Informatica Inc. (INFA) Earnings Call Transcript & Summary

Okay. Thank you, everybody, for attending. We're honored to have Informatica here. Eric Brown, Chief Financial Officer; and Vic from IR. Thank you both for attending our event this week.

Yes. Thanks for hosting us.

Good. Well, Eric, it wasn't that long ago that you were out with your 3Q '22 results. But just maybe to refresh those that maybe are less familiar with the story. Do you mind just taking 2 minutes highlighting what you think were the 1 or 2 key takes that you'd like everybody to stay with. And to the extent that you and Vic have been interacting with investors and analysts since, perhaps if there have been 1 or 2 common questions you're taking, I'll probably end up hitting on them with my question list, but maybe you could address them.

Sure. So yes, in summary, we had a really good Q3. In terms of our key metrics, we had cloud ARR growth of 39% year-over-year. And so it's not adjusted for FX. But 39%, which is basically within our range of growth, total sub ARR grew at about 27% and bottom-line non-GAAP operating income, we delivered well above the kind of consensus expectations on the strength of good overall margins, good mix, good cost controls as well. In terms of the follow-up questions, most particularly asked question, of course, is what are your thoughts on 2023, right? And obviously, we haven't given any formal 2023 guidance. But one of the things we did talk about is more kind of conceptual. We talked about the notion of having cloud-only approach next year. Now cloud only doesn't mean that we'll only sell cloud products, and we won't take any additional orders for self-managed products or perpetual license products. What it means is that as we sit here today, we have now fully completed our multi-tenant cloud platform roadmap for IDMC. Last product that we converted from single-tenant cloud to multi-tenant cloud with MDM. And so what this does is it sets us up for a real concentration in regards to all aspects of how we operate the company, to sell, support and lead with cloud. And so 2023 is like the first full 12 months that we have with that positioning. So we spent a little bit of time talking about that, what it meant and why now as it relates to roadmap completion.

Got it. I do have a couple of questions on that. So we'll defer that a little bit. I suspect the other common question you're getting, Eric, might be a little bit less Informatica-centric, and that is it's rough out there? And how is that environment trickling into Informatica's results? So my observation is that, I'd say, the headwinds are a little bit less than what some of the other software companies are talking about, but you're feeling it. Could you elaborate on the extent to which you are in terms of sales cycle, deal scrutiny, and things like that?

Yes. It's fair to say that not unlike other companies in this space, we have experienced, we specifically cited elongated deal cycle, additional approval levels, in some cases coming into the mix last minute where your internal buying sponsor was unaware until the last bit that there's a higher level of financial review, for example, the transactions normally within their delegated authority. So we're seeing that. But notwithstanding that, I think we've proven that we have a resilient business model. And we didn't get there by accident. We have painstakingly converted the entire business to a ratable model. And so if you look at our numbers, basically, our total ARR is approximately equal to our total GAAP revenue. And the things that really drive cash flow and margins is installed base, renewal rates, upsell rates and unit economics where are we in the gross profit margin line. All these things are proving out. As we sit here 3 quarters through the year, we, for example, seen slightly better than what we initially assumed renewal rates. And that's kind of counterintuitive to what you'd expect in the current macro environment. But what it does is it points back to the fact that our products are software cloud offerings are used by our customers to run mission-critical workloads. These aren't nice to have ad hoc analytic workloads, but companies rely upon our technology to move and transform data every day, every hour, every minute, so they can manage their supply chain, manage their customer reactions, close their books, those types of things. So we kind of -- we've been through these types of cycles before. And our prior incarnation as a public company, circa '08, '09, we went through a similar kind of tough macro environment.

I covered you back then.

And when we see our maintenance renewal rates back then held steady as a rock mid-90s. Again, just underpinning the fact that we're infrastructure software, not optional application software, and we drive mission-critical workloads. And so I think that, that all lends to a more resilient than average business model.

