Appen Limited (APX) Earnings Call Transcript & Summary

Thank you, and hello, everybody. Welcome to the call. It's nice of you to join us at short notice. My name is Mark Brayan, the Chief Executive. I'm joined today by our CFO, Kevin Levine, and our Head of IR, Linda Carroll. The call today provides an update on our business and specifically our evolution from being a services business to a product-led business. We'll also provide you with a fresh reporting framework that is consistent with our new segments of operation and focus areas, and henceforth, we'll report in U.S. dollars to remove translation issues. Finally, we'll provide a trading update and have some time for questions. To Page 4 to start. Appen continues to evolve in many ways. When I joined the business 6 years ago, we were focused on language data. Today, we provide many modes of training data for AI, including speech, image, video, Relevance, LiDAR, et cetera. Our delivery model is increasingly product led. You'll see later in the pack that we have a rapidly growing revenue stream from our products. We're also working to improve revenue visibility and earnings quality with a focus on committed revenue. Our customers are currently concentrated to the U.S. giants on a revenue basis, but increasingly diversified on the number of customers. All new customers provide a platform for more diversified revenue going forward. We're announcing today a restructure from the functional organization to one align with our customer cohorts, more of that later. And finally, we'll report in U.S. dollars, henceforth, and we'll discuss that in more detail later in the presentation. If you could turn to Page 5, I'd like to take you through the evolution and rationale of these changes. Appen's early focus was on high-quality annotation services. The bulk of this derived from us providing crowd workers to annotate data on customer platforms. Our acquisition of Butler Hill in 2010 and Leapforce in '27 (sic) [ 2017 ] strengthened and grew that capability. Our technology, Appen Connect, acquired with Leapforce, was essential to the provision of crowd workers at scale, but the work was done on customer platforms in the main. Over the page in the acquisition of Figure Eight in 2019 gave us 2 important things: more enterprise customers, that is customers outside of the U.S. tech giants and our own annotation product. This product and subsequent engineering investments gives us the opportunity to sell and support anyone that wants training data, not just those that have been invested in their own annotation product. Our product has underpinned the increase in our customer numbers, 136 were added last year. It's also enabled the expansion of work with our larger customers due to its much broader capabilities than their own platforms and more on this later. To Page 7, and our product and engineering capabilities has enabled us to open in new markets. China and the Government sector, adding to our presence in the U.S. tech giants or global customers, as we call them internally and the Enterprise sector. Appen Connect continues to underpin crowd provision, and our training data products are in use across all of our market sectors, in some cases, alongside customer platforms. Page 8 and to sum it up, expansion beyond and within the U.S. tech giants has been enabled by our products. Our products are essential to our growth and will underpin our future. Our products will evolve as well to provide more features, automation and value for our customers. To put it simply, we are accelerating our transformation into an AI-powered provider of AI data and solutions. This transformation includes 2 core areas on Page 9. The first is that we will be product led. This enables us to provide high-quality training data faster, at scale and with improved unit economics. It also provides a platform for future products and capabilities. The transformation also includes a realignment of our structure to be more customer centric to support the needs of our target customer cohorts. We'll take you through the structure shortly. I'd like to spend a little time on the products, first of all. Page 10 outlines our current product suite. We'll go into this in more detail tomorrow in our Tech Day, but I'd like to highlight a few things here. Appen Connect on the left helps us recruit and administer our crowd of over 1 million contractors. It's essential for an operation of our scale and is used across all of our customers and projects. Our annotation platform is the product that enables our crowd workers to annotate data. It's essential for customers without a platform and in use by our customers with their own platforms as well due to its much broader capabilities. We have some new products as well. Our Appen Intelligence suite includes 20 machine learning models that automate