job: Categorize all bank transactions accurately for cash positioning and forecasting

Desired Outcomes: - Minimize the time spent manually categorizing transactions - Increase the accuracy of transaction categorization to 90%+ on notional value - Reduce the frequency of unknown/uncategorized transactions

Job: Get complete cash visibility across all accounts at entity and group level

Desired Outcomes: - Minimize the time to understand cash position across all accounts - Increase confidence in the accuracy of categorized data for reporting

Domain Insights

  • Categorization philosophy: Focus on getting biggest categories right (90% notional value), can refine smaller categories over time
  • Intercompany vs intra-company: ZBA/book transfers within same entity shouldn't be treated same as true intercompany flows
  • Completeness is key: Lucia wants every transaction categorized for filtering purposes, even if some categories net to zero at entity level
  • Benchmarking approach: Compare Palm results against SKG rules mapped to GL accounts and budget codes

Action Items

  • [ ] On to share intercompany "cheat sheet" (account list to identify IC vs intra-company)
  • [ ] On to share opex categorization rules (budget codes, cash flow codes)
  • [ ] Palm to send full categorization handbook and data set
  • [ ] Palm to provide tool access for On team
  • [ ] On to spot-check wholesale transactions for potential refund misclassification
  • [ ] On to confirm no recurring payroll on these accounts

Notable Quotes

"I don't think I expect you to be much better than Kyriba. So I don't know if I want to take the rule-based categorization as the source of truth" - Lucía Galán Cáceres

"For me, when I think of categorization, I want completeness. So I need all of the transactions categorized" - Lucía Galán Cáceres

"The goal here is not to have you go through every single transaction and mark it as incorrect or correct. We want to get kind of that 90% overall on the notional value" - Gurjit Pannu

Full Transcript

00:01 Sofia Govoni: oh,

00:01 Amanda Mitt: I didn't get it. Sorry. Hi, everyone. Sorry.

00:04 Christian Sobkowski: Hi.

00:05 Amanda Mitt: How are you?

00:06 Gurjit Pannu: Yeah, good good.

00:08 Emma Sjöström: Pretty good.

00:09 Gurjit Pannu: To see you all again.

00:12 Amanda Mitt: Good to see you. Are you all in the US?

00:14 Emma Sjöström: No.

00:15 Gurjit Pannu: Christian and I are if we both lean over you could see that we're probably we're in the same exact hotel with just different rooms.

00:23 Sofia Govoni: yeah, there is the dog, the red

00:24 Gurjit Pannu: The dog. Yeah. So we're in San Francisco this week. It's been really interesting actually, really, really good sessions that we've had. And yeah, we're flying back this weekend. I'll be back in back in the good time zone by next week.

00:41 Sofia Govoni: Now, a knife.

00:44 Emma Sjöström: Yeah.

00:45 Sofia Govoni: And it was either the conference that you attended, or

00:47 Gurjit Pannu: Yeah. Yeah, so last week, we read a conference, you'll hear about this conference in Europe, called Euro Finance. We did the equivalent of that in the

00:56 Sofia Govoni: What?

00:58 Gurjit Pannu: US called AFP. So it's a big Treasury conference and once a year, so their last week for a couple of days and then this week we're in San Francisco. We had a few customer meetings. Demos investor meetings. Kind of fact schedule. But it's been really interesting. We were talking to Some I think are becoming household names in the AI space and it was surprising to see like the lack of resources they have on the AI front for the stuff that they do.

01:29 Gurjit Pannu: And so they're looking for tooling to support that has AI. And we're kind of like, at the end of it was kind of like Are you using X Y and Z. Like we're kind of using the other one, but we're using yours as well internally and it's it's really funny to kind of see but it's yeah, it's been really good, really good, really good time here in terms of productivity and meeting some right? Some cool folks.

00:00 :

01:52 Sofia Govoni: Now, that's nice.

01:53 Gurjit Pannu: No.

01:54 Amanda Mitt: I think.

01:55 Gurjit Pannu: Cool. I think I know I know Lucia it was gonna join, but maybe we just kick off for recording the sessions.

02:00 Lucía Galán Cáceres: I am here.

02:00 Amanda Mitt: I'm actually trying. Yeah.

02:02 Lucía Galán Cáceres: Yeah, I'm sorry.

02:03 Christian Sobkowski: I would do.

02:04 Lucía Galán Cáceres: I'm

02:05 Gurjit Pannu: Many laptops, I don't see, see everybody.

02:06 Lucía Galán Cáceres: No, that's good. Actually. Hang on. Give me a second and those six that issue and you can see where I am.

02:14 Gurjit Pannu: How. There you go.

02:17 Lucía Galán Cáceres: so, joining for my phone, because I'm

02:17 Gurjit Pannu: Good.

02:18 Lucía Galán Cáceres: at the airport.

02:18 Sofia Govoni: but,

02:19 Lucía Galán Cáceres: So at some point I'll just have my phone and go to the gate

02:21 Gurjit Pannu: Okay, cool. Sounds good.

02:23 Lucía Galán Cáceres: Thank you.

02:23 Gurjit Pannu: And we're gonna record this session. It'll be on the notion page for you to watch again, if you need to. So, so all good and cool team.

02:28 Lucía Galán Cáceres: Great. Thank you.

02:32 Gurjit Pannu: Thank you so much for the time as I mentioned in the, in the stock message, we're making some really great progress across categorization, we're thinking we're in a pretty good place now or we're Us to wrapping it up to where, now, we can start thinking about the forecasting piece on our path to going live with with these accounts by the end of end of before the end of November.