Okay. And then when -- Eric, when you set your financial guidance for the fourth quarter, what was the underlying assumption that you embedded in that guidance? Did you assume that the environment that you were seeing in September, October, basically flatlined through the end of the year? Was that the base assumption? Did you assume that things would get a little bit worse?

We assume that the worsening conditions that we saw in September, 8 month 3 of the third quarter we assumed 3 months of that in Q4.

Okay. Got it.

And so with that, elongated sales cycles, larger deals taking a bit more time to get across the finish line. Our assumptions on pipeline conversion rates on sort of types of transactions, we adjusted those accordingly. And so that was all kind of factored in. Of course, we updated for FX to make sure we're dialed into current rates because that's obviously been a factor throughout this year as well. So yes, we updated it for all the observed month 3 macro conditions from September.

Good. Okay. That's encouraging. Let's take a little bit on this resiliency that you've been talking about, Eric. So I'd love to understand a little bit more the resiliency of your data management use cases relative to, let's say, Snowflake's analytics use cases or how Tableau might be used or Mongo and Oracle databases. What is it about data integration, MDM, data migration that might make that less discretionary than some of these other data software companies?

Yes. Data integration at its core, it's about ongoing management transformation of data produced in one or more source systems moving through or into a persistence layer used for analytics to drive not necessarily just analytics and charts and things like that, which are visually interesting, but to drive operating business decisions like supply chain replenishment cycles or they lead to, for example, the closing of one's books or the balancing of a risk position at the end of the period for a financial institution. So all those things I mentioned are necessary and must be done every period. The data has to continue to flow, and in order for it to flow you have to have kind of a continuous subscription to the underlying data management, in this case, DI technology specifically because DI is once you map it out, it stays in place definitely. And when I talk about resiliency, it's a synonym for utility, right, of the application that we're powering. And so the concept of what is and what happens there in infrastructure where versus an optional analytic app layer, the 2 very, very different things. In terms of what's absolutely mandatory data around the business versus what may be somewhat optional to have, right? Because if I'm doing ad hoc analytics, I've moved a lot of data into the cloud data lake, I could have 10, 20, 50, 100 analysts working way looking for some interesting patterns. I can dial that up and without actually operating my existing work processes. Now if I dial it down, I may have less interesting discoveries over time. That's a fair point, but that's not quite the same thing as not being able to fully kind of service my supply chain.

And the DI subscription business is not incredibly data volume sensitive so that if -- am I correct that if you decide that in a recessionary environment next year that 10% to 20%, let's say, of those data analytics jobs are somewhat discretionary and they need to be cut. That doesn't translate to a 10% to 20% reduction in our data integration needs, does it?

No, it doesn't. It's been a topic of note lately this idea of optimizing cloud consumption. And I'll compare and contrast kind of the DI, cloud consumption use case versus a more traditional cloud use case. In DI, once you create these mappings and again, you're going to have an in-production applications let's just stick with supply chain management. It's not as though you could cut back by 50% the data sources that you're using to manage your supply chain, you need to have the totality of information, right? So you can't go minus 25% or minus 50%, expect to get pretty much the identical outcome. The app breaks down, our supply chain breaks down. So in contrast, and I can give you a real live use case on cloud consulting optimization. So we dramatically built our platform on top of AWS, GCPs or -- and coming next year, OCI. But about 3.5 years ago, we actually spun up an internal program to optimize our own cloud, i.e., AWS utilization.

You could monetize that tool right now, Eric.

