All right. I think we can go ahead and get started. If I could ask someone to help with closing. All right. Good morning. Thank you, everyone, for joining us at the Goldman Sachs Communacopia + Technology Conference. I'm Gabriela Borges. I head the emerging software vertical here at GS. I'm delighted to have on stage with me, TJ, Co-founder and CEO of AvePoint, and Jim Caci, CFO. Thank you so much for taking the time. So, TJ, I'd love to start with a little bit of the history of the company, because 2001 had deep expertise in building this company over that time.
Yep.
What has stood out to you about the way the industry has evolved since 2001? Help level set us a little bit, where does AvePoint sit in the enterprise technology stack relative to the ecosystem of Microsoft and some of the other companies that folks are familiar with?
Well, great, great question. So the industry goes through this cycles of centralization and decentralization. So since 2001, we actually now have gone from the distributor model to now cloud, which is centralized model, and more importantly, recently, obviously with aggregation of data and the massive increase in compute and cheap cost of storage, we now have generative AI, large language models that's driving further growth. So we have seen tremendous change in the dynamics of enterprise environment. We have been helping our enterprise customers managing their information for the last 20+ years, so we get to know all the legacy data stores, all the different data silos, hybrid deployments, multi-cloud deployment scenarios across enterprise.
So we were able to invest early, that's kind of the luck part of it, into Microsoft Cloud back in 2010 days, and build that sophistication and maturity globally, for the audience. And also because we started in the enterprise content management space, which historically has been very large enterprise and focused on highly regulated industry and public sector governments, that was, by nature, a very high touch experience, so we had to expand to 18 countries and serve those customers in person. Fast-forward to now, we have a global platform. So we're in Asia, we're in Western Europe, we're in U.S. and North, Canada, expanding to LATAM and Middle East and Africa through our Channel Program.
So we're able to basically cover the global market, but also very importantly, extend the segmentation coverage from just large enterprise, regulated industry to now medium to small businesses because of SaaS nature. So more accessible software deployments to the smaller customers because they don't have to install, maintain, worry about security. And very importantly, out of the SMB segment, there's also managed service providers, which is a new trend as well. In the last few years, that's the fastest growing segment for us. What that is, is essentially small to medium businesses don't have IT. They're now aggregating IT expertise by these consulting companies, and they are then providing the management capabilities for businesses.
Of course, the MSPs industry vertical is also consolidating, so we also see very large MSPs, global MSPs, and we actually also recently signed up a few of them as well. Crayon's one such example, SoftwareOne's another. They have global MSP practice. So that, that's the new development. Of course, the latest evolution is AI. My background is in machine learning and AI, so it's near and dear to my heart. And machine learning is no stranger to AvePoint. We've been actually a long-time consumer of Cognitive Services, which is in Azure, which is basically the front-end API to OpenAI. And we have done great projects, especially in Singapore, around EdTech, around GovTech, around leveraging machine learning and AI. But with any AI initiatives, 80% of the workload is information management.
It's data cleansing, data classification, data consolidation, aggregation, analytics. So when we talk about AI-ready information management practice, and that's where we now see massive demand as every enterprise is racing to build proprietary data models to, to better their business on top of the large language models. So those are some of the evolution key points in the last 20 years.
So let's stay on that last point. AvePoint as an AI-ready information management practice.
That's right.
Tell us a little bit about what that means in practice, and what are some of the conversations you're having with customers that will help them be more AI-ready?
That's right. So if you take OpenAI and even Google Bard, right? When you actually interact with it using parameters, it behaves very much like a very good summer intern. And the issue with summer intern is that, you know, it doesn't have domain-specific knowledge nor knowledge about your enterprise. What makes it different for actionable AI for companies is to be able to leverage the decades worth of domain data that your company have accumulated and build another layer on top, right, of the generative models to make it more actionable and purposeful for your business. Even something as simple as handling inbound customer requests, right? Customer support, case queries, Tier 1 , Tier 2 , and auto routing. But the foundation of AI is clean data in, actionable data out. Otherwise, you have trash in, trash out problem.
