Navigating AI monetization with Zuora Revenue

And, um, before we get started, I just wanted to share our, um, forward-looking statement. So just as a precautionary measure, our legal team asks that we share this on all our presentations. I won’t read it word for word, but please take a moment to read our forward-looking statement. Great. Thank you. And moving on, we are going to just go over a couple of housekeeping items. So please introduce yourself if you haven’t done so already. Please use the chat feature. It’s located at the bottom. We do have, um, a links in our resources tab. So if you click More, there is a Resources tab, and within that, you’ll see, um, a couple of documents, links that we’ll be sharing throughout and referencing through

the presentation. Use the chat function to ask questions. We are monitoring the questions, and we do have a devoted Q&A section at the end in which we will, uh, have our Q, uh, Q&A section portion. We’ll address those questions. If we can’t do so during that portion, we’ll do so, um, afterwards. We are recording this session, so we’ll share a follow-up email with the recording. And please, uh, partake in our poll questions. You’ve just taken one poll. We’ll have a couple more during the session, so please join in and answer the poll questions. Thank you so much. And next, I’ll be introducing our speakers. So I’m Lila Kambrosney, I’m a senior product marketing manager, and on this webinar today with me is Aaron Bueller, a solution consulting strategist. Aaron will be going over the topic of breakage and overage today, as well as presenting the demo

and then joining us for the Q&A section as well. So I’m gonna pass it to Aaron. All right, thanks, Lila, and, uh, welcome everybody. Uh, today we’re looking at breakage and overage in usage-based models, especially as it pertains to AI credits, because that’s what makes this topic particularly hot right now. Uh, but most of what we’ll discuss is applicable to any usage scenario. Uh, I think we’ve kind of kicked this poll off already. Uh, the reason we ask this is that accounting complexities make it difficult to tackle issues reactively, and, um, although accountants have no interest in delaying, you know, go-to-market opportunities, they also don’t wanna be caught off guard. And so we’re just curious to see how everyone,

uh, uh, feels about their involvement in some of those kind of things, uh, as they’re being, uh, set up. Uh, so just to kick it off, let’s, uh, talk about the concepts behind today’s discussion. Breakage and overage basically compare a customer’s commitment with the actual usage. Breakage, of course, is below the commitment. In other words, the customer is prepaid for rights or credits that may go unused. And overage, on the other hand, is usage above the commitment. Um, in other words, the customer consumes more than is included and may owe an additional amount. Conceptually, it’s pretty simple, but, uh, why is this a worthy webinar topic? Um,

both scenarios can affect the revenue recognition process, and they are especially challenging in AI models because historical data may be limited, customer behavior can change quickly, and your company’s approach may change frequently, even as we try to figure out the best way to sell this. So, uh, furthermore, the GAAP, uh, requirements can be complex, especially if they interact with other difficult rev rec areas like SSP allocations, contract modifications, et cetera. So let’s take a couple of minutes to walk through each area, and we’ll start with breakage. Uh, for breakage, the central question is: When should the unused prepaid balance

be recognized? In practice, there are three ways to think about it. There’s proportionate, a remote likelihood, and upon expiration. And let’s take them one by one. Uh, proportionate recognition. Un-under this method, if breakage can be reliably estimated and is not subject to significant reversal, then breakage is recognized in proportion to the customer’s actual usage. Now, this is theoretically elegant, but in our experience, rarely used because it requires those reliable estimates to begin with, uh, the ability to inject those estimates efficiently into the existing data streams, and then rinse and repeat throughout the contract term as those estimates change.

So particularly in the case of AI, it is difficult to justify this approach. Another alternative, uh, method i– allows for recognition of breakage when the likelihood is remote that the customer will use those remaining rights. Uh, this requires judgment, especially if the determination occurs before the term expires, but it can be a little bit more practical to execute. And then finally, upon expiration is a common operational application, uh, to recognize that breakage when the credits expire or the obligation is otherwise released. I suppose technically this could be considered a

special case of the remote likelihood method that we just discussed, um, just executed when the likelihood is near zero, right at the very end. Uh, most of our clients use automated rules to recognize any remaining revenue at term end date. But even here where it seems like, oh, this is a pretty simple model to employ, the lack of history may play a role, uh, especially for newer AI models. For example, will your company enforce the expiration, or will it grant some kind of rollover or a grace period, for example? So even at expiration, sometimes there’s accounting questions that still need to be, um, answered. All right, so before we continue on with the overage side

of things, uh, let’s pause for another poll. And this one is, how does your company establish those commitment levels? Um, are the commitments generally set, you know, below expected usage, above, or maybe you don’t even have, um, defined commitment in your scenario? Um, obviously it can make sense for both the seller and the buyer to establish a purchase commitment. Uh, typically, higher commitment levels can come with reduced per unit pricing. Uh, but customers can be wary of over-committing. So we work with companies who deal with really each of these, uh, as their typical scenario.