Really. Yes. Well, there's a lot of good tools out there. There are already vendors like we use cloud ability and other things at the time. But what we found was kind of fascinating, a lot of it is kind of behavior driven. So we use AWS, not just for production cloud, but we would have our sales team that would stand up POCs, proof of concepts and demos in AWS. And what they would do, they select like the highest performance, best database version of AWS and build a POC. We present it once or twice in the sales cycle, but then they would leave it live and running. And so it's kind of running continuously. It's taking the compute, thinking of storage. It's like the meter is running even though you've had just a couple of discrete needs for that particular demo application. And so the analogy I would use is like walking out of your house leaving, not just all the lights on, but all the water running. It's just pure waste. It's like, okay, if you're done with that instance from the POC in this case, just turn the whole thing off. And so we had actually developed that awareness and then some tools and some tech and some alerts and things like that to say, "Hey, this app has been running unattended for like 30 days. You should probably shut it down or take a look." And so we went through cloud optimization for what I just described for ourselves. And it takes some time. First, you have to have awareness. And I think what's happening more recently, 12 to 18 months ago, I think people were concerned about their OpEx and cost optimization. We're in a very different macro environment. Fast forward today, everything is under more scrutiny and some companies, I think, for the first time are actually looking into these issues. And there's other ways to optimize cloud as well. I mean if you're a cloud operator, I'll speak to now COGS optimization. And so many people may be familiar with -- I'll stick with the AWS enterprise license agreement. They call it EDP, their acronym, but it's basically an enterprise agreement. It's typically a 3-year commit, and they say the customer, in this case, would be us, we will pay X dollars per year times 3 years. TCV commit. Obviously, it's around over $60 million, so $20 million a year. And if you don't spend a full $60 million at the end of 3 years, you pay them the difference. It's take or pay. Now when you negotiate these agreements, there's a couple of ways to optimize. You can make a greater TCV commit, take it from $60 million to, say, $120 million, double it in this example. And your actual rate, discount from rate built into ELA may go from say minus 15% to minus 16% discount to minus, let's say, minus 17%, minus 18%, pickup 2 to 3 points. Don't we commit 2 extra points of discount. It sounds like an okay trade. Now the reality is, for optimization of COGS if you can architect your application, in our case, like IDMC, in such a way that you make maximum use of reserved instances and savings plans. These are basically dedicated clusters of servers and in some cases, in a specific physical data center. So like you buy those servers and everything for like 3 years, 1 year, 2 years or 3 years. If you look at a discount, not of 15% to 17% off a list when you do that, you'll get a 50% or more discount. So if you think about optimizing cloud COGS, here's like another twist on what one can do. In this case, it's us, the vendor who's using AWS as a platform, and we're optimizing the way that we spend AWS resources to serve up IDMC to our end users. So there's a lot of different takes on optimization. We have a long-winded answer, but I thought it was worth commenting on how there's like cloud optimization just for pure wasted use. There's also cloud optimization for analytic apps, which are optional and cloud optimization for embedded run time kind of COGS scenarios like the ones that I described.

So when -- and we hear this quite a bit from Microsoft and Amazon, they're talking about a significant uptick in the pace at which their customers are optimizing their infrastructure spend. So for the portion of Informatica that runs in the cloud, when large AWS Microsoft customers optimize, how exactly does that trickle down to maybe being a pressure point or not for Informatica? In other words, what is the sensitivity of their Informatica spend to their own optimization efforts. Is that a risk that we should be monitoring, Eric?

I think they are two independent things because their consumption of IDMC platform is a function of their data movement and their workloads. And like I said, once you set up these processes, you need to run them continuously with all the data elements. And so there's maybe some on the margin tweaking that you could do, maybe you changed the frequency a little bit but still the whole process has to run end-to-end. Whereas with just generic AWS or optimization, you're into some of the other scenarios that I described previously, like, well, how are you using it for, your internal use ad hoc applications, et cetera. And there are -- like we found ourselves, there's definitely an opportunity 3.5 years ago for our own cloud optimization. But we're doing forgetful, wasteful things.

Yes. Got it. Let's go back to the point you made earlier, Eric, about next year, although you haven't provided any formal guidance, having this cloud-first or cloud-only approach. So right now, Informatica can be deployed in a self-managed fashion on-premise or in the cloud. Yet only, I think, 2 or 3 percentage of that maintenance installed base have actually converted to the cloud. So I'd love to understand a little bit what the levers are to accelerate that, assuming that's a goal of yours over time?

