Skip to content

ON - Variance Analysis Working Session - 2025-11-11

Metadata

  • Date: 2025-11-11
  • Company: ON
  • External Participants: Lucia Galan.caceres (Team Lead), Amanda Mitt (Treasury Analyst - Reporting & Analytics), Yulia Ershova (Treasury Corps - Cash & Risk/FX), Federico Morando (Treasury Intern), Rodrigo Cabrera
  • Palm Participants: Emma, Gurjit, Jennifer, Giannis
  • Type: Customer Call
  • Domain Areas: Reporting, Cash Forecasting, Cash Visibility, Variance Analysis, Forecast Performance
  • Recording: https://tldv.io/app/meetings/6913098c86ce6c0013959f4c/

Summary

Context

Customer working session with ON's treasury team focused on designing variance analysis processes and dashboards. This was a collaborative session to understand ON's needs for both short-term cash monitoring and longer-term forecast accuracy analysis.

Key Discussion Points

  • Distinction between daily/short-term cash position checks vs weekly variance analysis
  • Daily checks are about confirming payments went through (anomaly detection)
  • Weekly variance analysis focuses on forecast accuracy over time
  • Need for 13-week historical forecast accuracy view
  • Interest in understanding model performance and explainability
  • Investment account data integration being developed

Pain Points

  • No formal variance analysis process exists currently - this is completely new
  • Need to manually verify that treasury-initiated payments (from Kadiba) went through
  • Can't easily explain forecast accuracy to stakeholders ("can I trust it?")
  • Model explainability is challenging - hard to answer "how reliable is your forecast?"
  • AP data not yet integrated, limiting variance analysis completeness

Feature Requests & Needs

  • Weekly variance check with flexible time frame selection (1 week, 2 weeks, custom)
  • Focus on balances for AG (multi-currency), holding, and EMEA entities
  • Variance threshold alerts (starting with 20% threshold)
  • Dedicated variance analysis section in Palm (not just sub-function of forecasting)
  • Model performance visibility - slider from simple to detailed ("how freakish do you want to go")
  • Ability to configure models per account (disable ML, include/exclude ARP)
  • Anomaly detection for unusually large payments that didn't go through

Jobs & Desired Outcomes

Job: Confirm treasury-initiated payments went through daily Desired Outcomes: - Minimize the time to identify delayed or failed payments - Reduce the risk of missing critical payment failures - Increase confidence that funded accounts have correct balances

Job: Explain forecast reliability to stakeholders Desired Outcomes: - Minimize the effort required to demonstrate forecast accuracy - Increase the transparency of model selection and performance - Reduce skepticism from finance stakeholders about ML-based forecasts

Job: Investigate variances between forecast and actuals Desired Outcomes: - Minimize the time to identify root causes of forecast variances - Reduce manual effort to drill down from summary to transaction level - Increase the ability to make informed model configuration decisions

Domain Insights

  • Two types of variance checks: Daily (did payments go through?) vs Weekly/Long-term (was forecast accurate?)
  • Verification scope: Treasury only verifies their own payments (from Kadiba); AP payments are tracked by accounting
  • Anomaly focus: "My assumption is everything works" - only flag exceptions
  • Trust factor: Other treasuries (HelloFresh mentioned) asking same question: "can I trust it?"
  • Feedback loop: Variance data enables better model configuration choices

Action Items

  • [ ] Palm to create two dashboards: (1) daily balance/payment verification, (2) 13-week variance investigation
  • [ ] ON to share updated maturity timeline files for time deposits
  • [ ] ON to share recent Kariba report for investment account ingestion
  • [ ] Next week session to focus on investment account data once ingested

Notable Quotes

"My assumption is that everything works. So either I'm told to check something or I'm going to say, okay, great, yesterday is closed. Let's look at today." - Lucia

"Can I trust it? Machine learning can be such a big thing. So I think this will be really cool to have that tap." - Lucia