If you have out-of-date, trivial, redundant data, then you will have more cases of incorrect answers coming out, right? This, in our industry, called hallucinations, right? So that's something that you have to try to control. So what we do then is, because we have been in this business of information management, enterprise content management, start from there, we can understand data from all different, different data repositories, whether it's file shares, Documentum, email systems, OpenText, chats, right, Slack, Teams. We're able to help companies aggregate that data and be able to correctly classify, label, and then tie a data lifecycle and governance capability to it. And very importantly, also, access control. Who has access to this set of data internally and externally? And then make it actionable. So for example, we, we actually mentioned this at our latest earnings.
We're able to help Australian government agencies achieve 20x productivity improvement by properly labeling all these information records that they put in cloud, and then make available to the public. Then they, they're now in the process of building AI models. I was actually just in Singapore last week talking to GovTech, which is their government technology arm. They actually said to us, for all their government agencies, they don't wanna use cloud, right? There's this data segregation issue and data security issue, so they want to build their own LLM, large language models, on top of their data. And wouldn't you know it, vast majority of the unstructured data that the government of Singapore stores is in SharePoint, which is where we came from.
Of course, yeah.
So they say, "Hey, AvePoint, can you come in and help us build an LLM on top of all the SharePoint data stores across government?" We're like, "Yeah, that's a fantastic project." And the premise of that ask was they want to improve the search capabilities across enterprise unstructured data. So it's basically how to find things better and also leverage machine learning AI to actually, you know, draw more context of your domain-specific knowledge. So these kind of asks are coming in fast and furious, and this is where we talk about AI-ready information management. Otherwise, you're gonna have trash in, trash out problem across your proprietary data set.
It's really interesting context, and this is a question for Jim as well: How do you compare the potential deal size for an enterprise customer that is working through some of the quality data in, quality data out problem? How do you compare the deal size of one of those engagements today to an, to a typical engagement if we were just to take your ARR and divide it by number of customers? In other words, because customers are going through this life cycle, what are the implications for your normalized growth over the next three to five years?
You wanna take that, or do you want me to go first?
I think that question is for you. You can go first, and I'll.
I'll go first.
I'll add in.
I'll go first. You know, it's interesting. If we maybe to break that question down into a couple pieces. First, we have three segments, right? In terms of we have enterprise customers, we have mid-market, and we have SMB. And so if you just took our ARR and divided by the number of customers, that probably wouldn't give you the right picture in terms of an average, 'cause we really have those three segments. And when we look at those three segments, the average ARR per customer is different per segment, as you would expect. Our enterprise customers are generally larger, our mid-market is a little bit less, and then obviously, the SMB customer would be our smallest. But kind of what you're, so that, that's the first setup for us.
And then what we're seeing, as you talk about, you know, kind of the projects or what we're gonna be doing for some of these customers, that's really where we're first seeing it is in the enterprise or the government customers that TJ alluded to, and those projects generally are larger. And some of those projects not only include our basic software, but then are including services as well, 'cause they're really looking for something beyond even what the current software would allow for. And so we're seeing those projects being much larger in terms of dollars and scope. But they're really—it's interesting because those projects then lead to future development and almost our R&D projects, because it's helping fund then. That knowledge gained is embedded then into the next generation of the software that we're producing.
So it's a nice, y ou know, it's helping those enterprise customers kind of get to the next level, but then also creating software for the not only enterprise group, but the mid-market and SMB will benefit from that same knowledge that we gained, and we provided to that enterprise customer and now make it available to everyone.
Yeah. So my commentary to add is, we are working to increase the ARR per customer across the segments. The easiest to think about is our Confidence Platform, it's a SaaS. We are actually, today, Microsoft Cloud ecosystem's largest SaaS data management governance player. So our Confidence Platform is that layer. Now, we're layering on top of that additional vertical solution capabilities. So information, AI-ready information management is such vertical capability. Of course, we also have EduTech, that's training management, digital assessment, and learning management also on top of this data layer, Confidence Platform, that's fully integrated with Teams, with Office Cloud, Dynamics 365, with their CRM Cloud, as well as Azure, which is a compute cloud. Today, we're actually an eight-digit consumer of Azure.