Okay. So generally, looks like we’ve got some level of commitment, uh, that we’re, that we’re working with here. Okay, I’m gonna continue on. Um, oops. Let’s, uh, talk about overage. So overage, of course, is on the other side of the, uh, coin, uh, when actual usage exceeds the customer’s commitment. So if the contract already specifies the price for this additional usage, for example, there’s a fixed per credit rate or maybe a tiered rate, uh, beyond that commitment p– uh, level, then overage is often evaluated as variable consideration. The amount is uncertain at the outset because it depends

on future usage, but the pricing mechanism is established in the contract. If on the other hand, the customer must separately approve or purchase additional capacity once the original entitlement is exhausted, then the arrangement may instead need to be evaluated as a contract modification. Uh, now I’m not gonna focus too much on the contract modification scenario, although contract modification can obviously be a complex area, but the particularly common and difficult scenario arises when the commitment is determined to be a stand-ready obligation that is recognized ratably. So if a customer is trending toward a material overage, the company may need to evaluate

whether that e-estimated additional consideration should be recognized, uh, along with the committed price. So let’s make this concrete with a couple of short examples, and then we’ll, uh, show, show these in the system. So example, uh, one includes, uh, a few different items. There’s a platform subscription that’s recognized ratably. There’s prepaid credits with potential overages. Uh, there’s another pay-as-you-go product included here and an implementation fee, which is recognized on a percentage of completion basis. Although we’re focusing here today on usage charges, it’s worth acknowledging that real-world scenarios often layer in

multiple products, which can introduce additional layers of complexity. Just to further set up scenario one, we’ll focus on just the two– first two validity periods, those first two quarters. Uh, the commitment for each is fifty thousand credits, and during the first three months of this arrangement, the actual customer usage was in an overage state by about sixty-five hundred credits. And then in the next quarter, the customer was about eleven thousand credits below the commitment level, uh, which resulted in a breakage. So we’ll trace both outcomes through the billing and revenue recognition flow. So for our first example, we’ll start in Zuora Billing,

although similar data could be ingested from other upstream sources. Um, as noted in the previous slides, the arrangement, uh, includes a quarterly platform fee, uh, with an annual term. It includes prepaid credits, uh, which is paired with a placeholder overage charge, uh, just in case there are any. It’s got prepay– uh, a pay-as-you-go usage, and then it’s also got an implementation fee. Now, I’m gonna focus just on the first two validity periods for that commitment charge specifically. You can see for both periods, the remaining balance is zero, so all fifty thousand credits have been

used. In the first period, the drawdown usage fully consumes the committed credits. Um, and you’ll see that’s all the way through, uh, January twenty-ninth. So at January twenty-ninth, we capped out. Um, now this section only shows the impact of the prepaid credits. There were additional usage, uh, that resulted in overage. If we look at the second validity period, however, the, uh, actual usage doesn’t fully consume the balance, so you’ve got a deplete funds action for the eleven thousand, uh, unused credits, uh, that are going to be released as breakage.

Okay. So moving into the revenue side, those billing charges become distinct contract lines, and each line was assigned to the appropriate performance obligation template. So this first one, uh, was recognized ratably. Uh, the implementation fee is gonna be recognized on a percentage of completion rule. Then the prepaid, um, portion here is recognized as the consumption occurs, which in this case could include actual usage or expiration activity in the way that our upstream was calculating it. And then for the other two here- Uh, these are initially booked at $0 because there’s no committed value, but each will be bumped up if any

usage actually occurs. And in this example, we have some, uh, loaded. So one thing I’ll say, in the interest of time, I’m not gonna dwell on the allocation impact, but you can see what we’ve included. The, the fixed components of this contract, we’ve included the allocation, whereas the variable components have been excluded. Uh, the revenue pattern for the fixed lines will therefore reflect that allocated value. So in the case of the prepaid, uh, credits, for example, uh, we’ll ultimately recognize that eighty-three thousand five hundred, uh, rather than the invoice value of eighty thousand.

So, uh, just wanna take a moment to look at that usage activity. We saw it coming from the upstream side. So on this prepaid line, you can see activity rolling in over the course of the various periods. If we look at January, um, you’ll see that that activity ends on January twenty-ninth. So if you remember, that’s when we capped out, um, for the, for the usage for the drawdown. And in the April period, uh, for the same line, you can see that as of the thirtieth, you can see that eleven thousand, uh, credit, uh, breakage. So that comes through, and we can release the revenue for that as well. If

we switch over to the overage line and look at the activity there really quick, you’ll see there’s only activity for one period because we only had ac– uh, we only flipped into the overage, uh, scenario in one month. And you can see it starts on the twenty-ninth, so the remainder of that day’s activity kicked us over into overage, and then we had a little bit more activity on the thirty-first. All right, so finally, let’s take a quick peek at the waterfall here and notice the various revenue patterns going on, including the pre-scheduled ratable line, which goes, you know, over the full term. Uh,