02:54 Gurjit Pannu: And with that, once we're there, then I think we're at the bit, we have the opportunity to then take everything else in as well. So, really excited for that for this session, Christian's gonna drive, and kind of share what we've done, what our expectations are, some of the outcomes, and then next steps and on this path towards towards going live.

03:14 Gurjit Pannu: And, and so, if there's any questions, feel free to feel free to hop in. And otherwise, I will hand this over to Christian.

03:25 Christian Sobkowski: Okay. Cool. Actually before before even going into into categorization of the data, maybe quick product update from our site and Emma and team have been Doing amazing stuff over the last few. Few days where We've we've just gone live with our new cash positioning feature. So as as we're loading all this data from you into the platform, you will see daily cash positioning recommendations in the platform.

03:56 Christian Sobkowski: Go live. Um and there's also a ton happening on the reporting side. So super, Super excited, that sort of asware sequencing, this around calculation first and getting the forecast done and then diving into someone's reporting for you. There's a ton to explore a ton to configure now and yeah, really looking forward to to basically all of this now is making sure that the data is in the platform.

04:27 Christian Sobkowski: The data is in a state where you're happy with it and we're actually represents your business and then I think all the fun begins and it becomes actually usable That sort of as a little start and kick off. Today is about categorization, which is a bit of I think fundamental for for Treasury platforms to function.

04:56 Christian Sobkowski: And to, to give context on what happened on your accounts as well as, you know, we're using a lot of this to look forward and forecast. and across these categories and then there's obviously reporting element on all of us where this funnels through directly into the reporting platform and you can for example a pretty detailed reports on some of these categories if you for example want To say, attack support across certain accounts.

05:34 Christian Sobkowski: Maybe starting now. So let me let me I need to quickly switch over screens. What what we wanted to do today is Al I'll quickly go through three categorization, but then it's really about showing showing updated outcomes on the data you've sent. You will recognize a lot of this actually from from the proof of concept.

06:00 Christian Sobkowski: We've run some time ago and really now it's about going through it in a bit more detail. And really, as the next step, capturing your feedback, right? We need to we need to go through one more round. And so I'm taking this session as a bit as a kickoff for us to To hand over and to start figuring out.

06:28 Christian Sobkowski: How is feedback coming back? And then sort of going live once signed off from from all of you. Does that sound fair? Amazing. I've seen Notting heads, very good. So. A quick refresher on categorization, I think maybe, maybe Sofia for you as well and the The ideas. We're, we're heavily heavily leaning on, on LMS to the categorization.

07:02 Christian Sobkowski: And the idea is that that means it's a lot more easier to to maintain. It's a lot easier whenever you switch accounts or open new accounts or go into new geographies and it should make things such as especially when we're looking at some of your operation expenditure, a lot easier than you would find with.

07:24 Christian Sobkowski: A rule space approach. That, that doesn't mean that there isn't a lot of engineering behind it. That is pretty and post-processing processing data view. The core is Is these prompts and what what we're basing this on so far is the onboarding session. We had now a couple weeks ago by now, where Amanda Rodrigo were walking us through the Swiss Swiss accounts and sort of the patterns you were expecting on those accounts as well as obviously basing it on the list of 16 categories that you send us.

08:14 Christian Sobkowski: Everything. Good.

08:15 Gurjit Pannu: Yes. Actually, have you all started doing the mapping rules in kairiba for the transactions? Are you still waiting for the bank statements? The first start coming through

08:27 Lucía Galán Cáceres: Yes. We have parallel Emma University.

08:32 Gurjit Pannu: Sorry, I missed.

08:32 Amanda Mitt: So yeah, go ahead. We did. Yeah, we started

08:35 Gurjit Pannu: Okay cool. Is it live already or is it still kind of being worked on?

08:44 Amanda Mitt: Um, yeah.

08:45 Lucía Galán Cáceres: I think it depends.

08:49 Amanda Mitt: we have the rules, but we're still kind of Rodrigo is working a lot on kind of improving them a little bit.

08:56 Gurjit Pannu: Okay.

08:57 Amanda Mitt: Because we didn't go live yet with the bank statement integration. So we have the rules that we use for skg.

09:04 Gurjit Pannu: Got it, okay.

09:05 Amanda Mitt: So we're like connecting now with our kind of Treasury codes trying to make this link.

09:12 Gurjit Pannu: Okay.

09:12 Amanda Mitt: but so yeah, that part is still a working progress, but we have like

09:15 Gurjit Pannu: Got it.

09:16 Amanda Mitt: the bank reconciliation, we have that ready.

09:19 Gurjit Pannu: Okay, cool. Thanks, super helpful. Just just curious and then one and something about more like potentially comparisons, but anyways just a question there. So, okay cool, thank you.

09:29 Amanda Mitt: With the rule for the forecast now as well mapped. So because we have to map for the statements for the forecast and like for the kind of Yeah, budget codes that we assigned. So I can share that with you as well if you want some Yeah.

09:47 Gurjit Pannu: That's cool. Thanks.

09:49 Christian Sobkowski: I was about to say, if, if you have known, correct. Categories. That's that's obviously a super helpful benchmark to to compare.

10:01 Amanda Mitt: Yeah, we do. Okay. Not many, but I can open them here so we can compare. Okay.

10:17 Christian Sobkowski: Got it, okay. Let's let's, are you opening it now? Are you, Is this an action point?

10:26 Gurjit Pannu: I think we take this as action point after the call, right? So like once

10:29 Amanda Mitt: Yeah.

10:30 Christian Sobkowski: Good.

10:30 Gurjit Pannu: results so I think we can keep moving forward Christian.