Actually, we just met with the Microsoft senior leadership in Silicon Valley two weeks ago. We're top 24 of their global 240,000 product partners, and we're top three of their global 400+ education product partners. So in this multi-trillion dollar ecosystem, you know, we have the sophistication and maturity that we built up over the last 22 years. We know how to really navigate and provide additional strategic value within the Microsoft ecosystem to help customers. Ultimately, this year is the theme of optimization, ultimately to optimize and maximize their return on investment on top of Microsoft Cloud.
That leads to two interesting questions. I think the first is, you all do know Microsoft incredibly well, and so would love to get your thoughts being one of their close partners. How do you view your strategy as being complementary to Microsoft's strategy as they continue to invest in what is also arguably an AI-ready information management practice?
Yeah. So to play in a hyperscaler space, whether it's Microsoft or Google or AWS, the role of a strategic vendor partner is to add value to the customers, to enrich that ecosystem. We all say the hyperscalers, they will continue to improve the baseline infrastructure, right? So when they have the Syntex release, when they have these base level, you know, API augmentation improvement on backup, for example, we will leverage our services. It's akin to a utility company layering bigger pipes in the street. At the same time, as you improve your house, right, your digital transformation projects, you still need to go to a prime contractor like AvePoint, to then go help you improve that actually services and capabilities to actually land Microsoft Cloud into your organization and enterprise. And that's really the analogy here.
And Microsoft, when they, you know, their latest, Partner Conference, they talk about, again, this whole, you know, construction and interior design kind of, you know, analogy of every layer have different type of, y ou know, the C-suites and then the director level, and then across different departments, there's different capabilities that we can bring to bear on top of the Microsoft platform, and this is where the value add that we bring, to the table. Having said that, the world is multi-cloud. We have actually a big, conference, industry conference, called #shifth appens, in D.C., in October. If you say fast enough, you know, it also means something else.
Oh, we got it.
Yeah. But that's technology, right? So it's actually established as an industry conference where we have customers, industry expertise, so we will talk about that. We'll also showcase some our AI capabilities on Google platform. So we're actually going to showcase a lot of new products, the capabilities, leveraging AI, across a multi-cloud fashion, because our customers, by definition, are multi-cloud. There isn't, t here's very, very few customers that's only using one hyperscalers environments for the exclusive purpose of business continuity, and also data sovereignty. As the world also increasingly going toward anti-polari zation, more localized approach, multi-cloud, multivendor approach, and also data-sovereign hybrid deployment, that complexity favors vendors like us that's been there and seen it and done it.
So I think those are good macro trends, tailwinds for us.
So the second part of the question is your comment on helping customers optimize their usage with the Microsoft ecosystem. I think it's especially pertinent in this environment where we've heard so much about optimizing IT costs. So talk to us about how that's a part of your go-to-market. What are the types of savings, and what are the types of cost buckets where you're helping customers? What is the order of magnitude of savings that you're helping customers realize?
That's a great question. I think, there's multi-layer approach to cost savings, right? Even look at just within the Microsoft Cloud context itself, you have different licensing types on Office Cloud , right? There's F1, E1, E3, E5, and the magnitude of price differentials could be as big as 4x-5x, right? And then now you have Copilot, which is another, you know, a whack at it, $30 per user per month. So that will effectively double your spend if you're already at $30 per user per month spend level, which is about E3 levels.
So we actually help customers kind of coalesce on entitlement management, license management, also optimize feature set, so that not everyone in your organization have to go to the most Cadillac version of Microsoft offering, but still make sense of the data holistic management, right? So for example, you take Walmart. You know, vast majority of their workers, employees are not desktop-based person. They're remote, you know, device-operated workers, and there's many temporary seasonal workers. So Microsoft has a term for that called retail license, right? F1 license. So how do you then actually make sure that the data that's generated out of these population of employees and the data that's generated from your C-suite, which are on E5, are treated the same, right? With the same holistic, you know, lifecycle, access control, and management.