"We would wonder if the account balance is much higher or lower than we expected. But I would still expect the talent and tax team or the AP team, the ones that inputted the payment at the other day" - Lucia (on who owns verification)

Full Transcript

Them: Hey.
Me: Hello.
Them: Everyone.
Me: H. I.
Them: Hello. How's it going? Good.
Me: Good.
Them: Nice. To meet you, yanis. Nice to meet you as well. How are you doing? Good. In Berlin, but yes, okay. I can imagine. Yeah. Fortunately, here, I'm in the Netherlands. By the way, we have some sun, so I'm really glad that we do, because the past two weeks, we are all cloudy and. Yeah, I mean, come on, it's November. Just got in. Right. You're right. Nice. Hey, lucia.
Me: He? Y. I'm just going to start the TLDV recording, just so everyone's aware it's being recorded. It's a bit slow, but starting soon.
Them: Is this everyone for today, from your side. I believe Yulia was also joining, but I think we can start slowly. Well, we can give her a couple of minutes.
Me: Sorry.
Them: Giannis is new. Maybe if we're going to wait for Yulia, maybe you guys are on side. Can do your introductions first, then Giannis can do his last, then he can meet everyone in the team. Amanda, do you want? Just looking it up. But one second. Sorry. I didn't mean to put you on the spot. Hi, Amanda. I joined on April of last year. And yeah, it was really cool because we just started the conversations with Palm a little bit after I started at on, so our journeys are pretty like together. And then it was nice because I did in the beginning a lot of the implementation we found with cadiba and now focusing more on reporting and analytics and yeah, working a lot on the dashboards and kind of being this bridge between the treasury team. And palm and the request the needs together with the deco as well. So. Yep. Nice to meet you and welcome. Nice. Thank you. Hi, everyone. Hi, Yulia. Hi, Elia. We're just doing some on introductions and then we're going to introduce Giannis to the team. Nice to meet you. Yannis has already introduced myself. I'm Federico. I'm the new treasury intern. I joined on the 1st of September. So now it's basically most two months and a half that I'm here and I focus more on the report and analytics side of treasury operations with Amanda and since. I joined I really the chance. With palm. Can you hear me? It was all blocked. Sorry. Okay, so as I was saying, had the chance to work a lot with Palm since I joined. It's something that I'm enjoying a lot. Because you get coffee insights. Step by step, and it's so really interesting. Nice to meet you. Did everyone already have the round? Am I the last one? No. Well, anyway. Yasu Yanu. So my name is Yulia. I'm a part of. Well, I'm a part of the Treasury Corps within the team. Mainly let's say from the business perspective, covering cash, a bit of risk effects. So from palm perspective, the forecasts which palm producer shows us this is one of the most interesting for me. And also let's say the overviews how we can have it from the different groups on a personal basis. I joined on this here. Previously I was in Zalando, so I live in Berlin. Actually located in Berlin already for 13 years. My husband is Greek, so that's why I have a connection to you. Well, welcome to the team. Happy to meet you. Yes. Pleasure to meet you. Nice to meet you. Very cool. So perhaps to have a new face. I'm Lucia. I'm leading the team. Giannis And I think I'm excited for all of the reasons everyone also already mentioned. I really love these collaborations. So super happy to have you on the team. Thank you all. Well, my name is Giannis as you guessed, so I joined Palm last week. Actually super excited because I really loved the product. Before even I joined the team, I had a couple of discussions with Chris and Rob. I'm in this, let's say, solution engineer role for a while now, let's say four or five years. So I consider myself your actual legal counsel to the team. Of engineers together with Jenna Emma. So anything, any feedback, any questions, whether you have on mind. Please forward it to us and I'll make sure to, let's say, represent you, let's say to the team of engineers. And what's your background? Where did you come from? Well, I'm Greek. I live in the Netherlands. For the past four years. My background is computer engineering. I worked in different, let's say, settings, from medical devices to telecom and energy in, let's say, more on the engineering side of things. But in product related roles. Cool. So you're a techie, but really good communicating with business. That's why you transitioned. That's a very nice way to put it, yeah. Cool vice profile. Welcome to the team. Thank you. Nice to meet you as well. Okay, great. Well, the folks of this session is variance analysis. So more of a direct session, if we're happy to just dive in. I thought it could be nice to start with, if you sort of question and then we can show. I'll share my screen show a bit of a dashboard that we've set up. And we