Latest / PwC's accounting podcast / Accounting for the cost of externally marketed software

Transcript

Accounting for the cost of externally marketed software · PwC's accounting podcast

0:00
SPEAKER_00: Thought leadership from PwC's National Office.
0:06
SPEAKER_01: Hello, and welcome to PWC's Accounting Podcast.
0:10
SPEAKER_01: I'm Heather Horn.
0:12
SPEAKER_01: In today's episode, we'll kick off the two-episode series on software costs.
0:17
SPEAKER_01: And as we'll discuss, there are two different models depending on whether the software is to be used internally or externally.
0:25
SPEAKER_01: In today's episode, we focus in on the external use model, which is applicable to software to be sold, leased, or otherwise marketed as a separate product or embedded within a product or process.
0:37
SPEAKER_01: We're seeing this guidance come up more often as companies are using software in their products.
0:42
SPEAKER_01: And it's definitely an interesting topic, even if you're not dealing with this particular issue.
0:48
SPEAKER_02: Now, all companies have software.
0:50
SPEAKER_02: A lot of companies that were developing software to sell and ship to people are now developing these big, you know, software as a service platforms that they're monetizing, but they're basically their own systems.
1:02
SPEAKER_02: So they're now kind of in the internal use model sometimes.
1:05
SPEAKER_02: So we have this kind of convergence of software being used in all different flavors, and we have these sort of two very called maybe old models with some very different cuts at where you start and stop capitalizing costs.
1:26
SPEAKER_01: That was Pat Durbin, a deputy chief accountant in PwC's national office, who's joined by Mike Coleman, a partner in the national office.
1:34
SPEAKER_01: Both have years of experience consulting with clients and engagement teams on this topic.
1:39
SPEAKER_01: In addition, Mike has some perspective from his membership as part of the AICPA software task force.
1:45
SPEAKER_01: They have some great insights to share.
1:47
SPEAKER_01: So let's get started.
1:49
SPEAKER_01: So, Mike, Pat, welcome back to the podcast.
1:51
SPEAKER_01: Nice to have you guys back on.
1:53
SPEAKER_01: And some great topics for our little mini-series for doing this time.
1:56
SPEAKER_01: And specifically, I think there's been a lot of talk about AI and just the general pervasiveness of software and products and services.
2:05
SPEAKER_01: And so it's great to have a chance to understand the accounting guidance that's applicable.
2:11
SPEAKER_01: And so, Mike, maybe just to start things off, can you level set for us about where we are with that?
2:16
SPEAKER_03: Sure.
2:17
SPEAKER_03: I I think this topic is important for our listeners in virtually every company because pretty much every company has software or uses software.
2:27
SPEAKER_03: Now, having said that, there are different accounting models for the cost of developing or obtaining software.
2:35
SPEAKER_03: And that's really dependent upon how the entity plans on using the software.
2:40
SPEAKER_03: So whether a company plans to license the software to its customers or whether it'll run it internally to run the business, that'll help determine what the accounting model is.
2:51
SPEAKER_03: Today's discussion, we're going to focus on the accounting for costs of software that's intended to be sold, leased, or otherwise marketed, which is covered by ASC 98520.
3:02
SPEAKER_03: And so understanding whether you're in the scope of 985.20 is essential to make sure we're capitalizing and amortizing the right amounts and which can have significant impacts on financial statements.
3:14
SPEAKER_01: Yes, I think anything related to recording entries can have a significant impact.
3:18
SPEAKER_01: Yes.
3:18
SPEAKER_01: I'm pinning outside.
3:19
SPEAKER_01: But I think, Pat, it's an interesting point that Mike made because I think maybe not half the battle in accounting, but a big part of the battle in accounting is just making sure you're in the right model.
3:29
SPEAKER_01: And in particular, obviously that's the case.
3:32
SPEAKER_01: And I think here, what's interesting is how do you really define what constitutes the software to be sold, leased, and marketed, and then differentiate that from other software.
3:42
SPEAKER_02: Right.
3:42
SPEAKER_02: And I think as as you're getting at and what Mike was alluding to is there are basically two cuts, right?
3:48
SPEAKER_02: There's the externally marketed model or there's the internal use model.
3:52
SPEAKER_02: And then we'll talk maybe a little bit later about some ongoing standard setting.
3:56
SPEAKER_02: But if we start with the landscape today, making that cut of externally marketed, you really need to go look at the standard and how it defines what it means to be externally marketed, licensed, sold, leased, etc.
4:10
SPEAKER_02: Um, so it's it's essentially that the entity has a substantive plan to market the software.
4:16
SPEAKER_02: So you're really going into this thinking I'm building something that I'm gonna try to sell or monetize from others.
4:23
SPEAKER_02: Once you have that arrangement with the customer, the customer is going to be able to sort of take possession of the software, meaning you're gonna have a license, you're gonna deliver to it the software that it will essentially run on its own servers.