annotation and administrative functions for us. It helps with speed, quality and unit economics. Our In-Platform Audit product helps our customers analyze their training data to detect bias and quality issues and improve the performance of their machine learning models. Our new mobile platform enables our crowd workers to engage and work with us via their smartphones and tablets, as well as a better and more flexible user experience, it opens up crowd markets for us in countries that are mobile-first and also enables more project types such as location-specific data collection. Page 11 highlights the value of our product suite. Our AI augmented annotation features improve the speed, quality and unit economics of data collection and labeling. Appen Connect enables at-scale crowd management and is essential to the operation and productivity of our business. It also solves an important problem for our customers. Our management of contractors on their behalf reduces their risk of employee misclassification. This is one of the reasons, by the way, that we do not see a shift to in-sourcing of data annotation. Large cohorts of contractors are in employment for our customers, and we manage that for them well. Our product suite, along with our 1 million-plus crowder contractors provides a strong competitive advantage. We have competitors, but in the main, they're focused on one class of use case such as computer vision. The extent and depth of our product suite and the fact that we've been in business for 25 years enabled us to tackle all data types and annotations and service more customers and projects. We'll go into more detail on our products in our Tech Day tomorrow, and we hope you can join us for that event. Page 12 provides a graphic of the new org structure. As of today, we have 4 customer-facing BUs, each with P&L responsibility. Our Global team is responsible for supporting and growing our 5 largest U.S. tech customers, Tom Sharkey, formerly our Head of Client Services, will lead this business unit. Tom's background in technology services, outsourcing and large account management, along with his track record of delivery for these accounts today is ideally placed to operate and grow our global business unit. Our Enterprise unit serves all other customers globally with the exception of China and Government. It has sales teams in the U.S., Europe and Asia and is focused on the tech sector and other developers of AI. Jon Kondo, formerly our Head of Sales, will lead this unit for now. Jon will be leading us to pursue his career as a large-scale global sales leader, and we have a search underway for a high-growth leader to run this team. Our Government and China business units remain as they are under their current leadership. These business units have P&L responsibility and can operate with the flexibility they require to support their customers in their markets. While the basics of data collection and annotation are consistent across these markets, the go-to-market customer culture and their relative positions in the AI life cycle all vary and require flexibility and agility to serve well. Our product-led approach creates a new product team. The product team has sat under engineering but now needs to be elevated. We're close to finalizing our search for a new U.S.-based Head of Product that will join the executive team report to me and greatly enhance our development of world class -- our world-class product suite. Our Engineering, Crowd and HR and Corporate teams remain as they are under the current leadership -- under their current leadership. These changes to the business require changes to our reporting, both segments and currency. To segments, first of all, on Page 14. There are 2 slices to our segments now: customer and product. And we combine these to give you a view of what we're focused on. The product segments across the top in the graphic are split into the work we do on our customer tools, i.e., where we provide crowd workers and the work we do on our product -- sorry, the split into the work we do on customer tools, that is where we provide crowd workers and the work we do on our product, where we provide technology and crowd workers. Our customer segments below product in the graphic include global, the U.S. tech giants, Enterprise, Government and China. The focus slice is Global Services and New Markets. Global Services is the revenue from our largest U.S. tech customers from the work we do on their platforms, the majority of which is their large Relevance programs. Although the work is done by our crowd on their platforms, Appen Connect is essential for the at-scale provision and management of the crowd workers. New markets includes all of the revenue we derive from our Enterprise, Government and China business units as well as the revenue from our global customers that