can go from then sort of redesign it, put some ideas together to make it something more worthwhile, to make this process as easy as possible for you guys. So I guess, Amanda, in terms of the process of variance analysis that you envisage, last week you mentioned a weekly variance analysis. Is there any process that you're doing at the moment that we're trying to replace, or is this a completely new process? No, it's completely new, I would say. And then also, we kind of discuss with the team as well. We prepare this table with kind of a summary of all of the different horizons that we would need for variance analysis. And they kind of follow the same kind of ideas. The forecast. Right. Like, we would have very short check that we do sometimes, for example, checking if a payment or investment occurred, like kind of checking that daily check. With the painter. Two files, but in palm. I think it's like, for example, when you see verified or unverified cash flow. And then I think what we discussed most last week was more on the short term, for instance, so the weekly one. But we also discussed with the team that it would be really nice if we could kind of select between one week and two weeks or even dreaming big, like, the time frame that we need. I don't. Know if that's possible, but it would be really cool because discussing with Yulia and the treasury cording that's actually doing the cash management that the cash positioning they check or with the forecast. Let's see if I know Pam already has a weekly one. And with some of the reporting, we actually have the 13 weeks week by week. So maybe we can play a bit with these reports and do a weekly two week experience. But even ideally the time frame that we need and then we just thought about checking. The balances for now. So like here we took some notes. For us, it would matter most now focusing on AG because of the different currencies and the spots. On holding and then the EMEA entities, so on uk, on Italy, like the entities that we need to kind of the funding sometimes and send the retainers or the standing orders, as we call them. So these were just some notes that we took. And. Yeah, excited to see what you built as well. So maybe we can kind of think a little bit about the process. So we would check, for example, the forecasted balance. The actual balance, and then we would kind of focus on variances above 20% for a start. That's what we kind of discussed with the team. That would make sense for now, since we don't have a process and we just have somewhere to start. Okay, so, yeah, that's super interesting. So before we move on from this first two points, which is super short term, I guess. I'm going to relay this back to you to make sure that my understand is correct. So the purpose of these two bits is predominantly cash management. So like intercompany fund and FX if need to do spot trades. So as much as it is variance analysis, What has happened. It's a flash. These payments have gone out. Especially if we're focusing on the core entities that you manage centrally. A lot of these things we're going to recognize. This payroll happen? This happened. That's what we were expecting to happen, rather that. Variance analysis. Is such that the forecast was correct. It's so short term. We know that this is supposed to happen. Has it actually happened? Yeah. Confirmation that the payments went through when it comes to the first one than any sort of. So the forecast is. I mean, the forecast for a C plus series, starting balance, expected payments, closing balance, and then we compare that closing balance to the actual balance that comes in the statement the next day. So it's a check. That the payments went through. For example. Operational. Sorry. Yes. No, I was just saying that, for example, we tried, for example, with the actuals and the forecasted for this week. And actually, that's why it was the main point. Also for me, for example, there were some major variances that it's just to understand with those percentages that I really like, the payments has been delayed or actually to have on the short term, like the monitoring view of what's happening and what's going to happen. Sorry, Emma, before I jump to you. Say in this example. We use payroll as the example. People get paid. On the 10th of November, and we're checking to say we don't necessarily want to know there's been 100% variance. That it wasn't paid. It's that we funded that account. The balance in that account is still high. Therefore, I want to look next step. Was that because my payroll payment didn't go out yesterday? No, it didn't. Then I'm taking actions rather than in the last week my forecast was this accurate. And this is why it's very granular in that we need to see these flags, and therefore, we need to take action. The payroll payment needs to go today. Does that make sense?