10:33 Christian Sobkowski: Amazing. Let's let's have that ass in the first action point. I think it's right with with this whenever you have something already where you know things are correct that's and always super interesting. Yeah. Benchmark to to figure out How is the performance? Go going in and actually, so I'll walk you through some of the results.

11:04 Christian Sobkowski: And really, what, what this light decades is we're seeing it as a bit of a handbook on the work that the system in the background does. And so there's a fair bit of data in these coming slides. And I'm not expecting to to go through everything on this call.

11:24 Christian Sobkowski: But as you're as you're going live with this, it's probably a good reference to go back. To in case you're looking at And looking to make sense of things. And with that. So what we've prepared for this call, as we've prepared, the or I'm looking to, to go through the, the Handbook for one account, the one ending in 1901 are Which right is your most, most active account of the ones you've sent over.

12:05 Christian Sobkowski: And and really showing going through some some of the, the thinking and the philosophy. That's underlying this What you're seeing here is, you're seeing an overview of the categories identified on this account. And really what would you seeing is on the left? You see a pie chart where you're seeing overall amounts in in a time span we what's here represented is May 23 to December 23.

12:42 Christian Sobkowski: So you're seeing 840 million Swiss francs in into company payments and this is plus and minus our netting. Netting out. But there adding to each other So this is absolute amounts. so, you seeing in terms of volume, Over 75% is in the company. and then you're seeing a pretty pretty chunky operational, expenditure payments, and then A collections is the next one.

13:15 Christian Sobkowski: And then you're seeing smaller smaller categories around tax and payroll. There's not not a ton of activity Um, and some of the unknowns we can go in later. This this is the same, same type of transactions. We already saw in the in the proof of concept. So really there's like two or three transactions that you will relatively easily identify and we can code it.

13:45 Christian Sobkowski: but basically on the right, you then seeing the table in In a really detailed manner of what is right, the number of transactions. How does that stack up in terms of percentage across the account? It's an emotional amount of each category and again, how does this stack up in terms of percentage? so really when the the way the way I like to think about this is And anything that where you're seeing a notional amount percent of Percentage of 1% or less.

14:25 Christian Sobkowski: Really, it's making very, very little impact on your On your understanding of what's happening on the account, from a cash, balance's perspective, right? And really it's it's a long tale of transactions where there may be some reporting merit and there's definitely counting merit in getting them, right? but in terms of positioning and forecasting, They are.

14:55 Christian Sobkowski: Not the ones that are the the important bits to to get right to to manage your cash balances. So that's a bit of and I like to use this overview key as a bit of a guide to what activity are we seeing in the account? And where, where the important things to? Zoom in on.

15:20 Christian Sobkowski: Does this make sense so far? Any questions?

15:25 Amanda Mitt: Yes. I just want to refresh everyone and say that this is from the last two years, right? So it's a total from Basically. 20, I think full 2023 until half of 2020 like yeah, October of 2024, I think when we shared so I just wanted to share this with Lucy and Sofia so they just, yeah,

15:46 Lucía Galán Cáceres: I actually want to ask you, sorry,

15:46 Christian Sobkowski: The.

15:49 Lucía Galán Cáceres: interrupting. Just wanted to ask internally, maybe that we don't have to answer now, but does it make sense? In terms of what we see in the account to see so much intercompany? I can't remember exactly which account this one, now refers to with the Ivan, but this could be a quick sense checks is.

16:05 Lucía Galán Cáceres: It's so much intercompany It should be something that makes sense we can check with knowledge with Casper and so on.

16:12 Amanda Mitt: Mm-hmm.

16:13 Lucía Galán Cáceres: Because I think that's gonna be a Bigger benchmark than comparing to kariba to be honest. Because I don't think I expect you to be much better than kiriba. So I don't know if I want to take the rule-based categorization as the source of truth, although it's good to check of course, but

16:34 Christian Sobkowski: That's really interesting. Great, good comment, lucía, I think. Yeah. So What? What we've included in the handbook. This this data is and again, I'll send you all of this afterwards in in written formats. We have it. But this is currently showing data from May 23 to December 23. happy to refresh that on the full data set where I think we have January 23 to April 24.

00:00 :

17:09 Gurjit Pannu: Yeah, and one other thing, I think to call that on this inner company, Pie chart is again, this is the notional absolute amount, so it's in and outgoing or added together, right? So, I'm not surprised that it makes a fair, at least this account, such a big piece of the the pie because every outflow or inflow, I would imagine that you have going on, you know, across your accounts.

17:32 Gurjit Pannu: And there's a lot of pooling activity that's happening. And so therefore, it's, it's making that number look that big I'm wondering something for us to take back, is How do you kind of maybe represent this chart to Make up for this kind of I don't say double counting but it's it's essentially Every flow is being added to a large making a larger inner company balance.

17:51 Gurjit Pannu: When I think the net net, inner company flows might not be as high as well.

17:58 Lucía Galán Cáceres: No, not for sure but still it's when we can check it. The account is really where all of the company payments go through just business. Check.

18:07 Christian Sobkowski: Great. Great sense. Check absolutely. And again, so so you will obviously write this is gonna live in the platform ultimately, but you're also getting getting an Excel sheet with all the underlying data. So you'll be able to slice entices on your own and that should make it relatively quick to sense check for example into competitions actions.

18:37 Christian Sobkowski: With that I'll actually move on and so. So the rest of the handbook is actually we're going into each of these. One, two, three, four, five categories. Again, see it as a guide for, for the data set, ultimately. And we're gonna start with with the big ones and then as we're running out of time, it's going to become less and less important.