comes from our product. New Markets is an important focus area for us to enable via products and benefits from the combination of at-scale crowd management with Appen Connect and highly productive, high-quality data annotation from our AI augmented product suite. An exchange to our reporting is currency on Page 15. Over 90% of our revenue and assets are in U.S. dollars and the translation to Australian dollars muddies our reporting. Reporting in U.S. dollars removes translation volatility and simplifies the comparison of performance over time. We'll commence reporting in U.S. dollars with our first half results this year and have provided a high-level 2020 P&L which is stated in U.S. dollars in this presentation, and there's more detail in the announcement that was posted on the ASX site this morning, along with this presentation. Pages 16 through 19 provide different views of our segments. Page 16 shows growth by focus area. The chart on the left shows half-on-half revenue growth in U.S. dollars from the services we provide to our global customers on their platforms. This revenue has been impacted by the deferral of a few large programs that impacted our 2020 results. The chart on the right shows revenue from our Enterprise, Government and China customers as well as the revenue from our Global customers that we derive from our platform. It has a much higher growth profile at 41% CAGR, illustrating the power of the product-led approach. Slide 17 breaks new markets into Enterprise, Government and China and product revenue from our Global customers. You can see that the product revenue from our Global customers is growing nicely. It underpins -- it is generally due to the growth in new projects from our major customers. But over time, we may see some large program shift to our platform as well. Although they have their own platforms, those platforms are quite specific to certain use cases such as Relevance. Our product handles a far greater variety of AI use cases across text, speech, natural language, image video, 2D and 3D data such as LiDAR and Relevance and enables our customers to develop a broader range of AI products. The right-hand chart shows nice growth from Enterprise, China and Government, albeit with a COVID hiccup in 2020. Also, China and Government are young businesses that are growing rapidly off a small base. Slide 18 provides another important view of our focus areas. Although growth in Global Services lags new markets, it is highly profitable revenue with improving margins due to scale and the benefits of Appen Connect. New Markets, on the right, included an investment phase in the first half of last year and broke even in the second half. This slide shows the thesis we've promoted for the last few years. Expanding margins in our large global programs will be reinvested for growth in new areas. The final segment view on Slide 19 provides revenue by customer unit and shows on the left that the combination of Global revenue on customer platforms and Global revenue from our products is growing nicely -- sorry, is a nicely growing, high-revenue business, again, supporting our thesis that we'll continue to get growth from our Global customers, but will get higher growth from our new customer segments. To Page 20 and some outlook statements. We expect our 2021 Global Services revenue to grow at mid- to high single-digit percentages on 2020 and be heavily skewed to 2H due to the pace of the return of work from the key projects delayed in 2020. We expect New Market revenue, that is revenue from Enterprise, Government and China, plus product derived revenue from our Global customers to grow at circa 25% this year on 2020, more in line with the broader AI market. Our new org structure gives us the opportunity to optimize our resources around our product-first future. This provides some savings that will flow through 2H 2021 and amount to USD 15 million on a full-year basis in 2022. Note that we may choose to reinvest these savings to further accelerate growth. To Page 21, an update to our order book. It now stands at USD 260 million and includes year-to-date revenue plus orders in hand. Note that this number is derived with a consistent methodology and timing of prior year numbers provided at the AGM. 2021 EBITDA will be heavily skewed to the second half of 2021 for a few reasons: Key projects that were deferred in the second half of 2020 are returning, but their delivery will be skewed to the second half. Our first half cost base reflects the full year cost of 2020 hiring and the benefits from our restructure won't be seen until the second half of this year. Overall, we are maintaining our underlying EBITDA guidance for 2021, restated in U.S. dollars on the page. Thank you once again for joining us for this presentation. I'll now hand it back to the moderator for questions.