4:36
SPEAKER_02: And that's kind of the critical distinction usually between whether you're in the externally marketed model or the internal use model.
4:44
SPEAKER_02: If you never transfer possession, if you will, of the software, of the code, the customer doesn't have the ability to run it on its own servers, you're gonna be stuck in the internal use model.
4:54
SPEAKER_02: But if you're gonna transfer that software, you're going to be in the externally marketed model.
4:59
SPEAKER_02: I think it's particularly interesting in this environment where there's an increasing prevalence of software solutions being delivered in the software as a service environment.
5:09
SPEAKER_02: Many traditional software companies are finding themselves actually in the internal use model for these software tools that are delivered through a purely hosted solution because they're never transferring the software to their customer.
5:25
SPEAKER_01: All right.
5:37
SPEAKER_01: And maybe this is not really a scenario we see, but if a company does develop something for its internal use and it realizes, oh, this is fantastic.
5:45
SPEAKER_01: I do want to put more into it and market it, et cetera.
5:48
SPEAKER_01: Can you do ever switch models or is that the question?
5:51
SPEAKER_03: Yeah, so the the accounting guidance will cover that scenario.
5:55
SPEAKER_03: So as Pat mentioned, your evaluation is what your plan is going to do, right?
6:01
SPEAKER_03: Because you're making the software with a plan to weather market it or keep it internally.
6:07
SPEAKER_03: So if you had a plan to just use it internally and you were the internal use standard, but then you decided, as you said, hey, I'm going to sell this, the accounting model will say, well, you started building this asset as if it was your own asset, like PPE, even though it's not PPE, uh property, plant, and equipment.
6:27
SPEAKER_03: I know.
6:27
SPEAKER_01: Yes, I do know that.
6:29
SPEAKER_01: But it's good.
6:30
SPEAKER_01: I don't like acronyms, so thank you.
6:32
SPEAKER_03: So if you're capitalizing under that standard and then you choose to sell, the accounting model will say, Well, what do you normally do when you sell a fixed asset or PPE?
6:45
SPEAKER_03: You record a gain and you reduce your basis.
6:48
SPEAKER_03: So the accounting model would say, if I decide then, hey, I want to then market it, what would happen is if you had capitalized any software balance, any proceeds in your sale would reduce that balance.
7:01
SPEAKER_03: And then thereafter you start recording revenue.
7:05
SPEAKER_03: If you made that switch before it was saleable, this the asset would basically transfer over.
7:12
SPEAKER_03: It gets a little tricky as to whether you have to then re-evaluate what's capitalized or not if I make the call before it's released.
7:21
SPEAKER_03: Most companies make the call after it's released.
7:24
SPEAKER_01: Okay.
7:24
SPEAKER_01: And I'm assuming this is not a huge scenario, but it's just anytime we talk about those two criteria, I at least didn't want to miss the first one.
7:32
SPEAKER_03: We'll talk about it.
7:33
SPEAKER_03: Yeah.
7:34
SPEAKER_03: The bigger scenario is because companies that have sold licensed software and then have moved to CES because that's the direction they're going.
7:43
SPEAKER_03: So the bigger thing is I have a plan to sell it, and then I decide I don't want to sell it anymore.
7:49
SPEAKER_03: Now I want to just use it internally, and we'll talk about that circumstance.
7:52
SPEAKER_01: Well, and I think, Pat, I left us with a little bit of a cliffhanger because you said basically it's being hosted by the seller.
7:59
SPEAKER_01: And so then you never actually do, the the customer never does take possession of it.
8:06
SPEAKER_02: Right.
8:06
SPEAKER_02: And I think it's we're in an interesting time because companies, in many cases, will still provide the software or an option to take the software, run it on your own servers.
8:17
SPEAKER_02: And as long as there's sort of any potential that somebody's going to take possession of the software, you're going to market it and sell it in that model, then you are still in the externally marketed model.
8:29
SPEAKER_02: It's only if you're exclusively a SaaS provider that you would be in the internal use model.
8:36
SPEAKER_02: And this is why I think we'll talk about it a little bit more later, but the FASB standard setting project is sort of relevant and timely.
8:44
SPEAKER_01: Aaron Powell Yeah, because I guess in the scenario you're talking about, then there's an incentive to the provider to give the option for someone to host it on their own servers or otherwise so that there is clear that you're in this model.
8:57
SPEAKER_02: Aaron Powell I don't know.
8:58
SPEAKER_02: It's probably really more a question of what the company thinks is the most economically beneficial approach for them.
9:04
SPEAKER_02: I don't think they're really motivated by which accounting model they want to be in.
9:08
SPEAKER_01: Aaron Powell Okay, fair point.
9:11
SPEAKER_01: I know that accounting shouldn't be driving economics.
9:14
SPEAKER_03: So or business.
9:16
SPEAKER_03: Not to go too far into a different direction.
9:18
SPEAKER_03: The the bigger motivation, at least for the seller, tends to be how they recognize the revenue.