19:06 Christian Sobkowski: Again, thinking about it, from a positioning, and forecasting Let's get the big ones out of the way perspective. first. You will see again. I'll take some time to walk through the slide and you will see this for every single one of the categories. So you will see this for intercompany, you will see this for operational expenditure collections, we've bucketed tax and payroll for you together, because they're both quite ir.

19:37 Christian Sobkowski: There on this account. and then there is a bit of a catch all of refunds and unknowns where We can talk a bit more when the time comes. Starting with Intercompany really? What would you see here on the left? You're seeing a zoom in from the previous slide across all of the four intercompany categories identified on the account.

20:06 Christian Sobkowski: What is the number of transactions? What is the notional amount across those transactions? And then then on the right hand side of that, you're seeing and A sample of the the largest transactions for each of the categories. Again, this is just a quick sense. Check. And I think here, here's already the first one, where Your, you probably want to want to double click into this one up here.

20:47 Christian Sobkowski: To to sense check, if, if that's actually an into companies transaction and or might be something else, it's relatively, it's comparatively small compared to the rest of the activity within two company. But Again, it's it's a quick guide to to go through these things. And then on the bottom, you're seeing two two bit of statistical measures to again, help you understand activity.

21:17 Christian Sobkowski: So on the one hand, you seeing a distribution, histogram? Where really what's what's important? This everything around zero. Doesn't make a whole lot of difference for for your cash balances. But any of the outliers. Those are probably the ones that are are helpful to sense. Check in this case.

21:46 Christian Sobkowski: you're again seeing the same thing on on a timeline basis, on the right hand side, which, I would imagine is probably a quick again since check in terms of visually. Does this make sense for in this case into company transactions? Which which you can double click into? When things feel off.

22:21 Christian Sobkowski: Does that make sense so far?

22:23 Amanda Mitt: Yeah, I'm just sorry I have a few questions. So this is from 2023 from that initial. that is that we shared or is this from the like Because I think what? I'm, maybe a little bit confused. It's because we have like, for example, this is I think it's the Yeah.

22:44 Amanda Mitt: Now we we use this account for basically Yet. Payables,

22:54 Christian Sobkowski: Mm-hmm.

22:54 Amanda Mitt: But no, because we have like, for example, we have a master chief account. We have available account, we have kind of like, a receivables account. So I think these three accounts that I shared, so I'm just thinking. How could I see how much is yeah? Like for example, when we see this zero balancing transfer This is IC Cash putting, but it's actually internally, like Intra company not between.

23:19 Amanda Mitt: I think. So, maybe I think this can enter a bit, what?

23:24 Gurjit Pannu: Got it.

23:25 Amanda Mitt: What Lucía said.

23:26 Gurjit Pannu: So, same entity just for that. You don't consider it your company.

23:29 Amanda Mitt: Yeah.

23:30 Gurjit Pannu: per se. It's just a it's actually like a

23:32 Amanda Mitt: Yeah.

23:32 Gurjit Pannu: trend. It's a book transfer more than a

23:34 Amanda Mitt: Yeah.

23:35 Gurjit Pannu: transfer.

23:36 Amanda Mitt: Yes, I think that's it. Actually, so this is like, actually the movement from the Participant to the Master.

23:46 Gurjit Pannu: Mm-hmm.

23:46 Amanda Mitt: Within the same.

23:48 Gurjit Pannu: With it within the same company. So you wouldn't.

23:50 Amanda Mitt: So this is on again, and it's all on again. So it's all inside on again.

23:55 Gurjit Pannu: What would you what?

23:55 Amanda Mitt: Sorry. Rodrigo, yeah.

23:56 Gurjit Pannu: What category would you label that as

23:57 Rodrigo Cabrera: Yeah.

23:59 Gurjit Pannu: from the list that you have?

24:04 Amanda Mitt: I don't think this would be a category because it's like a zero

24:11 Rodrigo Cabrera: All right.

24:13 Amanda Mitt: balance. so ideally, I think When I went to see. The casual, I wouldn't want to see. I think the cereal balancing like, you know, if it's so high level, I would want to see just what's flowing actually in or out. I don't know. Maybe this in a different bucket.

24:34 Amanda Mitt: Sorry, I let people.

24:34 Rodrigo Cabrera: I mean this. So I think in this case, as this is a I'm not sure if this is a cash, pull, the actual cash, pull master account. If it's the cash pull master then it can contain not only on a G2 on AG but also on a G2 on France, for example, or on as a year or chief.

00:00 :

24:55 Amanda Mitt: Okay.

24:58 Rodrigo Cabrera: Yeah. Well if it's CHF, then just on AG But with other currencies, it might be intercompany. Yes A I don't know if it will help if we provide that cheat sheet more or less just to okay, if The the code contains this account then it's within an idea or if we provide the list of all of our accounts will that help to identify, if it's intercompany, then we shouldn't include it or it will net out either way.

25:34 Rodrigo Cabrera: But if it's intercompany, then it should go to this. I see cash pulling In my other question, or comment. For example, the fourth one over there. It looks like and I see long term, but it's one of those that just by looking at the code. I know it's an interest payment if or an an interest outgoing payment, all also they like a Inter company cash pulling, but it's an interest payment and not like long-term financing.

26:06 Rodrigo Cabrera: So those are the ones. Like I don't know if it makes sense for us to share. or basically cheat or help the tool to work better by providing feedback or the current rules that we have for the tool to learn from them, a And basically perform better to avoid this mistakes from the beginning or just let it be and we fix as we see.

26:38 Rodrigo Cabrera: In the, In the tool.

26:42 Christian Sobkowski: This. I have response. I lucía has her hand race, so I'm gonna let her Speak first.

