Skip to content

Sonder - Cash Forecasting Philosophy - 2024-11-19

Metadata

  • Date: 2024-11-19
  • Company: Sonder
  • External Participants: David Watt (Treasury)
  • Palm Participants: Emma Sjöström
  • Type: Customer Call
  • Domain Areas: Cash Forecasting, Variance Analysis, Connectivity
  • Recording: https://tldv.io/app/meetings/673cbddceafea00013094313/

Summary

Context

Deep discovery call with David Watt from Sonder Treasury about their cash forecasting philosophy, process, and what makes a forecast trustworthy. This builds on a previous conversation where David described forecasts as "an opinion" - this call digs into what that means in practice.

Key Discussion Points

  • Forecast as opinion: Treasury must apply judgment to inputs from other teams, not just accept them as-is. Different teams have different accuracy levels (tax is reliable, sales is optimistic)
  • 13-week rolling daily forecast: Standard process updated daily in ~15 minutes; actuals replace forecasts with variance analysis
  • Collections hardest to predict: Outflows are controllable (payroll, rent, tax), but inflows depend on customer behavior
  • Stakeholder management: Different touchpoints per team - calls with real estate, Google Sheets with tax, automated reports from payroll
  • Trust in AI/ML forecasts: Need clear designation of data sources (color coding), track accuracy over time, ability to mark outliers
  • Conservative approach: "Rather miss low than high" on cash inflows - upside surprises are better than shortfalls

Pain Points

  • Collections volatility - Customer behavior is unpredictable (bookings affected by events like Super Bowl, elections)
  • Bank connectivity gaps - HSBC to RBC migration caused 2-day data lag; different banks provide different data quality
  • Stakeholder input quality varies - Sales optimistic, finance teams more reliable; need to sanity-check inputs
  • Outliers skew algorithms - Severance payments, one-time events need to be excluded from forward projections

Feature Requests & Needs

  • Data source visibility - Clear indication of what's actual vs manual vs ML-generated (color coding suggestion)
  • Outlier marking - Ability to flag one-offs so they don't affect future forecasts
  • Manual override capability - Always need ability to adjust forecasts with new information
  • Percentage adjustments - "Payroll's gonna be 10% lower from the 15th forward"

Jobs & Desired Outcomes

Job: Produce a daily cash forecast I can trust and defend to stakeholders

Desired Outcomes: - Minimize the time spent updating daily forecast (target: 15 minutes or less) - Increase confidence in inputs by understanding their source and historical accuracy - Reduce the risk of being caught by surprise on large variances

Job: Apply judgment to stakeholder inputs before including in forecast

Desired Outcomes: - Minimize the frequency of blindly accepting optimistic projections - Increase ability to sanity-check inputs against historical patterns - Reduce forecast error caused by accepting unrealistic inputs

Job: Run the business on minimal operational cash

Desired Outcomes: - Minimize idle cash sitting in low-yield operating accounts - Increase cash deployed to higher-yield investments - Reduce buffer requirements through forecast accuracy

Domain Insights

  • "Forecast is an opinion" - Treasury professionals should filter inputs through their own judgment
  • Conservative on inflows - Always safer to forecast lower collections; upside is better than shortfall
  • High transaction volume = lower volatility - Sonder's many small transactions make forecasting easier than lumpy B2B
  • Automation = reliability - Direct bank connections (NetSuite → bank) make process more predictable
  • Snapshot comparison - Track "what did I think on Nov 1st about Nov 20th?" to measure forecast accuracy over time
  • Color coding convention - Black = actuals, Blue = manual inputs, Green = system-generated

Action Items

  • [ ] Design data source indicators (visual distinction between actual/manual/ML)
  • [ ] Plan outlier flagging feature for one-off transactions
  • [ ] Consider forecast accuracy tracking (snapshot comparison over time)

Notable Quotes

"If you don't have an opinion about is that the right thing? Or how likely is that really to happen? It can go pretty wrong if you don't apply your own judgment. Just take a bunch of other people's guesses? Then there's no kind of logic to it." - David Watt