Of course, for a partner, especially MSP, who help small business manage different licenses, how do you actually have a more holistic approach to manage these licenses? So that value, one goal is, cost savings, right? We're talking about potentially 30% cost savings for the customer right there. And then, of course, you have other, cloud vendors, like take Box, for example, right? Box is $30-$40 per user per month. Effectively, we are now actually doing a campaign with Microsoft to say, "Hey, if you already paid, you know, $30-$40 on Office E5, what do you need another cloud storage vendor for?" When actually, in addition to, you know, Microsoft capabilities and AvePoint capabilities, you already achieved that for a fraction of the cost, right?
There, we can actually save customers another, you know, easily, like 60%-70% cost. So we see consolidation of different systems, of different providers, and we also see cost optimization. And also at the same time, storage costs. So if you actually look at the data's cost, just putting your data into SharePoint, right, which is Office 365, and OneDrive, there's cost to that. And that storage footprint continue to increase very quickly, faster at the rate of declining storage costs. Now, with AI especially, so Satya Nadella actually recently talked about by 2025, which is very near, 10% of all data generated will be done by generative AI.
So this is on top of the pace which we're already operating at, which is every two years, we are generating as a species, generating more data than all data accumulated, you know, since the beginning of mankind. So we're talking about exponential data generation already. So that means there's even more urgency to have proper, you know, data management, so that you can get rid of a lot of these out-of-date or trivial data so that, you know, t his term in our industry called dark data, right? Vast majority of time, 90% of your data are actually not actionable. So all these combined, so we actually, in cases, of—we have a case study on this with WPP, with one of the largest advertising.
Mm.
Company in the world. By signing on us, we're saving them millions in storage costs. Again, optimizing the storage out at the same time, providing data resiliency and business continuity because we're actually storing the critical data onto different platforms than just a singular, one hyperscaler cloud platform. So that's also part of the requirement. So we don't only operate and live within the Microsoft Cloud ecosystem. We actually already have significant footprint of data estate on AWS, as well as on Google, and also on-prem. So there lies the complexity, right? There also lies the need for enterprise customers and any business customers that don't want to rely only on one vendor to provide everything that they need.
It leads into a question on demand, because software investors have been debating for a couple of years now, where are customers in their IT optimization cycles? You all made some interesting comments on your earnings call, which is signs of life or early signs of demand stabilization. Tell us what the signs are. Where do you think customers are, both in optimizing their overall IT ecosystem budgets, and then more specifically, for AvePoint in your pipeline and your business, what are the early signs of demand stabilization?
Yes, maybe on the demand stabilization, what we were referring to during earnings was that, you know, last year we saw elongation of sales cycles. We saw really that kind of being the forefront of people's IT spend, right? They're being very critical over their spend. And so we kind of planned that in 2023. We kind of built that into our models, right? And our expectations. And so what we saw in Q1 and then further in Q2 was none of that getting worse, which to us was a really good sign, right? That we didn't see this continuation of, you know, more scrutiny or elongation further. We saw stabilization that our deal cycles seemed to lock in, and they got longer in Q4, but that did not continue in Q1 and Q2. So which was very encouraging, right?
'Cause we didn't, you know, we had planned that they would kind of stabilize, and they in fact did. So that was really good. So when we think about that demand stabilization, that's kind of what we're referring to. So that was a really good sign. And when we think about the journey of where people are, I think TJ's example of WPP is a good one. I think people are still in the early stages of now realizing they have massive amounts of data, and the costs are now starting to become real, and they're dealing with it, right? And fortunately, we have an opportunity to help them. WPP is a great example.
We worked with them for several months, working through the ROI, working through the potential savings, and it became a very cost-effective solution for them to implement, which we were able to help them do. And as TJ said, it's going to save, save them millions over the course of the first year, and that only gets bigger as the amount of data expands. So I think we're still in the early stages of people really focused on optimization, but we are seeing it this year. I mean, we're doing it in our own business. I'm sure most of you guys are doing the same thing, focused on controlling costs, and you know, and data is obviously a big component of that.