and then we’ve got the various usage patterns, uh, including the overage, which was only in a single month there. Yeah, you can see on the implementation fee, that was– took about four months to complete that one. And then we’ve run this scenario out, you know, not the full year, um, but you can see where the overage hit in January. And then the pay-as-you-go had some just various activity over the course of the various months. Now, um, if I pay attention to this first validity period, even though we might have anticipated overage charges were likely since we were recognizing the prepaid balance as consumed here, um, we typically don’t need to estimate

that overage because once the prepayment is consumed, we simply kind of replace that line of activity with the overage activity. Um, I point that out only because in our next use case, this won’t hold true because we’re not gonna be recognizing the revenue, uh, as used. Um, just one other thing I, I’ll call out here is that the billing patterns don’t impact the revenue timing here. So this usage, the, the annual platform fee was billed, um, upfront quarterly, so was the prepayment. The implementation fee was, uh, hundred percent upfront, and then the overage and, uh, pay-as-you-go were actually billed, um, a month in arrears.

Um, so we’ve got all kinds of billing patterns, but the revenue always can follow the earned period. Okay? So briefly returning to the slides, we have one more example. Um, in this example, uh, we have a hundred thousand dollar annual stand-ready arrangement that’s going to be recognized ratably. Um, one month into the term, the company estimates, uh, ten thousand, uh, dollars of overage. This overage won’t be incurred until later in the term, but must be included in the transaction price. So for the second example, we’ll just go directly to Zuora Revenue. Uh, similar to the last example,

this arrangement has a prepaid commitment line and a separate placeholder line for any potential overage charges. However, unlike in the previous example, the prepaid commitment is treated as a stand-ready obligation, and so the base amount is recognized ratably rather than following the consumption patterns. And under these circumstances, we need to be on the lookout for estimated overages. And in fact, shortly after the contract begins, we obtain evidence that the overage is likely. And so in this example, we added… Yeah. So in this example, we added a, uh, ten thousand dollar variable consideration

adjustment, uh, which increased the transaction price from a hundred thousand to a hundred and ten thousand. Now, remember, this line is still a prepayment for a hundred and twenty thousand hours, so consumption activity is still tracked, um, but it doesn’t drive revenue for this stand-ready obligation. Once that pulls up here. Okay. So what you can see here is that… Just gonna move my Zoom out of the way. That we’ve got usage patterns, um, that are all over the place each month, but the revenue is, uh, b- Rateable because it’s stand ready, uh, and it ignores that pattern. In fact, we can kinda

see that the usage is running a bit ahead of the straight line pace, so it does in fact it o-uh, appear that overages are likely. And because the VC adjustment was made a month in arrears, you can see that I’ve got a catch-up in the month of December, and then rateable, uh, thereafter, uh, in, in line with the rest of the committed portion. All right. So for this example, rather than show the waterfall, I’m gonna show you the revenue summary, which is similar to the waterfall, but it includes a couple of other things. It includes the billing pattern,

uh, as well as the balance sheet impact. And you can see here that the hundred thousand was billed up front, and then the, uh, with the rateable revenue pattern, and even see that ten thousand dollar catch-up, uh, due to the estimate. Um, this ex-this example is still in flight, but just to describe how it’ll play out, essentially the VC adjustment here can be recalculated as needed over the course of the contract. Um, once the… once we start the billing actual overages, probably later in the, in the year, um, those will be deferred, and then once the contract term expires, the VC estimate will be removed and the actual overages will be released. So the final revenue will match

the total amount actually charged to the customer. All right. So just to recap, and I’m not gonna go through all of these points in detail. There’s just a couple of points I do want to call out. Um, first, there is a diversity in the way contracts with usage components are constructed, uh, including expected billing and revenue patterns. So breakage and overage are fundamentally impacted by the intersection of commercial terms and customer behavior. For breakage, you’re typically paying attention to whether you have reliable estimates and if it’s practical to incorporate those estimates and actual business practices around expirations.

Uh, for overage, the biggest challenges often come when the contract establishes a pricing mechanism, uh, so it’s treated as variable consideration, and especially when that obligation is determined to be stand-ready or if overage rates are significantly different from prepaid rates. All right, so we’ll just, uh, kick, kick off one last poll before the Q&A, uh, and we’ll answer a couple of those questions before we close out. So we’re asking how you estimate usage. Um, if you’re still exploring, um, but have a pretty good idea what you plan to do, please indicate that. We’re not sure yet is a pretty popular answer, and I’m not at all surprised.

Okay. All right. Uh, Leila, any questions we wanna specifically address? I’ve– I’m sure we’ve been monitoring and answering some things in the chat, but… Um, none so far. So if you have any questions, you’ve so far seen a couple demos. Aaron’s walked through overage and breakage. If you have any questions, please go ahead and drop them in our Q&A section.