26:49 Lucía Galán Cáceres: Yeah. But go ahead because I need to pay for something so perfect timing, then I go.

26:51 Christian Sobkowski: Okay, amazing. Rodrigo apps. Absolutely. So this this is why exactly why we're doing these sessions? Right, we're ultimately. The. I I fully expect there to be Things here that we haven't learned about your business. Right? And that includes for example, seeing such as what what you identify as I see cash pulling and what you identifies, I see long term financing.

27:26 Christian Sobkowski: So if you have a cheat sheet here, absolutely helpful because then then we'll fix this. And it's it's gonna show up the right way. Um, in the tool.

27:35 Gurjit Pannu: Yeah, and I had one more question. This is me for you, Amanda. So, when we think about, Cash pooling transactions, right? Or sorry before going to cash between transactions. What we're thinking about in this categorization piece is around looking at every transaction that comes in the bank statement and putting a tag on a category tag on it, right? So to understand what, what is this is a pooling transaction? Is this a financing transaction opex or whatnot? It sounds like one of your questions was around like Well I don't want to see the pooling transactions on my kind of business activity because that's almost like a ghost transaction, right? So if you have a hundred and outflows, you have a hundred of incoming, zba, you don't want to see that, incoming zba in your report per se because that's not a true reflection of your business, right? So think, what's, what's important to keep in mind here in this exercise? This is more on the Of tagging.

28:32 Gurjit Pannu: The transactions in a way that we recognize what they are. There will be a next step around like well when it comes to the actual reporting of the activity, you want to actually remove all of the in the zba stuff, right? And and see kind of business activity, which you will be able to do That's one.

28:51 Gurjit Pannu: That's one and make sure that we're all the same page there. And then the second piece around Intercompany versus Intro company. I think it's important for us it, you know, after this call or maybe you can discuss a little bit on this call, is that understanding the purpose of differentiating between a cash pooling and an intercompany and intercompany pooling activity if you want to split that out, right? So do you want to see all of your cash flow activity? Which if you if you pull it out, it should all net to zero regardless of his intra or inter company.

29:26 Gurjit Pannu: Or is there an option to rather kind of be a little bit more granular and say? Well, I want to do all pooling but I also want to split it out between What's a book transfer like a intra versus inter, and if it's for reporting purpose or whatnot, Of course we could discuss and make sure we get it, right? So I think there's maybe a room for some, some clarity there for us to ensure that it is categorizing it in the granularity that you're looking for to ensure that it's meeting your kind of requirements around why you want to kind of see it that way.

00:00 :

29:52 Lucía Galán Cáceres: Yeah, and I what I wanted to say actually is that I agree with what I think Richard is thinking, I would actually want to see the intera company because the thing is for me, when I think of Casa Categorization, I want completeness. So I need all of the transactions categorized and this is, and it's a company cash pulling.

30:09 Lucía Galán Cáceres: It's just that, when we look at the entity level, we'll see zero. But if I look at the account level, I do want to see the cash out on the cushion because I might be reviewing if something went through. I mean, Amanda you're thinking of said Pa which always works, but what if it's a kiriba or semi-automated cooling? It's still an intra company cash pulling between Dancekebank and UBS, and this, I will want to check, right? So I would want to see every single category categorized, I still would argue that.

30:35 Lucía Galán Cáceres: This is in the company. Cash pulling. And I think it's just like Richard said, depending on the purpose of your report, you can't filter this out. But I think you will the moment that we look at the entity level, right? And if you want to look at the account level, then I do think it's important to have this as a crashing cash out, but we can we can check it if you want and then we see where what kind of report you have in mind where you would want to see this completely separately.

00:00 :

31:02 Amanda Mitt: But when you

31:02 Lucía Galán Cáceres: And that Rodrigo sorry. Yeah, I should go ahead.

31:04 Amanda Mitt: Sorry, you know, anything about it that when he said a budget codes or for example, when you're doing a cash flow forecast, you would want to see the deviate that because I think in

31:12 Lucía Galán Cáceres: Yeah.

31:12 Amanda Mitt: my mind the categories would be for maybe it's like two high level. Of course when they are.

31:18 Lucía Galán Cáceres: But I think we do this. Yeah.

31:20 Amanda Mitt: Yes. Okay, listening. But for I think this category initially thought they were from or something more like, Casual forecast, liquidity, learning something more like high level.

31:30 Lucía Galán Cáceres: I we can talk about it, I don't see.

31:30 Amanda Mitt: So, maybe here.

31:33 Lucía Galán Cáceres: the need for more colority because for me this is a category where you just want to be complete so that you can take it out to take the noise out of the account if you want to do analysis. But I don't want to necessarily see exactly here. If it's inter intra for that, I have the in-house bank.

31:50 Lucía Galán Cáceres: Account. So this is where I see if there was an intercompany and hit or not, and Because of how we, how we booked transactions there. But I can, let's let's check because you have something in mind and I think it's worth going until the end of that thought and seeing if we're missing something on the on the detail just to to answer Rodrigo on your point I mean we need to see what we expect this interest to be called.

32:15 Lucía Galán Cáceres: So again to me This interest for cash pulling is an intercom company financing interest, just a short term So I need to check, I don't remember the categories, but I don't know if we have anything more detailed than intercompany. Long-term financing. We don't have inter company. Long-term financing costs or anything like this, right?

32:34 Gurjit Pannu: Right. Yeah, yeah.

32:35 Lucía Galán Cáceres: Yeah.

32:35 Gurjit Pannu: He knows. One of the other takeaways that we had while we were working.

32:38 Lucía Galán Cáceres: Then I would keep it. Yeah.