Yeah, I agree. I think we're still early stages. Even a recent study that's, again, cited by Satya was, you know, 85% of the C-suite see the need for digital transformation and going to cloud, and yet only 18% of them are satisfied with their own firm's progress. We see that during COVID, everyone's going to cloud in a hurry because that's the only way to scale and enable hybrid work. We know customers, for example, that only use VPN in Japan, they have to line up at 4:00 AM, like, 5:00 A.M. to get on to VPN queue. It doesn't scale, right? But now that post-COVID, so during COVID, everyone go to cloud just so that they can work, right?
What that then triggered is this mass, this big transformation agent that's actually COVID, not any management, you know, books or management edict. But then once everyone go to cloud, then there's the concern about who has access to what, right? So this governance thing became a very important thing. Security became very important. Now the next wave, of course, is with AI. So people also realize, not only do I need to work, go to cloud to scale my work, but I actually have to go to cloud to leverage the latest and greatest of technology has to offer, which is now generative AI. Whereas if I just stuck with my on-prem data centers, which, yes, it's more cost-effective, because I already have my OpEx spend done, that's not where the action is happening.
That's not where the, you know, latest iteration, where you as a business have to innovate, otherwise you'll be left behind. So there's a second push, you know, towards using cloud in a bigger way. But of course, at the same time, you know, you have to have the guardrails. We talk about information management, we talk about security, access rights, so that, you know, you also don't, at the same time, lose the crown jewel of the business by submitting a lot of proprietary data to open, you know, you know, a large language model query, and all of a sudden, potentially some of your querying information could be embedded into some of other people's answers.
So there's concern around security, around where you feed these large language models, both proprietary as well as public ones, is what's driving a lot of the debates among our customers and demands as well.
So how do you reconcile some of the big picture trends that are happening with your go-to-market and practice? Because you've got the three product pillars, you've got 40+ products. So I'll ask the question two ways. One is: How do you enable the go-to-market for your sales folks? And two is: Out of the 40+ products, are there one or two that investors should be focused on that you think can really move?
Yeah. So our customers don't actually see the 40+ products. The 40+ product covers different deployment scenarios and different data sources, right? So the way we actually sell is we sell functional areas, and we sell platform. We think platform is really the way our customer are thinking about things in term of consolidation. So we don't have a holistic competitor. We have point competitors across these major functional areas. But then our power comes from the platform play. We can get into accounts various ways, right? I talk about during COVID, everyone rushed to cloud, and the tip of the spear became the governance story. And then, prior to that, the tip of the spear was a migration story, right? Data analytics migration. And now with AI, it become a data analytics story and the information management story.
So we have different ways to get into engagements and then leveraging the power of the platform, be able to then upsell and cross-sell. And we'll do more of this. So as, again, you know, a pitch for the October #shifth appens Conference in D.C., you will see how we're actually embedding more AI capabilities across our products so that they actually work better together to incentivize more upsell, cross-sell, and ultimately to improve NR, right? But the ultimate goal, of course, is to provide more value to our customers. So they do see by betting bigger with a platform provider like AvePoint, we can actually overall save them cost and get better returns on their investment on Microsoft Cloud.
All right, I'm going to pause for a moment and go to the audience, please.
[audio distortion]
Yeah, that's a great question. So obviously we're imbuing more AI capabilities into our product to improve the top line, but in tuning for our own business is to improve bottom line. So for example, we talk about the support use case.
Mm-hmm.
So we actually, in our industry, have recently, one of the outside organizations surveyed 2,000 enterprises out there, and they found that we offer the best support as well as the best backup service offering in the Microsoft Cloud ecosystem. And we do have a 24/7 support organization, and it's located around the world for data sovereignty reasons. So we see that the ability to automate tier one support to actually provide actionable routing and case identification, essentially suck in 20 years' worth of support data, as well as all of our extensive user documentation into an AI model and provide that support create extensive savings for us. The other side of it is coding. So we use Copilot, GitHub Copilot.