Sure. So first, at a high level, we rearchitected product, IDMC. We market in 2 forms: pure SaaS, where we're doing all the hosting on top of AWS or GCP. And then we also sell that same software to an end user for self-managed. And oftentimes, what they'll do is, they'll take the verse of the software, and they install it themselves and run it themselves on their own AWS systems. And they'll say, "Hey, we don't want the SaaS piece, we'll conduct that maintenance and oversight." Very different revenue models for us, one versus the other. The first is pure ratable cloud, the second self-managed leads to upfront 606 rev rec and acceleration. ARR is identical in either scenario. And then we have the installed base, which is the maintenance, north of -- a little north of $500 million for us, which is the outcome of 15-plus years of cumulative perpetual license sales. Now we are not forcing migrations. Again...

Like they used a hard stick.

No, because back to my earlier comment about DI being a mission-critical workload. It's like -- it's an analogy. It's like telling someone that, look, you're running our ERP system, we got like force you to migrate next quarter. It's like you just physically can't do it. And if I attempted to do it in an unorganized fashion, I blow up my business. Back to the comment of we power mission critical workloads, right? That's the recurring theme here. So in many cases, these power center applications have been running fault free for a decade that provide value every day, every hour, every minute, and people are really happy to keep paying the maintenance contracts. So we're not end of supporting these contracts. We're not taking a stick approach. It's more of a carrot. The carrot is as follows. One, the preponderance of our R&D spend, our innovation is on the new cloud DI says of.

add more features to the cloud version.

Correct. We're past future parity and now it's moving well ahead of old on-prem power center. So that's kind of incentive number one. Incentive number two is when we sit down, customers are again looking to optimize their spend. They're realizing that may not be cost effective to continue to run and maintain on-premise applications. And so the standard business case that we would take with the customers think about migration is a TCO, total cost of ownership based argument. So we'd work with them to say, today, you run power center, you have to maintain all this infrastructure, servers, databases, administrators, power utilities, everything. And we'd say, this is where you are today. Once you fully migrate to our cloud, pure SaaS, these are the costs that you jettison in between. And what we found on the migration deals that we've signed to date, we've achieved a 1 to 2 ratio, meaning that the TCO is sufficiently compelling that a customer that would pay, for example, $100,000 maintenance today is going to pay us $200,000 of cloud DI subscription revenue post migration. Now in between, there's a fair amount of work, right? Because this is, again, mission-critical production app, you don't just flip the switch. It's not a self-contained app. It's much more readily migrated. This is infrastructure, and it's tied into dozens of different data sources, systems inside the company. So it's a bit of an undertaking. It again underscores how resilient and mission-critical and sticky our installed base is. But at the same time, there's this compelling ROI opportunity. We think of over time, more and more customers will say, now is the time to -- we have a lot more high-level interest in migration than our actual stat life to date on the migration. So it sits a little under 3% of the installed maintenance space. And so we'd like to continue to press on the pace of migration. We have kind of a 3-part approach to it. Part one, we're basically through the first phase is using our own Informatica consulting services personnel. So these are badged employees. We have stood up and developed a series of translation tools, utilities and script. And up until now, all the migration work has been done by our professional services, Informatica badged employees using these tools and scripts. We are now starting to make those tools and translators available to third-party SIs. Again, we talk about our key ecosystem partners. One very important layer of that is global SIs. And as it turns out, they're very interested in pitching the big picture of digital transformation. So put your sells in the shoes, a senior Accenture partner got 1 or 2 Fortune 500 accounts, you need to pitch the next big idea for value creation for your client. Digital transformation is like one of those big topics today because, like all businesses, irrespective of the industry are thinking about digitally enabling their business. And it's not just I'm going to move certain things to the cloud or just stand up new in the cloud, is kind of an end-to-end transformation. It's like a perfect global SI. Inside that is a component which is your data management. If you're going to be thinking about cloud modernization, your whole data management strategy, has to support that, and that's where we come in. So having them trained and enabled now that we've proven out the program over the last 1.5 years or so, we're now kind of in this second phase of migrations. And we're looking to get over time more lift and more capacity, frankly, out of using third-party pro services people. And the third part of this is really ongoing, which is just the continuous reinvestment against tools, translator, script and technology and so that we can decrease the time that it takes to migrate individual assets or transforms, which are part of the installed power center base because, again, typically, as customers 8, 9, 10 years old, we've built up thousands of data transformations over time. We're all coded into power center, and we need to like more effectively and automatically unpack those, so it just ports over to cloud DI. So ongoing investment in tools, tech, translation is kind of the third part of the migration overall strategy.