Me: Yeah, Just the naive question, the forecast that you're looking at here. Are you expecting that to come mainly from you, like file upload or. Just to get a bit more nitty gritty, because we have the verify feature, right? But that currently relies mostly on either one offs that you input. And we can enable it also for the, you know, files or maybe, you know, specific items from the file. To be fully transparent with playing with this concept of tracking and not tracking specific payments that you really care about. I'm just trying to dig into a little bit. Like, where do you expect to.
Them: That's a good question. I think, for example, for investments or for the effects, we were discussing with Yulia and the team that they could also include the plan for the spots that they do in the week and they can use like, for example, this feature. But I think for AP payments, we were just considering Palm for now. And this, of course, one day. Ideally we will connect with BigQuery and I have this ERP data as well, but for now, no, we're just using the Palm forecast and the manual addings. That's but in the future, I think when we managed to have the especially the AP data. I think the big ones there are the ones that we could. I don't know, Track, Verify. Yeah, I don't know. I think it's. It's an interesting debate. Is it worth it to go one by one and say, you know, I don't know, can we not?
Me: Or is there. Will there ever be specific? Like if we. When we have that a bigquery integration in place, we're moving your AP data automatically.
Them: Let.
Me: Do you see? Like you said, everything might be a big push, but would it be what you to say? Hey, this big amount here. I don't know. This one is important and explicitly track. That.
Them: 's rare that we would know what's important because this comes from ap. So, you know, it could be once in a, I don't know, in six months that we hear like, hey, this is an urgent payment. It really needs to go through. And then we would follow it. But it's very rare that we're asked to do something like this. Usually the payments that we care about are the ones that we initiate from cadba. So these are treasury payments. So for those, we definitely need to make sure that they went through because no one else is going to check. But for the invoices, accounting is going to let us know something didn't go through. So it's more like, I think here is a good use case for leveraging AI and say, like, hey, this was an unusually large amount in CHF and it didn't go through. Do you want to check? And then we would, I don't know, check with accounting and so on. But either we are told or we see a huge variance in the account balance or we're not going to go individual payments level.
Me: Yeah, that makes sense. But then for the data you do in fact upload, let's say, in a file. Tax or talent tracker.
Them: Even those. Right? I mean, I think we. We would wonder if the account balance is much higher or lower than we expected. But I would still expect the talent and talent and tax team or the AP team, the ones that inputted the payment at the other day. I mean, we're inputting the forecast, but who is actually inputting the payment in whatever platform they would track.
Me: Yeah, I would love to return to that in a different context. When we. When we're a bit further on. Cool. Thank you.
Them: Yeah. Sure. So, yeah, thank you so much. I think that is very clear on the objectives of this. In the sense of. What is the purpose? To identify delayed payment. It's more of an anomalies focus. I was just thinking like my. My assumption is that everything works, right? So my assumption is that the payments went through counter funded. So either I'm told to check something or I'm going to say, okay, great, yesterday is closed. Let's look at today. Yeah, and there's definitely a way that this plays into the account positions page. You know, the cash positions page? Where the account that are, say their cash flow header account, they're not zero balances and they say at risk of overdraft or something like that when they shouldn't be on another side. You know, it's, it's those sort of things that we're trying to flag every day and. That could almost be the first thing that we want to see as well. We're playing into a conversation. That I've got a session tomorrow to redesign that home page of an anomaly. Focus there would help manage the cash because it's the first thing that you see. Okay, that is definitely. When we speak more about the variance analysis in a more purist sense, I think it's more the second or the third road from the excel de Amanda showed. That's more, I think, the variance that you and us speak about as real variance. Check. Yeah, yeah, exactly that. And this is one of the things that I wanted to get into, in the sense of the weekly check and how it'll be interesting to see it play out is to how accurate the forecast is week by week versus month by month, and how that'll change over time. And how we will adjust the report. Maybe initially it's over a bit of a longer time frame, and then it becomes more precise as the AP data goes in, as other things go in, and then it can become more