[Operator Instructions] Your first question comes from Quinn Pierson from Crédit Suisse.

Maybe just firstly on your guidance comments. So at the February result, I believe you said expecting a high-teens EBITDA margin. Is that still how we should be thinking of it now? Or should we be thinking of it maybe more on the margins, at higher margin, given some of those cost benefits coming through in the second half?

Yes. Quinn, I would keep those margins in that range. There's some ups and downs in the numbers. And overall, I'd keep them in that range.

Helpful. Maybe secondly, so there's some interesting drivers on -- in your new markets opportunities. I'm sure you get some questions on that. I was hoping to just ask a question maybe on the longer term in terms of your Global Services. And that division did have a bit of a kind of a flattening of growth during COVID. You've guided for CY '21. That's helpful, mid- to high single-digit growth. I guess, to what degree do we think of that kind of revenue growth in Global Services as sustainable? I mean I think we get the concept that these customer algorithms are data-hungry and data-intensive. But as the revenue base gets larger and larger, the ability to kind of continue sustainably growing at attractive, call it, high single-digit revenue growth rates, I'd just be keen to get your thoughts on kind of the sustainability of that type of growth level.

Yes. Quinn, we're confident in the sustainability of that because it's core programs for our major customers. The flatness that you see on the chart reflects the deferral of those major programs that we saw in 2020. Absent of that deferral, we'd see mid- to high single-digit growth, and we don't see any reason why that will stop. What we could potentially see down the track is moving some of that revenue across to our platform. Now that's not assured because our customers may choose to keep those programs on their platforms. But if we're able to get some of that moved across to our platform, we'll see an uptick on our platform versus the services revenue.

That's helpful. And lastly from me, so you've called out USD 15 million of gross savings. I was just hoping you could maybe -- I mean it's a sizable number. I was hoping you could maybe elaborate a little bit more in terms of what's driving that -- what's driving those cost savings? And then how much of that to expect to be reinvested?

So the majority of the savings are personnel savings, and in terms of the reinvestment to be determined. We just want to send the signal that this is not -- this is a -- we're aligning the business to the future. And we want to be investing in that future. It's not an exercise just to take cost out of the business.

Your next question comes from Lucy Huang from Bank of America.

I just have 2. So firstly, with the Global Services revenue guidance and the skew towards the second half. So just wondering if you're getting any improvement in the visibility in terms of work returning in the second half from the global tech giants? And I also noticed that you mentioned that the order book has grown about $20 million since February. So just wondering how much visibility do you have to have the confidence in that second half guidance for Global Services?

The currency has changed on the order book number, Lucy. That's the first thing. In terms of the pace of return of these projects. So there were a handful of projects that were deferred in the second half last year, which impacted our full year results, and they were deferred for reasons of -- a variety of reasons. One, for example, was just the allocation of engineering resources. Our customers are building more AI products and engineering, AI -- engineering -- AI engineers, the programmers, the data scientists, the data engineers are in high demand, and so they're allocating them accordingly. So it's really an internal scheduling issue. And we are seeing a return of those projects, and that gives us some confidence for the second half. But the main impact on the business has just been those sort of internal scheduling issues on those projects.

Wonderful. And then just my second question. So I think previously, you mentioned there's a bit more competition, particularly in speech and image using your product-led business. Just wondering how do you think Appen is going to respond to that? Could we see an uptick in R&D in the near term to beef up that capability? Or do you think the current investment spend is sufficient at this point?

Yes. So we have a good suite of capabilities, and you're absolutely correct that the competitive environment requires us to have leading products, including our AI augmented products. I'd invite you to join the Tech Day tomorrow, where we'll showcase a lot more of those capabilities. In terms of future investment, hence, the comment around that $15 million. We don't want to -- it's not an exercise just to put money in the bank, it's an exercise to reallocate resources and capital to where we need it for the future.

Your next question comes from Bob Chen from JPMorgan.

Just a couple of questions from me. In terms of the sort of guidance for FY '21, does that guidance now include some of the restructuring benefits from the org restructure? And if so, do you know how much that's contributing to FY '21 guidance?

Not materially.

Okay. Great. And then just a question on, obviously, seeing some of those deferred projects coming back potentially in the second half. I mean does that impact sort of the skew into FY '22 as well, given the cycle of data refresh that usually occurs?

It's a bit early to tell, Bob. We -- I think by the time we get to the half, we'll have a clearer view of this sort of phase that we've been through. We are -- this is sort of a hangover from sort of COVID, and we anticipate that certainly in the U.S., that the economies are pulling out of COVID, and we may see a return to "sort of normal business", but it's a little early to tell at this point. But as to how that will sort of lead into 2022, we're sort of optimistic that 2022 will be back to sort of historic patterns that we saw pre-COVID.