32:40 Gurjit Pannu: there's, there's also an opportunity if you wish to kind of add more

32:43 Lucía Galán Cáceres: Hmm.

32:44 Gurjit Pannu: categories, if you want that grant

32:45 Lucía Galán Cáceres: Yeah.

32:46 Gurjit Pannu: party, right? And that's what I think this

32:47 Lucía Galán Cáceres: Yeah.

32:48 Gurjit Pannu: will help us. Do is here's what we're does.

32:50 Lucía Galán Cáceres: Absolutely.

32:52 Gurjit Pannu: This make sense, or is there? Opportunity to add more or reduce based on what you're seeing from the

32:56 Lucía Galán Cáceres: yeah, and then

32:57 Gurjit Pannu: results.

32:58 Lucía Galán Cáceres: add more category so it really see what's the purpose of this. Because if the purpose is to identify how much we paid in interest, I think we have reports for it in the in-house bank and those need to be correct because those will go to accounting. Here, I want to see what we use this categories for.

33:16 Lucía Galán Cáceres: And then we see if we want to be able to double click over refine with this high level.

33:21 Gurjit Pannu: Yeah.

33:21 Lucía Galán Cáceres: That would be, that would be my my opinion. But again, Amanda Rodrigo, let's let's check this internally, I think it's worth checking.

33:29 Rodrigo Cabrera: Yeah. Yeah.

33:38 Christian Sobkowski: Amazing. And look, I think this, this was actually really, really good conversation. I think both in terms of Feedback for us right there. There's clearly there's clearly documentation you already have in terms of, right? What do you want to see us what? And then hopefully it's it's a little bit or it sounds like it's a little bit of Raise the question mark internally that you can follow up on and get some some clarity around what's happening on your accounts.

34:17 Christian Sobkowski: I'll keep going for now, but this is definitely right. This is this is one of the largest Notional amounts of transactions you're seeing on the account. So let's definitely have it as a follow-up and make sure that this is absolutely correct correct. Because really, it's not a whole lot of transactions and it's they all look very similar.

34:38 Christian Sobkowski: So we should be able to get this 100% right with one more iteration. I'll move on from into company unless there's anything else. I will move on. And I think this, this is actually probably my, my favorite group for, for any customer. Because what would you see here now is operational expenditure? so the the things that popped up in the on the account were inventory, other administrative expenses and payables and and I am actually convinced that in terms of improvements this is probably the is a bit that Has the biggest room.

35:36 Christian Sobkowski: For us to improve the model on. And I'll tell you why in a second. so right here, you seeing You're seeing the same thing again you seeing how much is happening across each of the categories you seeing the outliers, you see the timeline which which is already quite noisy.

36:03 Christian Sobkowski: And and you're seeing some of the examples, you seeing some transportation shipping under inventory, you're seeing some of the non-descriptions actions under admin expense. This is for example, an outlier. I would definitely look at that. I wouldn't be surprised to that something completely different. And, and you're seeing, what I would imagine is large marketing spend under on the tables.

36:34 Christian Sobkowski: So, where's the opportunity here? From, from the first onboarding session, we've, we've taken in some, some business context around. What is what? but really, the way, this one can be improved is to For us to become very granular, on what type of activity. Are you classifying under inventory? What type of activity? Are you classifying under Admin Expense versus Payables? Right.

37:05 Christian Sobkowski: Because in a S As categories as such they are. They can be quite overlapping without additional context. So really, the follow-up from from our side here for you is going through. A adding more context from your side and then be finding ideally finding examples that are representative for this for, for example, what an Act admin expense looks like for you.

37:44 Christian Sobkowski: A briefly pause. I'm not sure that was fully. Fully coherent.

38:01 Amanda Mitt: Yeah, I'm not sure what.

38:01 Lucía Galán Cáceres: I think it was

38:03 Amanda Mitt: Yeah.

38:05 Lucía Galán Cáceres: and actually, I think here basically, you could Be a good comparison because I think here and Rodrigo's where you probably have developed the SPD rules really, really well because you need them for custard consideration, right? So, I would expect that here. We do have something to benchmark. You on.

38:22 Lucía Galán Cáceres: And then if it doesn't fit, then this is what we need to provide edition context as to why. So that's something we can check.

38:29 Christian Sobkowski: That's amazing. So Rodrigo you have for this for this data set, you have Correct. Categories, across inventory, admin expense, and tables.

38:44 Rodrigo Cabrera: Hmm, not for every transaction but which will have a for quite a lot. I will share with you the same that we shared with Kiriva. With the budget goes and cash flow codes. And in the rules that should be assigned to each of them. And then, yeah, it can be a great support to compare the performance of this.

39:11 Rodrigo Cabrera: A locations and to build upon that. The logic for, for Going forward.

39:21 Christian Sobkowski: Fantastic. Look, I think. We we have two fantastic follow-ups across Intercompany and Opex. Now that should right being the the two largest. Categories on these accounts, should should give very, very good results in in one more iteration. Out of interest. How do you distinguish inventory, admin expense, and payables

39:57 Rodrigo Cabrera: A. By the vendor basically, and the description of each payment after a while we Get to find a pattern in the, in the payment description or methods, and then we create rules for that a rec

40:13 Christian Sobkowski: Yeah.

40:14 Rodrigo Cabrera: recently. We implemented a Nexus as a platform for all the All the factory payments or inventory payments and those have a really standard description. So, going forward, the inventory payments will be really easy to To identify a. but for the rest is just the understanding, how they pay or how we pay them a In the great rules based on that.

00:00 :

40:48 Christian Sobkowski: Fantastic inventory makes sense. So, so the difference between admin expense and payables is the method of payment,

40:55 Rodrigo Cabrera: And honestly, I'm not really sure.