9:24
SPEAKER_03: And so SaaS companies want a revenue over time as a service.
9:29
SPEAKER_03: So they're not motivated to give the customer the right to sell the software because from their perspective, then it's done.
9:35
SPEAKER_03: Now I've sold a license and that's a different revenue model.
9:38
SPEAKER_03: So there is attention there.
9:40
SPEAKER_03: Most of the times the customers don't really care one way or the other because they get the use of the software.
9:45
SPEAKER_01: And that's what they care about.
9:47
SPEAKER_03: And that's why this topic is important because how they use it is driving how the accounting is.
9:52
SPEAKER_03: So that's what we're going to get into.
9:54
SPEAKER_01: Okay, good.
9:54
SPEAKER_01: And I do think it's helpful to sort of understand some of the backstory before you start talking about the accounting, because otherwise it's just debits and credits without sort of a logic to it.
10:03
SPEAKER_01: So with that all said, why don't we start talking about the accounting model?
10:07
SPEAKER_01: And so let's assume now for the this discussion that you have concluded that you're in the externally marketed software model.
10:16
SPEAKER_01: Then what comes next, Mike?
10:18
SPEAKER_03: Okay.
10:18
SPEAKER_03: So um the guidance that you're in is ASC 99520.
10:22
SPEAKER_03: And and the the big thing with that guidance is there's a decision point about when you start capitalizing the cost related to the software that uh you're developing or in in cases purchasing.
10:39
SPEAKER_03: And that is when you reach technological feasibility in your development.
10:45
SPEAKER_03: Okay.
10:45
SPEAKER_03: And I I think the determination of what constitutes technological feasibility is key, again, because any development before that point gets expensed as RD.
10:58
SPEAKER_03: Any costs that are incurred after that point are subject to capitalization until the software is then released, and then any cost incurred after the software is released becomes an expense like maintenance.
11:12
SPEAKER_03: So the notion of technological feasibility is the key thing in 98520 for externally marketed uh software.
11:21
SPEAKER_01: All right.
11:22
SPEAKER_01: And so since that is the key point here, then Pat going back to you.
11:25
SPEAKER_01: How do you determine at what point that's reached, especially with software, which I feel like is constantly v1, v1.1, v1.1a, it's kind of a continuum at Suitsek.
11:36
SPEAKER_02: Well, it's interesting you you bring that up because this guidance that we're going to talk about was written in a time frame when, yes, there were different versions of software, but they happened over a much longer period of time and there was sort of a long planning process.
11:51
SPEAKER_02: You know, computer processing time was at a premium.
11:54
SPEAKER_02: So you sat around and did a lot of work and thinking and writing algorithms and um flow charts before you ever really did any coding or like true kind of development work.
12:05
SPEAKER_02: Um that's not how software gets developed today, and thus it's a challenge to apply this model.
12:11
SPEAKER_02: But the concept of technological feasibility as it was described in the standard is the point at which you've completed all of the planning, designing, coding, and testing activities that are necessary to establish that the product can be produced to meet its design specifications, including all of the functions, features, and technical performance requirements.
12:33
SPEAKER_02: So that's, I know you don't like to read, but that's right out of the standard, just to level set the definition of technological feasibility.
12:39
SPEAKER_02: And then it goes on to give a couple of criteria, I guess, that you would evidence how you've achieved technological feasibility.
12:48
SPEAKER_02: So you've completed a detailed program design, um, meaning you've kind of got this big master plan for how the software is going to be developed.
12:57
SPEAKER_02: And any high-risk development issues that might have been identified have been resolved through testing, like you think you can accomplish it now.
13:06
SPEAKER_02: That's one way.
13:07
SPEAKER_02: Or you've actually built a working model.
13:10
SPEAKER_02: And that's probably more typically how people are going to look at it today because this happens in a much more iterative, dynamic environment.
13:18
SPEAKER_02: Um so you've got a working model that has functionality and it's consistent with the design you had laid out.
13:26
SPEAKER_02: So those are the words in the standard.
13:28
SPEAKER_02: How you apply them in practice is I think where the real challenge is.
13:32
SPEAKER_02: But the important point is when you've achieved technological feasibility is a question of fact.
13:38
SPEAKER_02: It's not an accounting policy decision, meaning you have to demonstrate for each software project, am I a technological feasibility or am I not?
13:48
SPEAKER_02: Because that's the point at which, as Mike was saying, prior to that you expense the cost.
13:53
SPEAKER_02: After that point, you capitalize the cost, and then subsequently amortize them as you sell or monetize the software.
13:59
SPEAKER_01: Yeah, it's interesting, Pat, because as as you were talking, I was thinking, we do see other places where sort of what the business is doing and what accounting needs, there is a divergence.
14:09
SPEAKER_01: But I can in particular see the discussions here between the software developers and the controller or their team.
14:15
SPEAKER_01: And the controller is saying, okay, well, have we, do we have this working model?