Okay. Great. And then just finally, obviously, that $15 million sort of number that you called out seems pretty large. I mean which areas exactly are you sort of reducing heads in? Is it sort of some of the team managers or the sales staff that you're sort of pulling out in?

So if you could imagine a sort of a services-heavy business relies on a lot of people for delivery and the management of those people, et cetera. Also, in the back office, there's a lot of functions that our business has historically relied on. The deployment of technology to automate and enhance the way we do things just simply means less people in some of those areas. So we haven't broken down further than that, and we won't. But the short answer is a tech-enabled business needs fewer people in some of those service areas.

Your next question comes from Wei Sim from Macquarie.

I just had one question actually. And it's just in regards to this pivot that we're talking about going from -- our revenue model being from project-based to a committed model. Is this something that we've discussed with our Global customers? And what kind of feedback they might -- they have given from that?

Yes, it is, Wei, and we do derive some committed revenue from our Global customers. So it provides some benefits to them and us. And for the right projects, it's -- they're happy to entertain it.

Great. Is there any color as to what kind of level we may see being coming through as committed revenues versus project-based now versus previously?

So we reported on that at the full year, and we'll do so again at the half year.

Your next question comes from Paul Mason from E&P.

Just 2 from me. The first one, just on your revenue and work in hand. I've done some basic math and it's sort of pointing to about 10% up versus the same time last year in U.S. dollar terms. I maybe haven't got the number perfect there. But -- and so I'm just wanting to equate that like in terms of the split of your different sort of guidance buckets between Global Services and New Markets, blending that out, are they sort of correlating together at the moment? Or is there something else about that work-in-hand number that is maybe making it a bit higher right now because there's something about second half in there or anything like that. Have you've got any comments on that?

Yes. I don't know if I fully understand your question, Paul, but certainly, your back-of-the-envelope math sounds about right.

Okay. Great. And the second one for me, I know you haven't provided a number on this, but maybe just to steer, if you can, in terms of the quantity of restructuring costs. Like are they expected to be sort of around the magnitude of the savings but on a onetime basis or a bit less or a bit more?

Yes. No, we haven't disclosed that.

[Operator Instructions] Your next question comes from Ross Barrows from Wilsons Advisory.

Great. And just a question, if I could, regarding the breakdowns that you've got on Slide 16 and 17. I think in terms of the growth last year didn't grow as fast as you had hoped, certainly still grew. Are you able to give us some context around whether it was the second half. So the Global Services on Slide 16 or the Global Product on Slide 17 in the second half '20, where you hope to have stronger growth that didn't eventuate. Is there any color you can give around that would be great.

Yes. Ross, the -- from the services perspective, it's due to those larger programs. So the slowdown in that and the deferral of that work is directly visible in the Global Services revenue on the left-hand side of Page 16. Page 20 of the left-hand chart, the Global Product revenue is largely due to new programs that have been enabled by the product.

Yes. Okay. And that's where you're hoping to get incremental traction over time?

Yes. So our customers have platforms that are predominantly centered on Relevance data. And while we'd hope to get some of that work across onto our platform, their platforms are sort of stitched into their infrastructure. And so it's not as simple as it appears. Where will benefit the most -- where the Global Product revenue will benefit most from us winning new work for new use cases and new AI development. And yes, it's absolutely a focus of ours.

That's helpful. And just to revisit a comment from before, just around the revenue moving from project-based to committed. I'm assuming that's the -- well, I think you said before, it's the ACV number that you have disclosed, and you said you will continue to disclose. Just confirming that's kind of the conversations you've had with those larger players. And one has agreed to move on to it or at least quote some work with visibility and the goals to get more of those larger customers to do the same, is that -- just confirming what you said in the past.

Yes, that's right.

Yes. Can I just add to that because it's -- the comment came up here and it's come up before. And that is we've been -- we have been working towards this for quite some time. And when we reported at the full year number, I think we reflected that circa 30% of our revenue is now committed. So we're very much on that pathway. I certainly want to dispel the myth if some people think this is something that we are just starting from. We are moving through that. And we have had success, as we reported previously, with a large global customer with a large committed deal that we've done with them.