"For cash flows, if you want to be conservative on the inflow, right? So I'll put six in there, if you sign eight that's upside to the forecast. You'd rather miss a little too low than too high." - David Watt

"Algorithms have a hard time with sudden changes like that, right? So just being able to ignore this outlier, mark it as an outlier." - David Watt

"We try to do it in 15 minutes or less, right? You don't want to spend a lot of time on data entry. You should drop it in there and be able to quickly analyze the variances." - David Watt

Full Transcript

00:02 David Watt: well, I mean For Treasury. It's a couple things making sure you have the right amount of cash and the right place the right time is kind of our starting point. And you have to forecast, to be able to do that, because if you, Obviously have don't have enough payments will get delayed or fail.

00:20 David Watt: If you have too much you're losing out on, you know, interest income.

00:24 Emma Sjöström: Yeah.

00:25 David Watt: and so, That's the the Treasury kind of the tactical thing I think for the company you have to have a forecast because you have to Plan ahead and make decisions accordingly, right? So you have to You know, for Sonder, especially right or we don't have a lot of excess liquidity so we have to have a very accurate forecast.

00:43 David Watt: To guide our actions and Even down to what kind of contracts to sign and, you know, the terms of the contracts matter, right? If there's a large up front payment or if it's Monthly or quarterly that kind of thing. So, It depends on the company, but I think those are fairly typical consideration.

00:00 :

01:02 Emma Sjöström: Yeah, thank you. That makes a lot of sense. So like I remember from our last conversation you, you said that to you or forecast, this is an opinion.

01:14 David Watt: Mm-hmm.

01:15 Emma Sjöström: and I found that very interesting and I, I'd like to dig a little bit deeper in like, what what makes a forecast usable to you, like, what makes you Trust the forecast presented to stakeholders. Use it in your day-to-day like what are the yeah? What needs to be there for you to feel Safe with it or trust in it.

00:00 :

01:41 David Watt: thing is when you're if you're building the forecast or if you own the forecast, like you don't just take Inputs, or you're not gonna know everything. So you're gonna have to work with other teams to get insights into different line items. but you have to also, Judge, you know, the accuracy of their of their guidance.

02:04 David Watt: And I think that's what I've seen a

02:04 Emma Sjöström: Yeah.

02:06 David Watt: lot of people maybe struggle with that early in their career. They just say Well that's what so and so told me and I put it in there but If you don't. You have an opinion about is that the

02:14 Emma Sjöström: but,

02:15 David Watt: right thing? Or how likely is that really to happen? Right? Maybe they have it, it's gonna happen. But is it gonna happen when they said it's gonna happen? Or If the date is certain but the amount is uncertain, right? That's kind of the piece of The forecasting art, I guess, is that? You know, you should have, you should apply judgment if you own the forecast and be able to.

02:39 David Watt: Explain it to people because it is, it's just a guess in the future, right? Like, there's always unknown items. But I think it can go pretty wrong if you don't apply your own judgment. Just take a bunch of other people's guesses, right? Then it's It's not, there's no kind of logic to it.

02:58 David Watt: Because different people are gonna have different. Levels of confidence and stuff, but if you just take it all as equal, then it's really not as useful. You have to I think kind of filtered through your own judgment and put it in there. What you think is most likely to happen?

03:11 Emma Sjöström: How do you do that? I think it was

03:14 David Watt: Yeah.

03:14 Emma Sjöström: What? Yeah? How? Yeah. What? What do you need to be able to make a judgment call? If I If I come, and I tell you that, oh, I'm making. Doesn't that investment then that, you know like well what would you?

03:28 David Watt: Yeah.

03:28 Emma Sjöström: steps would you take to kind of? Sanity check me basically.

03:33 David Watt: Yeah I think always asking for like How did you get to that right? Like Oh is that contract signed Or is that just you expect to sign it right? Like that gives you an idea of like Oh it's signed and it's got real terms, okay? That's pretty certain But especially independence who you're talking with, you're talking with sales, right? They're very aggressive by nature.