of a weekly check. But Jennifer, would do you imagine, do you envision at some point the variance to be like its own section within the Palm tool. Like, I don't know, a place where you can, you know, you can go nerdy and say, ok, I want to check my model performance one by one. Because right now it's just like a sub function in the forecasting, right? So you can double click and see the variance. But would it be cool to have, like, a complete section just on variance? Analysis. Yeah, I think that would be great. And, you know, We can also make as many views of different dashboards as we want to in the sense of we can. Have a currency born in this module. This one we can have a Category 1. We could have this. We could have all the different time frames, you know, but I think. Which would always lead back to when the investigation comes to you, I want to see the actual transactions in this category that is driving this variance. To link back to that transactions page at the app, which is my favorite page because it's so easy to use, it's so easy to filter. But when you get to okay, I can see. So I'll flatter ice cream for a second as an example. Taken this as an example. I can see this is strange. I want to get into that right now. As it stands, I would always find it much easier. Even though I've got all the data in here. I do like to have that transactions piece. Sorry. Team yannis.
Me: Mute. D.
Them: Oh, you want me? Always the same issue. Sorry. Probably I want to say the same thing as Emma's going to say. I'm not sure if it's going to be like a main base on the left of the sidebar as you have on Forecast, but what we're definitely planning to do is have the reporting. So because of variance, dashboards can be built very differently for many customers. And what essentially we want to provide is the flexibility that you are able to choose which charts with entities, with transactions you want to pick. And based on that, you can play around and create your own custom reports specifically for variance. But I think for me, the variance that you can get in a report is usually okay, so you know, the forecasting is happening on the background and this is the actual versus forecast. But what if I want to understand the more details about the model statistical model right behind it. So wouldn't it be cool to have a section where and it could even imagine like a slider and say how freakish do you want to go. You want the simple view of your forecast is good. You ask us to do 90% and we're hitting it. Or do you want to check the models that are being selected? So also I'm trying to think of when we are asked as a Treasury team, hey, how reliable is your forecast? How confident, do you feel? The answer also varies depending on who's asking. So to someone, you can just say, hey, it's good, we trust it, and they're fine. And some other people are like, no, but which models are you using? What's the statistical error that you get? And then those data from the problem right now.
Me: Yeah, 100%. So I'm just gonna rephrase slightly. So the dashboards, actually how we're using them, a lot is built when we find there is like a very customer specific thing we need to build. And that's a great. But we're actually also using them to learn a lot. So I think investments will be a great example. For example, we're now building it in embeddable fast. Our plan is definitely to have it in the app as a very like much more sophisticated investment like view in the app that's connected to everything, lets you drill down having that holistic view of your day dying up and I think we'll definitely looking at improving the experience across like the app as well for variance analysis. We're currently actually. I know Rodel mentioned this in the workshop, but there was a lot of information shown at you guys. So we're actually rebuilding our internal architecture around or. That's done. So it's about. We have, like, automated metrics, evaluation metrics, and the background that forms the foundation for. Picking automatically which model. What we definitely want to look at is what you're saying, like, provide more. Views for you in the app. Ways to judge, like, how accurate has it been historically? Maybe even, you know, we could look into stuff like, hey, if I would have used that forecast instead, how accurate would I have been? The explainability part is always a little bit challenging when it's depending on the model used, right? But that's definitely something we want to explore more. Because 100% is our priority to make like the forecast, trustable and use 100%.