14:19
SPEAKER_01: Have we met this?
14:19
SPEAKER_01: And then saying, well, we're still testing, we're still doing this.
14:23
SPEAKER_01: And so I guess as we really think about sort of that divergence between what's happening in fact and what's required for accounting, then how are we applying those concepts in practice?
14:34
SPEAKER_02: Yeah.
14:34
SPEAKER_02: So I think I made the point that it's not just an accounting policy decision, but I think as a practical matter, in the context of how a particular company operates its development process, they're going to have some procedures and policies that sort of define for them when they've met technological feasibility.
14:54
SPEAKER_02: So maybe it's a bit of a semantics distinction, but I think you have to figure out some way to sort of operationalize those accounting concepts in the context of how your business is operating.
15:05
SPEAKER_01: Yeah, no, I definitely think that makes sense that this is a place where some the accountants can bring maybe a little order to this process to help with that.
15:12
SPEAKER_01: And so I think, Pat, you talked about again just the difference between how software was developed when this model was put in place in the way it is now.
15:21
SPEAKER_01: But if we're thinking then about putting in place some policies about what a working model is, what does that look like?
15:28
SPEAKER_02: So the the standard defines a working model is basically an operative version of the software that's completed in the same language that's a software product will ultimately be produced again.
15:39
SPEAKER_01: Okay, I have to almost laugh.
15:40
SPEAKER_02: Right.
15:41
SPEAKER_02: We're going back to like, you know, the days of whatever early programming languages and there weren't a place, big number of choices.
15:47
SPEAKER_01: But also the thought, like, okay, I'm gonna write it in one program, but let's translate it before we anyway.
15:53
SPEAKER_01: Okay, sorry.
15:53
SPEAKER_02: I think that had a lot to do with you know functionality back in the day.
15:56
SPEAKER_02: Exactly.
15:57
SPEAKER_02: Um so it can perform all the major functions that you had planned for it, and it's actually ready for customer testing.
16:03
SPEAKER_02: So we might call that like the beta release, right?
16:05
SPEAKER_02: So it's not just in the lab, like we could sell it to people, they could start using it, maybe we're gonna give give some disclaimer that it's still in beta mode.
16:15
SPEAKER_02: But you know, again, this is an old definition.
16:18
SPEAKER_02: So applying it in practice, you know, the agile development environment that we have uh at most companies today makes it very challenging.
16:27
SPEAKER_02: You know, you really have to think through, okay, where are we at in this process?
16:31
SPEAKER_02: And I think the practical reality is there's a very short window between when you believe you have a working model and when that software is available for release, general release to customers.
16:42
SPEAKER_02: So you end up with a very narrow window of when costs will actually be capitalized.
16:47
SPEAKER_01: So then, Pat, again, I can see a lot of challenges here because thinking even when you talk about the fact of major functions planned, if you think about the way software planning is, often you start with you're gonna have 10 items, you wind up with four items.
16:59
SPEAKER_01: And so is that major or not?
17:02
SPEAKER_01: But assuming then that you have reached this technological feasibility, then during this window, what types of costs are capitalized?
17:12
SPEAKER_02: So I think the general concept is you could think of it sort of like we think about one of my favorite topics, inventory costing.
17:19
SPEAKER_02: It's sort of the productive costs you incur to produce the thing.
17:23
SPEAKER_02: So in the context of software, that's mostly going to be people's time.
17:27
SPEAKER_02: So it's the actual developer's time.
17:29
SPEAKER_02: If you're purchasing third-party services to support that effort, those would be um eligible for capitalization.
17:36
SPEAKER_02: It doesn't really include a lot of detailed guidance on the examples of those types of costs, but I think those are the way we would think about it is it's the costs that are directly related to the software development.
17:48
SPEAKER_01: But so how it's also interesting because again, if you think about software development, you're finishing the first version even as you're working on the second version.
17:58
SPEAKER_01: So how does that work?
17:59
SPEAKER_01: And you're gonna have to do some again, you made me think of it with your inventory allocation or your inventory reference.
18:05
SPEAKER_01: Do you have to do some type of allocation then and say, well, this is related to finishing up the first one?
18:10
SPEAKER_03: Or yeah, so what winds up happening, and you know, not not to beat a dead horse per se, but again, when this was was written in that working model definition, it was meant to be objective to the development process, but it was objective to the development process at the time.
18:26
SPEAKER_03: So now that the development process has really shortened and you start working on the next thing fairly quickly, the the standard does address what do you do if you create the next version and you have a capitalized balance related to the prior version.
18:47
SPEAKER_03: And it has this notion of shared code.
18:50
SPEAKER_03: So when you develop the next version, often you're not developing it from scratch.
18:55
SPEAKER_03: You're using some of the code from the prior version, and then you're adding some new features and functionality and deleting some code.
19:02
SPEAKER_03: So the standard basically says if you're going to continue selling both of them, then the capitalized balance needs to be allocated between the two software programs, and then the incremental capitalized cost gets added to that.