00:00 :

03:53 Emma Sjöström: Yeah.

03:53 David Watt: So they're gonna assume things are gonna happen but if you ask him like Oh that's great. Did you guys sign all those deals already? And then it turns out Well, no, but we have, you know, the goal of signing 10 of them this month, you know, okay. How many did you sign last month? Well we signed six right? And I so why would you think we're really gonna sign all 10 like I think trying to understand why people, you know, the little background don't just take their input as is true, but You know, so how many of you sign the last three months? Like, Well we've signed six each of the last three months.

04:35 David Watt: Well, Why would I put 10 in? Then right now, right. Like maybe, you know, oh, because four of them are gonna You know we're almost signed but we signed them on the first instead of the 30th or something. Right. There might be a good reason for it. Like Yeah.

04:40 David Watt: Okay. Yeah. They just signed But you know some groups are always optimistic. Some are probably too conservative, that kind of thing. And then you can also use the history, right. That's when I think we talked about before but always try to build the forecast that's gonna match how many get my actual data so that I can easily compare.

04:56 David Watt: What act what happened previously to what they're saying is going to happen in the future and Because that, you know, I can't always see what the other groups are doing, but if If I can go backwards and check and same idea, right? If they've signed six every single month, but they're saying 10 and they don't have a good, you know, they say, Well that's we think we're gonna hit the target this month like From a forecast standpoint.

05:17 David Watt: For cash flows. If you have you want to be conservative on the inflow, right? So I'm like, Well I'll put six in there if you sign eight that's upside to the forecast. That kind of thing. You'd rather miss a little too low than, too high on the cash flow.

00:00 :

05:31 Emma Sjöström: That makes a lot of sense.

05:35 David Watt: Because a lot of the people we deal with on inputs, right? They're not finance people necessarily they're definitely not forecasting.

05:42 Emma Sjöström: No.

05:42 David Watt: so you kind of have to make sure you understand how they're coming up their number because it's not Probably the same approach that a finance person would take.

05:50 Emma Sjöström: Yeah, just so I understand like what like kind of type of orca style. We now I'm guessing we're not like very, very short-term but a bit more like midterm maybe or like how because I'm also assuming, right? So if a tax team comes and said Hey we have this huge tax down payment, that's more reliable than the sales.

06:09 Emma Sjöström: Person.

06:10 David Watt: Yeah, exactly.

06:11 Emma Sjöström: Coming with their hopes and dreams to you, and trying to present it as a fact. But, so just interesting for me to understand like, I assume like the closer you get to the actual date, like the short term, the forecast, and more accurate data is available either VSM system or your colleagues and co-workers will be in a better place to actually

06:34 David Watt: Yeah, no that's 100% right. So And you're right like that's a big piece of it is who are you talking to, right? So the tax team, Our finance people. So they typically have a better understanding of numbers and timing and And in their world too. It's very like statutory.

06:50 David Watt: So it's like Yes, it's due on the 15th and you know, that's true, right?

06:54 Emma Sjöström: Yeah.

06:55 David Watt: Salespersons, like I'm gonna sign all these this month. And they're, you know, but they'll probably sign eight out of 10 in the last two days of the month, right? They don't space it out, like they have a they often You know, it jammed and tried to hit

07:07 Emma Sjöström: Yeah.

07:08 David Watt: their goal at the very last minute.

07:10 Emma Sjöström: Yeah.

07:11 David Watt: I think for treasury like there's different links of forecasts, but a very standard wanna be like a 13 week rolling forecast, right? So basically one quarter at a time. So, that's Pretty tactical. You can then kind of match that the longer term forecast would be like a three or five year forecast but you'd probably show it by quarters.

07:28 David Watt: So you can take your 13 week and kind of try to match it to the long term. Just to make sure they're not going on opposite directions kind of thing. but you're also right like the first you know this week on Tuesday I should have a very high you know person probability of getting it right for where my Friday balance is going to be I'm not as concerned about where my year end balance, is going to be, right.