Them: You have to damify it a bit for us non computer scientists, of course. Because there's a lot of information we're not going to be able to digest. Or explain right otherwise. But it's such a fundamental piece. I mean, also for transparency. We had a call yesterday with Felix or HelloFresh and he is also speaking with you. And he was asking, like, how do you know you can trust it? I said, that's exactly what we're working with the Palm team on that's. Exactly the question. And it's him and it's every other treasury out there that we're going to say, but can I trust it? Machine learning can be such a big thing. So I think this will be really cool to have that tap. And then something else that we spoke with him and we spoke in the workshop is based on this variance analysis. It also gives us the tools to make choices as to how we want to configure the models. Like we say, hey, for this account, please deactivate machine learning. Or for this account, please make sure that machine learning definitely includes arp, whereas for this one our ARP is shit. Accounting is superlated booking everything. Ignore it. And I think for this, we need that data. This feedback loop that allows us to make better choices when it comes to forecasting. That's why I see this as a separate function. But yeah, sorry, I digress from.
Me: Amazing feedback. Super good and all aligned.
Them: Sorry that we go back to your. Yeah, because, I mean, Emma shared some very cool designs for this next v2, which I will say Emma, share another day, but it does do all of those things that you just said. And I think you also shared. When we had our sessions, right? Yeah. Okay. Well, you guys.
Me: Yeah, but we're actually more, like, thinking we want to actually find them even more visionary. Picture of it. So just getting started on that, hoping to get your feedback and then, yeah. And. Sorry, sorry. I know I'm hijacking. Just quickly, if possible, Julia, especially, we would love to see how you're working with the caribou forecast currently. It's also very good input for us on anything that you think is just falling short with the experience or accuracy or trustworthiness and. All of that.
Them: Of course.
Me: Thank you.
Them: Okay, and then we have a few minutes left. Eman. Do you want to talk the last point? Because I think this dash morphed the more traditional variance analysis which is the last point in your sheet. And then I will then prepare a more short term dashboard to hit the first two points. For next week. Emma, right? No. Oh, sorry. I did say Amanda, but actually, I don't want to make big assumptions, but do you want me to quickly share the dash? If it is more of a 13 week analysis. That is the premise of the proposed dash. Yeah, I mean, we were discussing more the weekly one with the team. And also Federico like this kind of quick. We did a first version of it, kind of resisting graph. But we just use, for example, the actuals from this week and compared with the forecast from last week. So something like this just initially would be already super cool for the weekly one, but for the 13 weeks, I think we need. Yeah, maybe we can discuss a bit and we can check your proposal based on what Lucia said. And what we've been discussing as well, but for the weekly one. Yeah. Maybe we can already start. Yeah, this is great. Ok. Ay. So this is obviously very easy to do. In the cash. My next question is sort of. What next? So if we pick an example one. On the Shanghai entity. Is it? We get that information? We know that. There's variance there. Might be the timing of the statements, now that I think about it, but we might. Probably be too high, so we need to consider that as well. I don't know. I think the timing of the statements from APAC right. Yeah. Sorry. Fred? No, just wanted to ask, obviously, it would be nice to have now see how to assess the methodology with also the dashboards and to see how to structure the processes. But like, for example, these are like big shifts and for example, sometimes it's about the timing of receiving the data or like payments like to see if they went through or not. So it would be. I wanted to ask if do you think that would be like in a hypothetical variance analysis section, just as Lucia was saying, to actually make the model learn for some entities or for some payments based on the timing of when the data is received. Or like, if we know that some payments are like they have a buffer of some days usually and make the model learn based on that. Or do you think should be more manual every time and check. Yeah, of course. I think the models will definitely learn, especially as the more data feeds into them and they also recognize patterns in payments. So if they're a consistently saved one, the APER data goes in, some are consistently late. It will learn that say we say on career, say that they're going to pay their AP on this. Day, but actually they pay usually a week later than that. It will pick up those patterns too. Where I was thinking of going with this is if we. And actually, I'll share the dashboard now, because I think this makes a little bit of sense. Once we get that data, how best to investigate. What's caused those issues. So I've split this dashboard. I tried to make it as visual as possible and split it into lots of sections. We don't have to do this. We can summarize it, but you already mentioned that the certain entities that you would be looking at