And just to clarify, the ACV is for the current year, and there's no publication or you haven't made any comments around committed revenues beyond the current financial year, correct?

That's correct.

Your next question comes from Andrey Mironenko from Alphinity.

Mark, just a question following from a question from Ross on Page 17. What exactly is this Global Product revenue? Would the bulk would be the same crowd data annotation services but done by large global tech customers over the -- but over your platform, but the bulk of revenues, not for the subscription for the platform itself, but for the data annotation services? Or is it something very different?

So the Global Product revenue is revenue that we derive from our global customers use of our product, whereas services is when we're providing the crowd workers on their product. Is that -- was that the question, sorry, Andrey?

Yes, it was. But I guess -- but what exactly does that mean, right? So when you're saying $26 million in second half, for example, that's $26 million, does it include -- like if a big global customer uses your Figure Eight platform to purchase that annotation from your crowd would only the portion that they pay for the subscription to Figure Eight be part of the $26 million? Or would all the actual work and annotation of the data be purchased on -- through Figure Eight, would all of that included in $26 million.

Yes, sorry. Yes, the technology, it's a combination of the technology and the crowd workers because that's the powerful piece in what we sell. Technology on its own is very, very powerful, but the crowd workers are the folks that do the work. And having them on our platform enables us to deliver a broader set of data modalities. It enables us to deliver higher quality data because the platform manages the quality. For example, it will automatically kick out poor quality data. It also enables us to improve the throughput and the productivity of the crowd workers. So yes, the short answer is that the Global Product revenue includes both the technology and the services component. And the combination of those things is what really makes us highly competitive.

Your next question comes from Stella Wang, a private investor.

Just 2 questions, please. The first one is further to the questions in terms of the cost saving. Now your restructuring sounds more like orienting it toward accelerating growth. So how should we reconcile this with this cost cutting? Is this cost saving would still happen without the structural change because you've got the in-house technology advancement, you just [ choose to ] clarify a number on this together with this restructuring? Or is there things you are looking at reducing size of and something you're looking at increasing size of?

So Stella, think about this as a realignment of our resourcing to our product-led future. It's not -- and yes, there are some cost advantages to doing this realignment because product businesses typically use or rely on fewer people than service businesses. But the rationale behind it is not to save cost. The rationale behind it is to realign our resources to our product-led future. We call out that figure of $15 million in full year potential savings in 2022. But as we say, we may choose to invest some of that back into the business to further our competitive advantage in sales, marketing or product. So it's not an exercise to cut cost, although there are cost advantages. It's an exercise to realign the workforce and the business to that product-led future.

So some of this figure would have happened anyway because of the internal efficiency gains through the technology?

Yes. Over time, we would see the need for fewer resources due to technology. However, we are accelerating that because the figures, for example, the New Market chart on Page 16 shows you that, that's where the growth is. So we're going hard into that and accelerating that transformation and with it the realignment of those resources.

So I guess in terms of the product development and R&D expenses, we shouldn't expect that to tail back in any form?

No.

Okay. Just a last question, please. In terms of the order in hand, in USD terms compared to last year, around this time, it's up just about 10%. Reconciling this back to your top line guidance, is this current figure including the deferred project? Or are some of the projects still to be factored into a second half updated number?

So it relies on the methodology that we used in prior years and includes everything we see at the moment.

So it's whatever deferred project is already back on. You do have already factored that in?

Yes. To the extent that we see those projects, they're factored in.

There are no further questions at this time. I will now hand back to Mark for closing remarks.

Thank you, and thank you, everybody, for joining us on the call today. I'll invite you all to join the Tech Day tomorrow, where we'll go into much more detail on the products, including some demonstrations of some of the new features that we've developed recently and are deploying in the market. So we look forward to talking to you then. But thanks once again for joining the call, and looking forward to speaking to you real soon. Thank you.