07:56 David Watt: There's a lot that could happen here. In our current forecast goes the end of January, right? Like I'm it's there but it's no. I don't have nearly a little confidence that January will be as correct. Right. It closer inch you get higher and higher confidence level.

08:08 Emma Sjöström: So, if you were to like produce a forecast, it's today's Tuesday, let's say it's Monday. And you wanted to produce a forecast for your week or for your day, even let me know your

08:17 David Watt: Yeah.

08:19 Emma Sjöström: process. Does this super interesting or like how you if it was part of your process, would go about doing that. Just for the day today, like the positioning of the cash now, right? Or this week, how would you like what steps would you Take, how would you go about producing such a forecast?

08:36 David Watt: Yeah. Well, we have it. We have that in exact process, right. So it's a 13 week forecast, it's by day. So every day we come in, We update the actuals from the prior day and again, it's not just yeah,

08:46 Emma Sjöström: Every day. Yeah.

08:48 David Watt: it's not just copy paste. It's When you're, you know, putting the actuals over the top of what you thought yesterday was going to be. You have to judge. Why did you miss right? Did something just not happen. I should put it on today's date. Or did it happen? And it was less

09:01 Emma Sjöström: so,

09:02 David Watt: than expected and it's a true you know variance.

09:06 Emma Sjöström: that's very understandable, is Analysis.

09:08 David Watt: Yeah, as you're inputting the actuals, you're judging. Is this a pull forward or this get delayed? to kind of reset the forecast each day and then so that's like

09:17 Emma Sjöström: Yeah.

09:18 David Watt: know, that's a pure Treasury function takes We try to do it in 15 minutes or less, right? You don't want to spend a lot of time data injury, right? So you should drop it in there and

09:28 Emma Sjöström: just,

09:29 David Watt: Be able to quickly and analyze the variances.

09:31 Emma Sjöström: So you just have a way to quickly pull the actual somewhere and then get it into your Excel sheet, I guess. And then

09:37 David Watt: Yeah. Google Tree for us but yeah, that's right.

09:39 Emma Sjöström: Yeah.

09:41 David Watt: And then you kind of say Okay, does this make sense? Okay, yes. For the actuals and then you start reconfirming your, you know? Okay, well, I can see that. Payroll is this week and so I should reach out to the payroll team now because I know payroll hits on Friday but they submit their reports on Wednesday.

10:00 David Watt: So then you start refreshing your for, you know, the forecasted items because certain things become available. If in a real run place they would automatically send you the report. You don't need to go ask for it, but if you know, that's what you should know what forecasts that items should I get today.

10:19 David Watt: Because people forget right and so again for an under we have to be super precise in our cash. So At a place with more money Maybe you don't bother maybe you do this weekly instead of daily because they've got enough cushion that if a lot of things just move a day or two.

10:33 David Watt: So right we do more work because we do it daily. If we had a bigger cushion we wouldn't have to we could do spend time doing something else.

10:40 Emma Sjöström: Yeah.

10:41 David Watt: But that's where Sanders right now.

10:43 Emma Sjöström: Yeah, that makes it makes a lot of sense. So in the So given where you guys are at? I think that's very interesting, like then, What part of the like, Cash 4 casting process? Do you find most critical and why? Just like arriving at that focus like the process of arriving like in terms of thought what parts are more critical for you guys?

11:07 David Watt: Well, I mean, they're all important. I think the hardest ones for us to forecast are collections Because that's out of our control, right? Our for our business. People book today and we get the

11:16 Emma Sjöström: Yeah.

11:18 David Watt: money in two days, right? But knowing Or in some cases, people are checking out in two days and we get paid the day after they check out or something. Depends on the the channel that That's the hardest part for us to know because people instead of checking out today maybe they extend their trip a day and it pushes you know they that's good for the company if they extend.

11:39 David Watt: But it delays our collection.

11:42 Emma Sjöström: Yeah.