Eric, is there any thought or effort underway to play the price card here in order to shift the TCO and economics in favor a bit IDMC cloud product? Is that under consideration or being used already?

It's under consideration. There's two ways to approach it. You can have outright maintenance price increases.

That's what it feels like Oracle is doing.

Or you can start to offer value-add extended support, which is just another flavor of price increase, but it's slightly -- because you're giving something additional as opposed to, I'm going to give you the same thing and just charge you more. So we're doing the latter, not the former at this point. We clearly reserve the right to do outright price increases over time. We haven't taken that approach yet. We haven't seen it as necessary. We'd like to kind of lever the SI channel, continue to work on the tools and tech and let the road map investment in cloud DI really be kind of a pull.

Makes sense. Eric, let's talk a little bit about a different subject, and that's what's inside that IDMC product suite because back in the old days when I used to cover Informatica in its first incarnation, it was really a power center play, mostly on-prem, actually. But one of the amazing things is, as you went to private, you undertook this broad product diversification. So now there's a lot more to the portfolio. Now you don't -- the company doesn't regularly break out that mix. But perhaps you could give us a refresher on what parts of the portfolio are growing in mix where maybe because of this environment, adoption is actually picking up. And is there any part of the portfolio that might conversely be decelerating a little bit. So some sense of the -- how the mix is shifting these days.

Sure. Yes. Back at the time of privatization circa 2015, the vast majority of the revenue and installed base maintenance ARR was power center related or denominated. We had kind of an early-gen MDM business there, completely different architecture, but very early stage, and that was basically the company. So it's really over-indexed to power center. Since then, we've built out not just MDM SaaS, we built out C360, Supplier 360, Product 360. So specific versions of MDM. We've built out data catalog, data governance, data privacy, app integration. And if we look at -- to your question about like where is there interesting high growth rates in the areas of data catalog and data quality? And so this is like typically what happens. You'll have perhaps a data integration or an MDM-centric initial sale that will be put into operation, data democratization, which is kind of an objective of these digital transformation -- digital demonetization, it's just making more curated data available to more people. So the total order gets higher value out of the data that you're putting there in a data lake. So once that starts to occur, then the concept of data quality is important. And so we see the highest year-over-year growth rates if we look at subscription ARR by general product family in data quality and data catalog. Data catalog is interesting because, again, when you undertake the digital transformation, question one is, okay, what are my existing data source is? We want to convert those sources to assets. And so you understand what you have, where it is and what's in each database. And so an enterprise data catalog basically does just that. It can actually not just identify and count and say, okay, there's 10 Oracle databases out there, it can look into the Oracle database, look at the table structures and directly extract information about row and column headers and tell you what kind of data it is. It's like, okay, this is a database that had customer data. This is a database that had customer ticket information. This is a database that had marketing outreach on customer prospects, which ultimately converted to customers over here. And so a data catalog is really important. So it's that and data quality where we've seen -- again, they're smaller books of business relative to DI and MDM. But that's where we're seeing the highest year-over-year growth rates.