19:18
SPEAKER_03: What's not there is how do you allocate it?
19:22
SPEAKER_03: It's the allocated based upon lines of code.
19:24
SPEAKER_03: It's the remaining balance, so it's not the initial piece.
19:28
SPEAKER_03: And what if you only take a portion of it?
19:29
SPEAKER_03: So that's where there's a lot of judgment and it gets a little difficult to do.
19:34
SPEAKER_03: The the nice part about it is since the working model is so late in the development cycle here, you don't get so much cost on the balance sheet to then have to allocate anyway.
19:46
SPEAKER_03: So that's a good thing.
19:48
SPEAKER_03: But if you're in that circumstance, it's a judgment as to what's a reasonable ration method to allocate.
19:54
SPEAKER_01: And then one of the key points here is that we talked about the beginning is which model you are in between the external and the internal.
20:01
SPEAKER_01: And so what are the differences between the types of costs that you can allocate in those two models?
20:05
SPEAKER_03: Aaron Powell Yeah.
20:06
SPEAKER_03: Uh probably the biggest difference between the two, uh as Pat was discussing, this notion of technological feasibility is an externally marketed software thing.
20:15
SPEAKER_03: It's meant to be an objective standard in the development process.
20:19
SPEAKER_03: Internal use software is well, when do you think it's probable?
20:23
SPEAKER_03: And so you you know, it says you need to exit planning, right?
20:27
SPEAKER_03: So once you've exit planning, and then it becomes probable you're going to finish and use the software, that's the point you start capitalizing.
20:38
SPEAKER_03: And that is a big judgment in the standard as to what would be considered probable.
20:45
SPEAKER_03: Right.
20:45
SPEAKER_03: So that's probably the biggest difference between the two relative to what is allowed to be capitalized.
20:54
SPEAKER_03: So what we wind up seeing in the internal use guidance, because it's a probability notion, not an objective measure of the development process that happens later, more gets capitalized in the internal use world.
21:07
SPEAKER_02: Aaron Ross Powell I mean, if you think about the sort of history of this guidance again, going back to the externally marketed guidance, that was developed in like the 1980s when you know a lot of computing was still happening on like mainframes, you know, software was being you know printed on disks or CDs and shipped to people, like there wasn't all this online cloud activity.
21:31
SPEAKER_02: And then the internal use model was really designed for when companies started building big in-house ERP systems to run their business, right?
21:42
SPEAKER_02: So they weren't software companies, they were just normal commercial industrial companies that were putting software in to run their business, right?
21:49
SPEAKER_02: And spending a lot of money doing it.
21:51
SPEAKER_02: So it was kind of like borrowed from a PPE model.
21:54
SPEAKER_02: So it kind of had two very different purposes.
21:57
SPEAKER_02: Now all companies have software, a lot of Companies that were developing software to sell and ship to people are now developing these big software as a service platforms that they're monetizing, but they're basically their own systems.
22:11
SPEAKER_02: So they're now kind of in the internal use model sometimes.
22:14
SPEAKER_02: So we have this kind of convergence of software being used in all different flavors.
22:21
SPEAKER_02: And we have these sort of two very call it maybe old models with some very different cuts at where you start and stop capitalizing costs.
22:30
SPEAKER_01: Well, and you guys have alluded to some new developments.
22:35
SPEAKER_01: And I have to say, I feel like this is like a book that I just want to get to the end so I can find out what happens.
22:40
SPEAKER_01: But before we get there, because I really uh do think that's going to be key uh to all of this, maybe we should, I think we've hit a lot of these, but what are some of the other practical challenges with the current model, Mike?
22:51
SPEAKER_03: Sure.
22:52
SPEAKER_03: Well, some of the ones that we hear uh quite often are first is again the judgment as to what constitutes a working model isn't always easy.
23:00
SPEAKER_03: Um a lot of companies say beta, a lot of companies look at a prototype.
23:05
SPEAKER_03: Um but even that means different things to different companies.
23:09
SPEAKER_03: So uh trying to figure out when you meet the working model becomes becomes difficult and and a little bit nuanced.
23:17
SPEAKER_03: So you always try to go back to what is the standard trying to make sure you achieve.
23:22
SPEAKER_03: I've got something that's operational in the language that is going to sell, that has everything that I think I've planned to do, and it's ready for customer testing.
23:32
SPEAKER_03: So whatever your definition is has to have those characteristics to get to working model.
23:38
SPEAKER_03: The next one we deal with is a lot of times companies will develop software, but then we'll buy pieces of functionality from some other party to then integrate into their broader program.
23:52
SPEAKER_03: And so that question becomes well, is that its own software program?
23:57
SPEAKER_03: But it's being integrated into a larger program.
24:01
SPEAKER_03: And so there's two considerations that companies need to work through.
24:05
SPEAKER_03: One is whether there is an alternative use for the software.
24:09
SPEAKER_03: Um, and that would suggest you have to account for that alternative use, not just your development.