11:42 David Watt: Or. It's just they don't, you know, they just don't we have a bad day of booking. It doesn't mean the whole week is ruined or the month. It's just There's quirks, right? If it's The Super Bowl or something, everyone's watching a certain event or something, or the election Recently, there's probably less people booking trips because it's such a national event that our whole bookings could drop a day just because people are distracted by something.

12:06 David Watt: But we also flip side, see it when there's vacations like national holidays. We usually have better bookings because it seems like people are at home with their family and they think about what we should do as a family and they start booking more. So we get a little bumps from long weekends.

00:00 :

12:21 Emma Sjöström: That's really interesting.

12:22 David Watt: So, but that's hard to predict, right? That's very hard to predict all the outflows. Hopefully we control. So we have internal touch points with people paying rent or the payroll team or whatever. So we should know the outflows with a high degree of certainty.

12:38 Emma Sjöström: And use would you say you how quite a lot of process in place to make sure you do know that? Or is it? How does it work? You guys Is it like you put accountability on another team to inform you. Or are you like chasing the other teams you see, do you

12:52 David Watt: Both. Yeah, they're supposed to give a certain things. We we kind of know, but we have to monitor and make sure that we get it and follow up if we don't

13:00 Emma Sjöström: How do you?

13:00 David Watt: Yes, we've kind of broken them down, right? We have

13:03 Emma Sjöström: Yeah.

13:03 David Watt: Regular calls with the real estate team to figure out how much rent is going to get paid. Payroll is is you know, runs like clockwork. Basically. So they just send us reports when they submit it to the payroll processor. They send us a copy as well. So we know that.

00:00 :

13:15 Emma Sjöström: Right.

13:19 David Watt: AP team. We have they have a weekly budget and then they can ask us for more so we just kind of have a placeholder budget and if they go above they have to ask us to spend more so then we can adjust the budget.

13:31 Emma Sjöström: Right.

13:32 David Watt: So yeah, we've kind of line by line. We have a different touch point. there's certain unknown items like utilities just debit us when it's due but that's where we have kind of a historical track record of You know, utilities don't change that much month over month.

13:47 Emma Sjöström: No.

13:48 David Watt: Unless it's, you know, summer is very different than winter, right? Especially in northern places, like Stockholm, right? So the heating bill would be very different but that's where you cannot just look month every month, but you kind of have a seasonal track record, right? And

14:03 Emma Sjöström: For sure.

14:04 David Watt: But they're also not the most material. So the bigger items for us will be rent and payroll and vendor payments. Tax payments, they have a whole schedule. Right? Again, that's very scheduled, they know that these payments are due on the 15th of every month, or this one's due by the 31st, or whatever it is.

14:20 David Watt: So we've worked with them to build out a whole schedule and then they populate it. We and then we look at Actuals and we still double check. Hey, Did you did we get a credit or something? We don't see this paint payment happening. So there's some back and forth and you can just You know, the real estate team.

14:37 David Watt: They're not finance people, so we actually have calls to talk it through. The tax team, get it. So it's just a Google sheet and we can comment in there, right? We don't generally need to take 30 minutes to talk about it. It's just like Specifically this item. Did it pay yesterday or you know, we didn't see it.

14:52 David Watt: So depends on the group on how

14:54 Emma Sjöström: Very cool.

14:55 David Watt: you want to. You know, I was trying to be efficient, we don't want to have Four, 30 minute calls every week just to talk over stuff. If it's We don't talk with the payroll team hardly at all, because it's just a, you know, we know the cycles for each country and they give us the reports and there's nothing to talk about.

15:10 David Watt: We know,

15:13 Emma Sjöström: No.

15:14 David Watt: Yeah.

15:15 Emma Sjöström: It sounds like you don't encounter surprises start often.

15:19 David Watt: Um, I mean that's kind of the goal is

15:20 Emma Sjöström: Forecast.

15:21 David Watt: not to have big surprises. At least.

15:22 Emma Sjöström: Yeah.