first, so I guess choose. Sorry. I've had this open for a few minutes. When I say a few minutes, probably like a few hours. You often want to zoom in on one particular entity or two or three that you manage. So you could see a top level sort of variance and then have the option to try and locate what is causing. What is the main driver? So this, to me, looks like if I was using this in real time, looks like the big variances are in sales. Cash in and then to drill down further. I've already left by entity, but specifically in USD. And then. Oh, I might want to actually export this data. And do additional analysis, but I'm wondering now if. Based on the things that you've shared, we can make this even more specific. Where when you look at this C, I can preset some options. I like to look at these few entities. I fund in these currencies. Therefore, when I first open this dash, this is where I want to direct my attention. And then beyond that, you can always adjust the filters as you choose. But to make this something that you want to look at every week and is quite easy and visual. To highlight the anomalies and I could. Maybe, based on what you'd set up, I'm going to add some key numbers as well. Maybe the valence variance fall on AG and some of the other entities. We could even add some for currencies too, if that helps to make this as sort of. As snappy as possible for you to then take actions or get on the phone and speak to someone to understand what happened with that. What do you think? That is sort of a first pass. This is back end at the moment, so this is still in development, but I can make those updates. And let you have a look at them, especially in relation to the process that you've just shared as well. I'll recreate that table, too. And then you can also see if the numbers match so you can get quite comfortable with the report. Yeah, I think that's great. It would be nice to see kind of the snapshots, right? I don't know if you can see it somewhere. Like, for example, is this showing forecast as of. I don't know. Maybe something like this, like, visually, I think would be super useful as well. And I do see. The balance, like you said, being also very helpful. But I think that even more detail than we expected. So really cool. Yeah. Let's refocus in on the let's have two different dashboards. I propose you have one focused on variance investigation and one focused on the differences and balances, the big payments more of the daily check and this one more to investigate, sort of the bottom end of that table, the long term variances. And I think it is quite important to segregate them because of how the. Sorry, I know we're running over trying to rise the scene quick. But obviously, as you update the forecast over type, the forecast becomes more accurate. However, if you want to do a full variance analysis, you want to look back over the last 13 weeks. You're looking at the forecast from 13 weeks ago. And that's what this dashboard does. This is looking at how accurate the 13 week forecast was. We'll set up the other dashboard to hit your first two points. Looking at the most recent forecast. What actually happened. And what was supposed to happen yesterday, do we need to take actions on that? Right now. Does that make sense? Yeah. You start with me. What you've shared is really helpful context here. And I'll get both of those two dashboards in your environment sort of over the next few days, and you can sort of. Hopefully we can do a little feedback in the next session to see if they've been useful. I did also want to talk a little bit about next week. If we have one minute. That Sarah, our engineer, she is working on ingesting the investment account data and she said that could be even up and ready by the end of the day, the end of the week. So I'd really like to purpose next week's session to look at that data in the system and was wondering if you could share a more recent report from Kariba that we're using to do the ingestion. I can send you the historic gone that we have, so you've got the last date, and then we'll get that in and we can start looking at that as well. Get some. I've already got my investment dash sort of designed, but once the data is in, we can adjust it and we can do another session to sort of make that help with that sort of investment planning. Which will be next slide. Perfect. Yeah. Great. Yeah, I think everything else will sort of summarized in the Slack message. I think everything from last week was pretty much closed off. And is there any more points for anyone for this week?
Me: Just.
Them: Now we're meeting. Yeah, sorry.
Me: Sorry, I was just going to ask who can send us an updated like the maturity timeline files for the time deposits.
Them: Please.
Me: Perfect.
Them: Yeah, I can share that.
Me: Just so cool. Thank you.
Them: But, yeah, thank you for this and look out for our message with the new dashboards. Thank you so much. Thank you. Nice to meet you. Giannis welcome. Thank you. Nice meeting everyone.