But we do find that, similar to what Microsoft's reporting to us, privately as well, it's very helpful, for senior developers to use, GitHub Copilot to actually generate automatic routines. But it's not very helpful for junior developers, new developers, because you have to be able to cross-check and understand what's generated, before you deploy it, right? Because there's always chance of, based on probability and statistics, that what's generated is actually, you know, part of that quote, unquote, "hallucination," right? So it's helpful. We see productivity improvements up to 30% for our senior developers. For junior developers, they should still learn the ropes and learn how to actually do proper, you know, DevOps and, containerization and, cloud development capabilities.
One thing we do see is that, also in cloud, everything has a cost, right? Compute, bandwidth, storage. There's a joke in the industry, not so much a joke anymore, that Microsoft is laying off a lot of people to buy more GPUs. GPU is very expensive. So Microsoft has a Copilot program, preview for Office 365. Not only do they have to hand select, this is all done by the business side. You have to be a big enough client for them. To become part of that co-sell, Copilot program, the minimum entrance fee is $100,000. So it's expensive, right? So even if you're running experiments with GPUs, like what we do with Google and Microsoft. It's also costing them. So everyone's become very, very cost conscious.
So what that means, translated to code optimization, is that you need to make sure your code is optimized in such a way that it's cost conscious. Every time you query an AI model, there's a cost to it, right? When you do the parameterization. Also, the best practice in prompting for something like OpenAI ChatGPT is that your prompt should be no less than 400 words to be a good prompt. But when you pump in a lot of long-term prompts into the engine, every few tokens has a cost because of the way transformer works, right? It will run the model.
So all of these, your programming, have to be very cost conscious, and that's also something that, you know, yes, you can use AI to cross-check for you, but ultimately, there's also some type of, training, for our developers to be mindful of. But yes, absolutely, there's a, you know, bottom line improvements leveraging AI for the business.
And then maybe just one added to that is on the kinda, say, admin side of the house or our internal uses, we've been doing a lot of automation over the past several years, and so AI is almost just the next generation of that. So we're using it in some of our accounting systems, some of our operations. We're trying to embed as much AI as we can to make the, and streamline kind of the admin side of the house as much as possible. So we're evaluating a lot of tools, and we're starting to implement some of that stuff now.
We think it's only gonna be positive. As I mentioned earlier, any AI project, 80% of the lift is actually data cleansing, data aggregation, and data analytics. So, and that's our wheelhouse for the last 22 years. So this is why you will see new messaging from us in term of highlighting this AI-ready information management practice. It's really about aggregating of all your corporate enterprise data from different silos, hybrid and multi-cloud approaches, to make them a clean data set to pump into proprietary data model. So we think, AI, it's gonna drive further need to go to cloud and also further need to, you know, raise awareness of information management and governance even more than before. So we think it's only a net positive.
All right, I think we can end with a finance question.
Sure, sure.
So, target of Rule of 40.
Yes.
In 2025, and break even from a profitability standpoint.
Yes.
What are the one or two biggest needle movers that will get you there from a unit economic standpoint?
Yeah, so maybe just to level set again. For us, Rule of 40 is ARR growth, plus our non-GAAP operating as a percentage of revenue. So those two components make up our Rule of 40. So when we think about 2025 and getting there, I think there's a couple of things that are really gonna help drive that. First is our kinda shift to the channel and our focus on channel, 'cause that does two things for us. We're seeing real good success in terms of driving growth and helping us on the top line, so it's helping us with that ARR growth number. But it's also helping us from an efficiency play too, for our sales and marketing. So it does two things for us that have been, at least so far, very positive, and we expect that to continue.
And then again, on the efficiency side, when we look at our G&A costs, you know, the question before about AI and really just automation. So we spent a lot of money, obviously, from a G&A point of view, on becoming public company-ready, and over the past several years have really invested and built an organization that not only can be and support us being a public company, but we don't need to make those same investments that we've made in the infrastructure going forward at the same levels. So we do expect to see significant improvement in our G&A efficiency as well moving forward. So really, those two key drivers help us not only achieve the top-line growth but also the bottom line profitability.
Fantastic. We'll leave it there. Thank you both for your time.
Great.
Thank you.
Thanks, Gabriela. Thank you, Gabriela.