15:23 David Watt: there's always a variance but yeah, no, we're that's we're pretty accurate. So what we can do, we bring our cash balance way down, right? So we can run the business on. Very little operational cash flow. Which is, which is my goal, right? Again, we don't have a lot of money, so we have to be efficient even if we had a lot of money.

15:44 David Watt: It's still a Treasury goal to have as little as money as possible because then you can put it in, you know, an investment account, or earner, higher yield, then you can in your operating accounts.

15:52 Emma Sjöström: Hundred percent. Do you see like any? So any reason why you guys are so good at keeping that like operations quite tight. And the sinks thinking quite tight in between teams is it related to the the scale of your company or your number of employees or Add geographic, geographical location, even like, I'm just really curious as to What do you think? Are the factors that have made it work for you to be? So if and, and avoid

16:17 David Watt: Yeah.

16:20 Emma Sjöström: the surprises in terms of large one-offs and such

16:24 David Watt: Yeah. Well, I mean first it's it's necessity. So we've had to build this to make that happen. But second is again some of the nature of the business, right? It's a lot of people staying but individually transactions a thousand dollars, right? But we collect millions of dollars per day.

16:42 David Watt: So, the high number of small transactions, makes it less volatile. If you know, if it's not we get paid

16:47 Emma Sjöström: No.

16:48 David Watt: once a month from Whatever a huge contract. And if that doesn't come in, it moves the number a lot, it's things don't move that fast, you know, same thing with payroll right. Doesn't change that much because we have a thousand employees. And so if 10 people quit, the number doesn't change that much.

17:05 David Watt: So I think we have a pretty easy to predict business in that sense. Because of the high number of small transactions but also we've been working this for a while. We didn't wait until we ran out of money to build the forecast right. We've Been very intentional from all the number, you know how the bank accounts work and how do we connect them to the systems? So a lot of this is automated which makes it easier to predict, right? If you're relying on people to login to portals and manually enter things, And then they're sick that day and then nothing happens.

17:37 David Watt: And you're like, what happened here? But we've built the connections. And so, a lot of it is Pretty automated. Which makes it more reliable for a forecast.

17:48 Emma Sjöström: Right. And when with connections, you mean like pulling actuals into your forecasts and

17:54 David Watt: There's that. And even how do we send the files to the bank, right? We've set up direct connections with our banks, so that netsuite, you know, when's approved in net suite, it pushes the file to the bank and it's received. And so Yeah, so it's, you know, again kind of we try to make it run like clockwork, right, where it's just gonna keep going, even if no one's touching it.

18:14 David Watt: but we still have to kind of at our level, we have to still monitor it and make sure that You know, that it nothing broke because things do break, occasionally, right? Or we change banks and we don't have the same automation for a period of time, so, Certain accounts operate differently, we're dealing with that in Canada right now.

00:00 :

18:33 Emma Sjöström: Yeah. Okay, interesting.

18:35 David Watt: Yeah.

18:35 Emma Sjöström: how long does it take for you guys to remedy that or like, how do you

18:41 David Watt: Well in the Canadian situation HSBC was our bank and they sold their Canadian operations to RBC which is Royal Bank of Canada. So our accounts moved, they just cop, you know, they created new accounts on the RBC system. We worked with it with the tech team to get the the file transfer process in place to RBC.

19:04 David Watt: So that took a while, but we got that in place. But what we've realized now is the reporting coming from RBC to Sonder is different in both the quality and the timing. So it doesn't come as quickly as it did with HSBC. Also, we don't get as many details in the data that we used to, so we've had to adjust things.

19:24 David Watt: Internally to get more details from our own system netsuite. And then we're also working with RBC. And in our workstation, Kariba To improve the connection how that data comes in so it can get to us faster.

19:36 Emma Sjöström: Right. Yeah, makes

19:37 David Watt: Because right now there's like a two day lag, which makes it confusing that today we receive Friday's data. Right or? And so it's a little everything else gave us all the other banks gave us Monday So it's a little can, you know, that part makes it harder and we're trying to fix that to get everything back on the prior day basis.