Got it. And Eric, the competitive environment is different in those new areas. So back when I was covering you when you were mostly a DI, data migration vendor, it was IBM DataStage. It was Ab Initio. So maybe a two-parter for you on the competitive front. Within the DI book of business, which you just said is still pretty large. Are there share gain opportunities still sitting out there against IBM and Ab Initio that might fuel DI growth? And then part two is on these new categories like data quality, data governance, et cetera, there's a cohort of new younger vendors kind of like Calibras, and how does Informatica differentiate against that those upstarts?

Yes. I feel good about our competitive environment. So with regards to competing against, I'll call them, I want to say late. I'll say incumbent vendors, kind of the IBM DataStage. Clearly, we've massively invested in a pure multi-tenant cloud architecture and platform. And it's not just platform on a PowerPoint slide, it's a platform because at the very beginning, we built a shared micro services coming at a data management and AI throughout the entire platform. It's not just a bolt-on module. There's AI throughout IDMC. And it's AI that allows our platform for the next incremental customer to be that much more accurate in the sense that it recognizes anonymously, of course, because we're running all this through a centralized cloud system. When we look at data mappings, we discover like data mapping, we didn't quite see before. We find out what they are. There's a bit of a learning algorithm that occurs. And so that incremental customer -- end customer just the benefit of those that came before it. And so AI is built into the platform. And we just don't have that at the incumbent competitors that you mentioned. The other thing too is just one of simple focus because we've done one thing for 25-plus years, just data management. And we've consistently worked harder than anyone to have the greatest library of data connectors. In other words, we've built up 50,000 meta data where data connector salads are. There's no permutation of data sources to data target transforms that we haven't seen and built a translator for. Let's talk about kind of the newer vendors like those that are more app-specific data governance to stick with Calibra as an example. So the good news is that really wasn't the concept of data governance in prior years. It's become an issue and that it's created an opportunity to expand the TAM. So different competitors come in like look to have a data governance product because it's now on a checklist for enterprise risk management or CDO's best practices list, et cetera. We obviously believe in data governance as well, and we've built a product into our platform and to support it. But Calibra is typical of the condition that we have. And I see one product family out of 7 that we offer. By definition, it's not a platform. It's a product. It's really an app. There is no upsell, cross-sell afforded to them. As a private company, I would -- it's fair to say that 12 -- up until, say, 12 months ago, they were in -- growth in any cost mode. These private companies raised money at very high revenue multiples, 20, 30, 40x in some cases. Obviously, that environment has changed. And they never operated profitable. So they really fully proven out unit economics and sustainability. And we like to compete against them for the simple reason that I can display the vendor stability card. So the sales teams like these guys are burning money, capital is tight, are they going to be there 2 to 3 years from now to support you. And by the way, if they are, you convince yourself of that, it's single-purpose tech. You don't have a platform. It's going to cost you more to own them plus 4 or 5 other vendors, data catalog violation, for example. DI, we could -- MDM, we could rattle off a couple of competitors like Reltio. And like stitch all that together versus having best-of-breed products on a true unified platform. And so ironically, I think as things kind of tighten up here and people look to get better leverage out of their IT investment dollars, placed your strength a platform, right, with best products has slightly more compelling total cost of ownership argument working for it.

Makes sense. Interesting. I've got a couple more for Eric in our last minutes. But for those of you in the audience that want to ask a question, you've got a QR code in front of you, you can scan it. Shoot me a question and it will pop up on this tablet for me. But while we maybe gather a couple, Eric, you can't talk to a CFO without giving you a chance to brag about your margin, cash flow story. So you just contrasted your story with that of some of the pre-IPOs that might not be in that same margin position. But when we all look to your adjusted EBITDA margins next year, you're not going to give guidance, but can you lay out what are a couple of the things that we should be keeping in mind as we think about margins next year? What are the 2 or 3 of the variables that you can leave it up to us to measure precisely, but maybe you could list what a couple of those things are?