24:14
SPEAKER_03: And then the second is, well, when did you incur that cost?
24:18
SPEAKER_03: Was it before you hit technological feasibility or after?
24:21
SPEAKER_03: Because that's going to indicate whether you capitalize or not.
24:24
SPEAKER_03: So that's something that we get into a fair amount.
24:27
SPEAKER_03: A couple others, uh, that onamortized code sharing thing we talked about.
24:32
SPEAKER_03: So that that we get a fair amount of discussion on.
24:36
SPEAKER_03: At an operating level, um, operational level for companies, it's the tracking of costs.
24:41
SPEAKER_03: So in through this process where you have people developing, it's making sure there's some way to understand what they're doing from a nature of the cost perspective.
24:52
SPEAKER_03: And then did we hit that?
24:54
SPEAKER_03: So do the accountants know which cost was incurred after the developers hit technologic feasibility or not?
25:01
SPEAKER_03: So the whole process around trying to accumulate that information becomes difficult operationally.
25:08
SPEAKER_03: And then the last one would be uh that we get a lot of questions on is a software company is going to be licensing software and then wants to move to Saa.
25:19
SPEAKER_03: And so the notion is okay, I'm gonna move to SaaS in my next version.
25:23
SPEAKER_03: And so I have a plan to just sell it via SaaS.
25:28
SPEAKER_03: Okay, well, what about a customer that has the first version and then has a subscription or PCS contract, post-customer contract support, sorry, thank you clause, maintenance clause that enables them to get the additional versions and take possession of them?
25:47
SPEAKER_03: So when the company has the plan to just do SaaS, but if they have the requirement to deliver for a payment that same software to a customer who has been a licensed customer, does that mean they're still marketing the software?
26:03
SPEAKER_03: So it's those types of things in the switch to SAS.
26:06
SPEAKER_03: When is the cutoff really happened?
26:08
SPEAKER_01: Aaron Ross Powell And Mike, do we see cases where you're getting questions from companies that would just say, I just want to expense everything?
26:16
SPEAKER_01: Or I mean I which I know you can't, but it just seems like given the complexity here, there could be some incentive.
26:22
SPEAKER_01: Trevor Burrus, Jr.
26:22
SPEAKER_03: Yeah.
26:22
SPEAKER_03: So so that question we get and we get a lot, and we get it in the frame of, hey, I've analyzed my development process, and my working model comes very late in the game.
26:35
SPEAKER_03: And the only thing is a little bit of work on the testing side.
26:38
SPEAKER_03: So there's really not anything to capitalize.
26:40
SPEAKER_03: So I'm just going to expense everything.
26:42
SPEAKER_03: It's like, okay, that's not your like it can't be your policy just to expense.
26:48
SPEAKER_03: So you still need to keep your pulse on what your development process is, if it changes at all.
26:57
SPEAKER_03: Where you may have working models that you get to quicker is when I do version one point whatever.
27:07
SPEAKER_03: I may get to a working model of that faster because it's been developed, and I may be adding some functionality that I know works and just combining them.
27:17
SPEAKER_03: So you so you can't just say expense.
27:20
SPEAKER_03: You have to do work to say, all right, my point in my development process is very late.
27:27
SPEAKER_03: So you can't abandon all of the process tracking.
27:32
SPEAKER_03: Well, and that's you have to actually still understand what everybody's doing and make sure your point in time is late enough where you don't have much to capitalize.
27:42
SPEAKER_01: Yeah, and that's actually why I asked that question, because I wanted to make sure it was very clear that you can't sort of default to that, but you still do have to make your way through this model, even though it may not align very well with current day.
27:54
SPEAKER_01: And I guess speaking about aligning with current day, then I think the other question that's probably top of mind is you know, you can't open the, I'll say newspaper, although I guess it's really the newspaper app without seeing headlines about AI, this, AI, that, AI models and everything else.
28:10
SPEAKER_01: And so how does the development of AI models then intersect with this as well?
28:16
SPEAKER_03: So we'll we'll take because we're talking about externally marketed software.
28:20
SPEAKER_03: So first, AI isn't, it may be its own class of software relative to what it does, but from an accounting perspective, AI is just another type of software.
28:30
SPEAKER_03: So what is going to matter first is what are you going to do with the AI?
28:35
SPEAKER_03: If you are going to externally market it, it follows the externally marketed standard.
28:39
SPEAKER_03: If you're going to use it to be just internal and use it for tools or provide service, then you follow the internal model.
28:48
SPEAKER_03: So if you're file, if you're developing AI in the external model, the same model applies.
28:54
SPEAKER_03: You're going to figure out when you have technological feasibility, the costs incurred before tech feasibility happens, those are expensed, after are capitalized.
29:05
SPEAKER_03: The big question we get is relative to AI development.
29:10
SPEAKER_03: Normally, that development requires a lot of data to train the AI engine.
29:16
SPEAKER_03: And so often that data is purchased.