41:01 Christian Sobkowski: Okay.

41:03 Rodrigo Cabrera: Yeah. I don't know what we consider as other admin expenses. Lucía Mendez, Sofia. A.

41:15 Lucía Galán Cáceres: Hey, but I think the answer is in the sheet that we built together Rodrigo's. So which jail accounts did? We assign to to admin expenses and then we see what's what's behind there? So that's what I would do. To be honest, I would, I would take pounds outcome and then compare it to So when we look at the categorization, we compare it to what we would have expected looking at the transaction.

41:42 Lucía Galán Cáceres: Looking at the applicable skg rule to which jail account, is it sending it to? And then we compare that to you like around. We have already done mapping of TL, account to budget, flow, budget code, right? So this is what we can verify. That connection was already.

41:54 Christian Sobkowski: Fantastic. Straight.

41:56 Lucía Galán Cáceres: Done. We can close the circle. We just have to do lots of vlookups. That's fine.

42:04 Christian Sobkowski: we'll, We're happy to use for you.

42:13 Lucía Galán Cáceres: Sorry, trying to thumbs up. Not the hands up.

42:15 Christian Sobkowski: That's okay.

42:16 Lucía Galán Cáceres: Thank you.

42:19 Christian Sobkowski: And good look we we have around 15 minutes left. I think now things are getting smaller so I'm gonna speed up a little Just so we get to the next I think for slides and then you can quickly summarize the call sounds okay. Go going. Forward collections, obviously, right.

42:47 Christian Sobkowski: This this is the other. Important one for you on this account. This this one, I think we were honestly pretty successful within a the onboarding call of defining. What is dtc? What does dtcr? What is wholesale? and and then I think also obviously you were You're labeling or how you set up your.

43:17 Christian Sobkowski: Your payment providers is giving a lot of context on where things be belong and which which store they're coming from or which website they are, they're coming from. So I'm I'm relatively confident that we are. Quite accurate on this one. Should be again, relatively simple to to spot check.

43:41 Christian Sobkowski: I think the one thing that requires a bit of attention from you is the wholesale category. where, I have a feeling that the model might have mistaken. Some refunds for wholesale sales. and I think that's, that's one of the tricky bits that that you're gonna see with your cash, in cash out, differentiation Where some? some large refunds will look very, very close to wholesale revenue to A larger chain.

44:26 Christian Sobkowski: And so, I think, What would I would suggest here? And again, I'll send you. This is for you to have a look at the largest whole search transactions. And double check that there's nothing in there that really, you know, might be something else. I think you're seeing it here there is it mentioned See, Dzb bank you.

44:50 Christian Sobkowski: We will you will probably know. Whether that's actually a whole sentence action whether that's some other incoming flow funds.

45:05 Amanda Mitt: I have no idea actually.

45:10 Christian Sobkowski: Okay.

45:12 Amanda Mitt: Let's check.

45:14 Christian Sobkowski: So yeah. So I think these are the things. This is also part of the, the feedback loop around How important it becomes for you to get this, right? And how important it becomes for you to be 100% correct on wholesale. Or whether there's some leeway, because ultimately, in terms of cash positions, it probably doesn't make a whole lot of difference, except for a handful of transactions that would get service to you anyway.

00:00 :

45:45 Rodrigo Cabrera: The Christian story, Like I lost my connection. Can you repeat the that last question?

45:53 Christian Sobkowski: And we were talking about their, Some of the whole search transactions look will look suspiciously like refunds.

46:03 Rodrigo Cabrera: Hmm. Yeah.

46:04 Christian Sobkowski: For other incoming, right?

46:06 Rodrigo Cabrera: Yeah.

46:07 Christian Sobkowski: Most most commonly, refunds in some form.

46:09 Rodrigo Cabrera: Yes.

46:11 Christian Sobkowski: And I think the the way I would approach it is for for you to spot check the largest wholesale transactions. Or again, if you have a true data set, obviously we can compare. But so for example here, you're seeing the largest whole search transaction identified was from DCP Bank, which might well be something different.

00:00 :

46:37 Rodrigo Cabrera: Yeah. Okay. Yeah. And they. Just curiosity. For the refunds, if by any chance, the description of the refund comes with a bender name or a vendor code. The the totally able to identify all. This is a vendor name. It's probably a refund and allocated to, or don't allocate it to sales, right? Or other income.

00:00 :

47:07 Christian Sobkowski: for some where, where the vendor

47:10 Rodrigo Cabrera: Category.

47:15 Christian Sobkowski: names are Meaningful, right. So Let me try to find example if you get a tax refund, right? that's that's getting classified as tax because the tax agency's name is there and that is very obviously not A Wholesale Transaction, I think something like like a bank name, that's likely different because that can be a large contract that that has been fulfilled.

47:46 Christian Sobkowski: Unless you have information on those things.

47:50 Rodrigo Cabrera: Okay.

47:52 Christian Sobkowski: Does that make sense?

47:53 Rodrigo Cabrera: Yes, yes. Clear.

47:55 Christian Sobkowski: Awesome. Hey, I'm just looking at time, I'm gonna speed up a little more still. And now we're getting to the small stuff. We bundle tax and payroll, Have a look. I think the the one thing we mentioned last time was there doesn't seem to be Recurring, payroll payments per se on these accounts would love confirmation on that.

48:27 Christian Sobkowski: But otherwise, I think this is especially the tax bits are the ones that, that hopefully are going to be quite helpful for you, because there's a lot of meaningful descriptions and it's it's working across languages. So whether it's French or Italian or German, That will work. But we're really going into the small small categories now.