00:00 :

19:55 Emma Sjöström: Yeah. Makes sense makes sense. No, it's interested to also find time to dig a little bit deeper into just So we're building, we're trying to build a forecast that contains both like the, the automated inputs, the human inputs and the machine learning like the AI.

20:11 David Watt: Yeah.

20:13 Emma Sjöström: And those are three very different types of input. Like we covered. For example, the the human input like you mentioned for full cost to be be trustable. You need to feel like sort of vetted that that you get from

20:25 David Watt: Yeah.

20:27 Emma Sjöström: other humans. How would you feel? So let's say That's one. Okay, Sure you've already got process in place for that and whatever you would input. Let's say Imports forecasting tool would come from. That it would be vetted already by you guys and your process and your like, sanity, checking skills, the other part, they may be from some ERP system or something like that.

20:48 Emma Sjöström: Just pulling all the known actuals but then that part, which consists of AI or like this machine learning model, right? Wow. What would? It take for you to feel that you would trust a model like that. There's not. There's that question, make sense? oh,

21:08 David Watt: thing with forecasting it for in like one is If I'm looking at the forecast, Because we also have layers of people, right? It helps to have one person build the forecast and then you have someone else review the forecast. So you have a fresh perspective on it. So if Palm is now building the forecast and we're just reviewing it, one of the things that's helpful for the reviewers to understand, which number is, which type right.

21:43 David Watt: And so, One of the easy things I kind of learned early on is just using colors and in the spreadsheet, right? So like black numbers are actuals. But blue could be an input. Right? And maybe you make it green if it's a system generated input or something, right? So you can tell Well, I don't need to challenge the black items because these are actuals, right? Like in the payroll example, right? If they've submitted a file, And they give us a copy.

22:20 David Watt: Like we know, that's the exact number. It's been submitted, so that can be a black number. Versus what we think collections are gonna be, right. That could be, you know, whatever, I use color blue. And then if a machine is doing it, maybe you designate that a different color, right? Or if they if the algorithms doing it so that we can First off, just understand when you're putting this whole thing together, how much of this is, is a guess and how much of it is known.

22:27 David Watt: And then Yeah. How do you get comfortable? Like It's just then it goes back to well, every day we're doing a variance analysis, right? So, how accurate is this been? And Yeah, we know different groups, have different levels of accuracy, right? And so in them, the algorithm should be the most accurate over time, but That's you know, that's we'll check it every day and we'll figure out.

22:50 David Watt: and I think also it's we look at it daily and update it, but we also kind of take snapshots and so It's pretty easy to be accurate about tomorrow. And then I updated again and it'll be accurate the next day, but our current was I on November 1st, for what? What the 20th is going to be right.

23:07 David Watt: That's knowing Is this forecast good for a month and, you know, and it's probably less good for three months or is it just

23:13 Emma Sjöström: Right.

23:14 David Watt: for a week that kind of thing too? So, I know, I think it starts with with just being able to Either through colors or just because you know this line you can designate you know, where the source is for each line in the forecast or something. Right? Is that That gives you as a review or a level of confidence, that okay, I understand where this is coming from.

23:37 David Watt: Historically they've been very accurate so I can trust this or historically they haven't. So I better spend more time looking at it.

23:43 Emma Sjöström: Right, just make a crystal clear what's coming from where and

23:48 David Watt: Yeah.

23:48 Emma Sjöström: Maybe. Yeah, you can drill deeper into stuff in various ways. I

23:53 David Watt: Yeah, and there's always usually

23:53 Emma Sjöström: suppose, right?

23:55 David Watt: overlay too or sometimes there's just new stuff that happens and CFO calls and says Hey we're getting a hundred thousand dollars tomorrow from so and so

24:02 Emma Sjöström: Yeah.

24:03 David Watt: They always the manual input override as well, right, and having that in there or an adjustment, it's not going to be. It's gonna be 15% lower for some reason than what we previously thought.

24:12 Emma Sjöström: Yeah, or this didn't happen. So,

24:15 David Watt: Yeah, like hey, revenue teams gonna

24:15 Emma Sjöström: Yeah.