29:19
SPEAKER_03: And so what we've got is many companies who don't capitalize a lot, their first reaction is well, the data I purchase is just going to be expensed because I'm going to purchase it before tech feasibility and I'm just going to expense it and move on.
29:36
SPEAKER_03: That's typically not how the accounting winds up being for two reasons.
29:41
SPEAKER_03: One, the data itself is its own intangible asset apart from software development.
29:46
SPEAKER_03: It's not just coding.
29:48
SPEAKER_03: So there's an argument to be had that it's a purchase of an intangible that gets capitalized on the 350 generally.
29:56
SPEAKER_03: And then the other thing that we see is in the RD standard is when companies are developing and purchase intangibles for development.
30:07
SPEAKER_03: If those intangibles have an alternative use, and an alternative use would be other development projects, it's required to be capitalized under 350 and the amortization becomes development.
30:19
SPEAKER_03: So what we typically see in practice is when a company is developing AI and buying data, they don't restrict the purchase of the data to one AI development project.
30:30
SPEAKER_03: It's normally data that's bought with rights to do multiple AI trainings over some period of time, and sometimes even to sublicense the data.
30:41
SPEAKER_03: So when that happens, typically an alternative use exists.
30:45
SPEAKER_03: So then the data gets to be capitalized.
30:48
SPEAKER_03: So the big takeaway from this is if you're developing software that's AI software to be marketed, yeah, the costs of the software development that are the traditional are going to follow the same type of model that you would always do.
31:03
SPEAKER_03: Any data that's purchased will be an intangible purchase to develop the software.
31:08
SPEAKER_03: If it has an alternative use, you're gonna have to capitalize.
31:11
SPEAKER_01: All right.
31:12
SPEAKER_01: A lot to think about there.
31:13
SPEAKER_01: But now we're finally going to get to where I actually have wanted to jump ahead for this entire episode, which is what's going on with the FASB project.
31:21
SPEAKER_01: Because it definitely seems like it's time for some new guidance.
31:24
SPEAKER_03: Yeah.
31:24
SPEAKER_03: So the FASBE project, they it's to, you know, to they took on a project for software cost, and the goal was to modernize the standard.
31:35
SPEAKER_03: And for those listening at home, I am using air quotes around modernize.
31:40
SPEAKER_03: And that gets back to when the guidance was written.
31:42
SPEAKER_03: It was written at a time when there was a linear waterfall type of approach, and most companies have moved to an iterative, agile type of approach.
31:51
SPEAKER_03: And so the question is we want to modernize to have the software standard accommodate those different types of development approaches for the most part.
32:00
SPEAKER_03: And we also noticed that there is different outcomes if you account for software development in external versus internal.
32:09
SPEAKER_03: And so there was a thought that said, well, maybe one of the ways to eliminate some of those differences is to have one standard.
32:16
SPEAKER_03: So that was one of the early thoughts is hey, maybe we can reduce some complexity by having one standard for uh software development.
32:26
SPEAKER_03: You know, what wound up happening, and I think there's like a story, an interesting thing happened on my way too.
32:33
SPEAKER_03: So an interesting thing happened on the way to developing this new accounting model, is through all the research and and and work that the FASB did, they realized most of the complexity was driven by applying 350-40 to more commercial type Says applications.
32:52
SPEAKER_03: And they actually realized that 985 was flexible enough, even with all those long definitions of what tech feasibility and working model is, and was producing outcomes people liked.
33:06
SPEAKER_03: Not a lot of capitalization and people could work with that and understand that.
33:11
SPEAKER_03: So what the FASBE ultimately decided to do was not have one model for all, is keep the two models and leave 985-20 untouched.
33:23
SPEAKER_03: And so the the project now is a focus on targeted improvements of the 350-40 internal use model.
33:31
SPEAKER_03: So effectively, 985 has no change to it.
33:35
SPEAKER_01: All right.
33:36
SPEAKER_01: So, Mike, just to clarify then for our listeners, because we started throwing a lot of numbers in there, which I know we're accountants, but so 985 is the model we've been talking about.
33:44
SPEAKER_01: Right.
33:45
SPEAKER_01: And then 35040 is the model for internally developed software.
33:49
SPEAKER_01: So we've spent the episode talking about sort of this misalignment between the way the model is set up and the way that software is actually developed.
33:59
SPEAKER_01: But if I understood you correctly, they're not changing the model?
34:02
SPEAKER_03: That's right.
34:03
SPEAKER_03: What they've what they've concluded in doing all their outreach with the software preparers, the developers, and the community that does the buying is that while the guidance was written for a linear approach, over time practice has sort of uh evolved to give answers that people actually like.
34:27
SPEAKER_01: And that makes sense in the circumstances and that are good information for users, all those types of things.
34:32
SPEAKER_03: The software developers have interpreted what working model means today in an agile environment sufficiently enough that the answers that preparers and investors are getting is what they like and don't really want that changed.