48:56 Christian Sobkowski: And then, you're seeing a bit of a catch all Of refunds and unknown refunds isn't isn't meaningful. Unknown is meaningful, but You will see there's 305 transactions in the unknown bucket, and there is literally, I think three different descriptions that come up again and again. So, this is one of the action points around, have a, look at I'll send you the description.

49:33 Christian Sobkowski: they will remove get removed from If you have context on those, then unknown and into one of the buckets where where they belong. So I'm pretty confident that we can get this one to to close to zero for you with an acceleration. That's all and that concludes the account.

49:58 Christian Sobkowski: We obviously have the same thing for for the other two accounts that you sent over a lot less activity, largely large into company on both of them. I'll send that over to you. I think in, in terms of what, what the way we usually work with this is, again, this is this is the goal of this is to get you life with a data set that Feels realistic and good enough to start things with.

50:35 Christian Sobkowski: so, and what that means is that And we're, we're gonna send the full handbook, across all accounts, as well as a full data set over to you. And and really I think we identified the I'll write down The action points from the call after an email. But I think what we had identified was already pretty clear around the intercompany cheat sheet.

51:04 Christian Sobkowski: It was around the true opex categories. And then, I'll Do the two or three callouts on the smaller smaller categories, where I think. A spot. Check of your eyes is helpful. and those those are the big things that That we would recommend. Obviously go ahead and explore as much as you wish and We we will incorporate any feedback.

51:39 Christian Sobkowski: But then the goal is to get this within the coming days into a form. Where It, it's signed off as a V1. And we can then use our to, to create V1 of our forecasts for you. And then really put your life almost accounts and you can start seeing these things in the app, where it's a lot easier to click around in, then in Excel sheet.

00:00 :

52:10 Gurjit Pannu: Yeah, and just to kind of reiterate, right? Like the goal here is not to have you go through every single transaction and market as incorrect or correct. We want to get kind of that 90% Overall on the notional value and have that starting point, right? So once we have this data, Christian mentioned in the platform, you will have the ability to wear in the future if you see a transaction.

52:31 Gurjit Pannu: That's all from this categorized to go ahead and read categorize, it correctly as well. The goal is definitely to get the biggest chunk, the biggest impact and impactful categories and transactions. And so then we can continue to move forward and then over time, you know, of course, we'll get more and more accurate with more and more context, more more feedback and interaction within the app that the ultimate goal here is, Let's get, let's get it in there and as accurate as possible.

53:00 Gurjit Pannu: Without necessarily having to say, it has to be a hundred percent day one. We'll get there eventually through that.

53:15 Christian Sobkowski: Wonderful. That's that's it from us for today. Any any closing words from from the on team?

53:31 Rodrigo Cabrera: No, how clear from my side?

53:31 Christian Sobkowski: Okay.

53:32 Rodrigo Cabrera: I think really looking forward to the next iteration and see. Where it leaves us?

53:40 Christian Sobkowski: Likewise. Likewise look good, expect an email from me. Let's yeah.

53:50 Lucía Galán Cáceres: Hey actually I wanted to say It looks great, we can definitely give you feedback on the categories, but I think I know like thinking we're expecting on experience with Kariba, I would really love to get access to the tool to start playing with it because what we're seeing with kiribas that we did a lot of design, but then until we are into the tool, we are not really learning and how it works.

54:12 Lucía Galán Cáceres: So, I think, if possible, if we could do both in parallel, kind of like, review the data and work on the categories and on the context. But then also good familiar with how we would check this in tool that would be amazing. And I think also regarding the integration.

54:27 Lucía Galán Cáceres: So right now we're doing this test with three accounts. Honestly, I want to work on the cash visibility as we improve the rules. So this as soon as we have the statements flowing into quadiva, I want to forward them to you, and I think that's gonna happen next week.

54:42 Lucía Galán Cáceres: So, if you're okay with it, I would love to start processing all accounts because then Amanda, then we can have the picture of the entity level, the group level, and I think this is going to help so much to understand. I don't want to have Everything. Super well categorized, it's more about is Pound.

54:59 Lucía Galán Cáceres: Helping us get the bigger picture that we don't have right now. That's For me that's the beauty of the of the pilot, right? And then we can go into into Morgan rarity in parallel. So that will be that we might suggestion.

55:15 Christian Sobkowski: Amazing. Well, Yeah, well, well understood he is what we're doing. I think you will still get an email from from me with the action points from this call. I think there is probably two or three in there that are That we can take off literally within the next five days.

55:40 Christian Sobkowski: We're gonna start working on getting this, this life for you. In hell.

55:46 Rodrigo Cabrera: Yeah.

55:47 Lucía Galán Cáceres: Amazing. Super looking forward to it.

55:49 Christian Sobkowski: Okay. Thank you so much as always. And yeah, keep an eye out. Tomorrow from, from us. Okay.

56:01 Amanda Mitt: Thank you.

56:02 Lucía Galán Cáceres: Thank you. work.

56:03 Rodrigo Cabrera: Yeah, I think.

56:03 Lucía Galán Cáceres: Thank you for the presentation.

56:04 Rodrigo Cabrera: Thank you both.

56:05 Sofia Govoni: Thank you so much.

56:05 Amanda Mitt: Yes.

56:06 Christian Sobkowski: Have one evening.

56:06 Emma Sjöström: He?

56:07 Lucía Galán Cáceres: Enjoy the US, right?

56:08 Gurjit Pannu: You, I

56:09 Christian Sobkowski: Feel safe.

56:09 Sofia Govoni: Bye.

56:11 Amanda Mitt: Thanks.

56:12 Christian Sobkowski: Right.