24:17 David Watt: offer a sale and they're gonna discount everything for the next 30 days, right? Just to see how the impact on collection on bookings or something, right? So, That's the other piece of giving yourself the flexibility to. You know, I guess that one we're dealing with now is sometimes payroll.

24:32 David Watt: You'll have large severance payments, right? If you do a reduction. But payroll go up.

24:36 Emma Sjöström: Ayah.

24:38 David Watt: But you know that's not the future payroll. The future payroll actually drop. So you have to have kind of the manual adjustment of like Yes, it's fully automated and I trust the report. But it's not something I should play forward. It's actually the opposite, right? It'll, it's a, it's a instead of straight straight line.

24:54 David Watt: It's gonna be a blip up and then a Drop like heartbeat kind of thing and go up and then but drop to a lower level going forward. So

25:00 Emma Sjöström: so if it's fed into a model, maybe a way to tell the model to ignore this bump,

25:05 David Watt: Yeah, exactly. Because that's algorithms that have a hard time with sudden, changes like that, right?

25:12 Emma Sjöström: So, just been able to ignore this Outlayer, this is like market as an outlier, even maybe.

25:19 David Watt: Yeah. Or say you know hey Payroll's gonna be 10% lower. From the 15th forward or something, right? Like you

25:25 Emma Sjöström: Well yeah.

25:26 David Watt: You know, you have to be able to adjust it.

25:30 Emma Sjöström: Make some makes a lot of sense.

25:32 David Watt: Yeah.

25:34 Emma Sjöström: Okay, final are there any like specific metrics or feedback that you used to kind of evaluate, but your forecast for meeting expectations. We've touched on where I varians right.

25:45 David Watt: Yeah.

25:45 Emma Sjöström: The is there anything more? Like How do you evaluate how? This was a great forecast or This is you know, you know what I mean?

25:53 David Watt: Again, with us, it's tricky because it's so daily. But in other companies that you usually, set the forecast like before the start of the quarter and then you judge your, your accuracy by how accurate were you at the end of, you know, predicting the end of quarter balance because

26:07 Emma Sjöström: they put treasury like, it sounds very Hard to work. Or am I? Yeah.

26:12 David Watt: Yeah. No, that's the game. I mean that's why they we get paid. Right is Treasury manages the cash and being able to tell management what your cash balance at the end of the period is going to be

26:23 Emma Sjöström: That's true.

26:23 David Watt: Is important. Because they can make decisions, you know, in normal public companies that helps you decide Do we need to borrow more? Do we need? Can we buy back stock? You know, like How is they're thinking of in a quarterly cycle for public reporting? How is this gonna You know, what kind of questions are the investors gonna ask us if our cash balance is dropping or if it's increasing Because that they have the input for what the investors are expecting.

26:47 David Watt: And then they we can give them. This is what we think is going to happen and that's they can reconcile in their

26:51 Emma Sjöström: Yeah.

26:52 David Watt: head saying Okay well That's not good enough. Our this is we need to generate more cash or something so they can incent the groups to do something different. Or, you know they say Hey, great, we're killing it. We know that the stock market does not. So let's go buyback shares, right? This is gonna be we're undervalued right now.

27:11 David Watt: We should go buy back more shares or something. That's the kind of that's again. Saunders. A survival level. And we have to do this to make sure we can get, but more established companies are going to be doing that to optimize performance, right? That that's what the kind of they're using it for.

00:00 :

27:26 Emma Sjöström: Yeah, 100%. so,

27:31 David Watt: Sorry, I do have to hop here. I've got another calling can't be much

27:33 Emma Sjöström: That's, that's no worries. This was

27:33 David Watt: late, but

27:35 Emma Sjöström: super valuable valuable and thank you again and hopefully see you soon.

27:42 David Watt: Yeah, we can do it again if you need another time.

27:44 Emma Sjöström: Thank you so much David. Thank

27:45 David Watt: You're welcome.

27:47 Emma Sjöström: Bye.