So first and foremost, we talk about next year really being a year for cloud only. And again, I touched upon that earlier. I mean that we've completed the first iteration of heavy lift on platform reengineering. We're going to spend much more time marketing and supporting Depot sales enablement of cloud in favor of -- well, not just perpetual license but also self-managed. The second thing is that there's been a notable change in mix in our overall pipeline. If we went back 12 to 18 months, maybe we had a roughly equal mix of cloud versus self-managed in our pipeline. So this is growth pipeline at all stages, all close periods. Fast forward today, roughly 3/4 of our total pipeline is now denominated in cloud. And we've seen very specific examples like in Finerv, for example, like 18 months ago, it's fair to say Finserv in general was a laggard in terms of cloud adoption. I've seen major banks calls. We only want self-managed on-premise. Don't even talk to us about cloud go from that position to we only want your SaaS solution in multi-tenant cloud.

And Eric, how does that mix shift that's building up now almost 3/4 cloud, how does that translate into a gross margin difference?

Right. So it plays out as follows. To the extent that our incremental subscription bookings tend to be more denominated in cloud versus self-managed, that will lead obviously more cloud ARR, which is what we want in the long term. I'll come back as to why in a moment. It will lead to less GAAP revenue recognizer front and less short-term non-GAAP op income because -- we have a couple of slides in our investor deck, Slide 34, 35, I think. We actually compare side-by-side $1 million a year, self-managed deal an extra $1 million a year cloud. One case, with the self-managed deal, you'll recognize nearly $1.5 million of GAAP revenue upfront, upon bookings, it's a whole 606 effect versus the cloud product, you only get 1 month out of 12 worth of revenue. The ARR addition is identical, one is obviously in the cloud ledger versus self-managed. And the cash flow is identical because we -- both contracts, we invoice them 1 year of ACV build up front. So what this means in the short term, you might see as you do that mix shift, away from self-managed to cloud for net new subscription bookings, slower -- lower GAAP revenue, lower non-GAAP op margin, but comparable unlevered free cash flow flow-through. Over the mid- to long term, the cloud has advantages. So notwithstanding that deferred revenue, cloud renews at a higher rate. It has a better expansion characteristic relative to self-managed. And for those reasons, it compounds at a much better rate than self-managed. And so yes, there may be kind of a bit of a short-term trade-off on GAAP revenue, non-GAAP op income, but I think more value is created when you sign more cloud versus self-managed. So that is our stated intent. And again, that's not to suggest that all our bookings next year will be cloud and that's self-managed. But we expect it will be a higher mix of cloud relative to self-managed versus what we're expecting for full year 2021 because we see the demand expression change in the form of the pipeline rebalancing majority towards cloud.

And Eric, you've got levers to try to offset that very near-term margin pressure. You obviously can control your personnel costs and non-personnel costs. Are there 1 or 2 levers that you could lean into to sort of buffer that mix shift you just talked about?

Yes. We've started doing that and talking about that a little bit. So last quarter, we gave a specific kind of stat. We said in Q3 2022 compared to Q3 2021, we hired about $250 million incremental people worldwide, and over 95% of which were hired in low-cost locations. So that's like outside of North America, Silicon Valley, right? And so one of the other advantages we have because we've had captive operations in India for primary R&D, customer care, virtually every single back office mentioned, G&A, finance, tax, AR, HR, everything. We can -- holding our hiring rate constant by location shifting it produces significant safety. So that's one of the things that we've already started to do. And in the spirit of optimization, we'll look to continue to do more of that going forward.

Makes sense. Okay. Good. I see we're less than a minute out of time. We went for a full 40 minutes. Why don't we end it there, Eric? It sounds like you've got an exciting couple of year journey in front of you as this cloud mix increases. So look forward to watching that play out. Thanks again to you and Vic for attending our event.

Thank you very much.

Okay. Our pleasure.