34:49
SPEAKER_03: And so the decision was: well, if everybody's happy with the results of that and how practice has interpreted that over time, we're just gonna leave that one be.
35:28
SPEAKER_02: I think some of the feedback they heard from users was we don't really care that much about any asset that would end up on your balance sheet anyway, because people are going to make different decisions.
35:39
SPEAKER_02: Ultimately, what we'd like to know is what companies are spending to develop software.
35:45
SPEAKER_02: And so we might expect to see some more disclosure about that.
35:48
SPEAKER_02: But I think that's how you kind of bridge, because you sort of start with this is an old out-of-date model, and you're telling me they thought it was fine.
35:55
SPEAKER_02: Right.
35:55
SPEAKER_02: Right.
35:56
SPEAKER_01: Right.
35:56
SPEAKER_01: Yeah.
35:56
SPEAKER_01: And I guess yes, that's a helpful perspective, Pat.
36:00
SPEAKER_01: And I I know Mike, you were on the working group, and this can be a Mike Coleman opinion, not a PWC opinion.
36:07
SPEAKER_01: But from your perspective, having worked with these models and then being part of the working group, what's your view on this path that we're going down?
36:14
SPEAKER_03: Aaron Powell Yeah, I I I think I'm comfortable with the path.
36:18
SPEAKER_03: I think it's a good idea.
36:19
SPEAKER_03: It's certainly much easier to apply.
36:23
SPEAKER_03: I I look at uh a model where people like the results, people know how to do it, people have been doing it for a while, to change just to change that.
36:33
SPEAKER_03: Or to change to, all right, we're gonna change the words to match it to what you're doing today, but the outcome be the same.
36:40
SPEAKER_03: Okay, I I I like the it's better left where it is, and put the focus in on what is really causing the difference, which is well, now I'm not transferring my software, I'm now providing a service, and that model's much harder to apply to that.
36:57
SPEAKER_01: Okay.
36:57
SPEAKER_01: I do slightly feel like I should have read the ending of the book, but that's okay.
37:02
SPEAKER_02: So uh but it's not published yet, so you can't apply it.
37:05
SPEAKER_01: So that's also helpful.
37:07
SPEAKER_01: So then with all of that said, you guys both know the last question that's coming, which is just some final pieces of practical wisdom.
37:13
SPEAKER_01: So, Mike, maybe I'll go to you first.
37:15
SPEAKER_01: That in this period while we're waiting for the exposure draft, or it sounds like even after we see the exposure draft potentially, what do you recommend for companies that are dealing with some of these issues?
37:24
SPEAKER_03: Aaron Powell Yeah, I I I do think uh the biggest thing is some of the last things that we talked about.
37:30
SPEAKER_03: So I'll just reinforce those, which is you can't just expense, you can't say I'm externally marketed and my policy is expensed.
37:38
SPEAKER_03: So make sure you're keeping your eye on the development change.
37:42
SPEAKER_03: We've seen one development change over a period of time, or many development changes over a period of time to get to here.
37:47
SPEAKER_03: It's not to say that's gonna happen again.
37:49
SPEAKER_03: So this is a development focused standard, so keep your your eye on that.
37:54
SPEAKER_01: And how about Pat from your perspective?
37:56
SPEAKER_02: Yeah, I would just maybe uh highlight the points Mike made around the costs of AI development and making sure you understand what you're talking about, because AI is just another evolution, maybe revolution of software, but the data component is not necessarily just software development.
38:15
SPEAKER_02: It's got its own existence as an asset, and you need to pay attention to you know specifically what the costs are that we're talking about.
38:22
SPEAKER_01: All right.
38:22
SPEAKER_01: Well, gentlemen, as always, very interesting to talk to you.
38:25
SPEAKER_01: Great topic.
38:26
SPEAKER_01: So thanks so much for joining me today.
38:28
SPEAKER_03: Thanks for having us.
38:30
SPEAKER_01: That's our show for today.
38:32
SPEAKER_01: Tune in next week for more fresh episodes.
38:35
SPEAKER_01: So that you never miss any of our audio content.
38:38
SPEAKER_01: Follow the PWC Accounting Podcast wherever you listen to your podcasts.
38:42
SPEAKER_01: And to stay up to date on all our latest accounting and reporting news, sign up for our newsletter at viewpoint.pwc.com.
38:52
SPEAKER_01: From Thought Leadership at PwC, I'm Heather Horn.
38:56
SPEAKER_01: Thanks for tuning in.
39:08
SPEAKER_00: PwC refers to the U.S.
39:10
SPEAKER_00: member firm or one of its subsidiaries or affiliates, and they sometimes refer to the PWC network.
39:16
SPEAKER_00: Each member firm is a separate legal entity.
39:18
SPEAKER_00: Please see www.pwc.com slash structure for further details.
39:25
SPEAKER_00: This podcast is for general information purposes only and should not be used as a substitute for consultation with professional advisors, including accountants and lawyers.