Operator: Ladies and gentlemen, thank you for standing by, and welcome to the Appen Limited FY '21 Full Year Results Conference Call. [Operator Instructions]. I would now like to hand the conference over to Mr. Mark Brayan, Chief Executive Officer. Thank you, and over to you, sir.

Mark Brayan: Thank you very much, and hello, everybody. Welcome to the call for our full year results for 2021. I hope you're all doing well and possibly back in your offices as we get to -- used to living with COVID. I'm traveling regularly once again, and I'm pleased to be able to spend time with our customers, partners and staff in our offices all around the world, so life returning to some sort of normal. My name is Mark Brayan.

I'm the Chief Executive. I'm joined today by our Chief Financial Officer, Kevin Levine; and our Head of IR, Rosalie Duff. The presentation was loaded to the ASX website this morning. I'll be referring to that throughout. The presentation will take 30 to 40 minutes, and then we'll open up for questions, and we aim to finish

at 12:00 noon Sydney time.

As a reminder, before we get going, all the financials are in U.S. dollars. This year's presentation is a little richer than prior years. It includes an update on our market, an overview of our business and importantly, the outcomes of our growth strategy review that we undertook last year and, of course, the financial results. So to Page 5 to start.

Our vision is to make AI in the real world -- work in the real world. And AI relies on training data, and it needs high-quality data to perform well. We aim to be the #1 provider of data for the AI life cycle, and by doing so, we'll help our customers build high-quality and responsible AI. We're a full service provider. Our products cover the data step -- data-heavy steps of the AI

life cycle: data sourcing, data preparation and model evaluation.

We deliver our products with a combination of our tech platform, our expertise and our crowd. We're pleased to announce record revenue of USD 447 million this year and EBITDA of $78.9 million, and that's before the impact of foreign exchange. We've grown revenue at 40% per annum over the last 5 years, which is an extraordinary achievement and a credit to our talented team. We delivered into the second half skew that we forecast at the first half. This was underpinned by a 32% half-on-half increase in Global Services revenue.

This is revenue we derived from providing crowd services to our large tech customers. The sharp uptick in this revenue returns it to its year-on-year growth trajectory and shows the health of our customers. We're also doing very well in China with revenue up 422% on last year on the back of strong sales to the tech giants, autonomous vehicle and mobile sectors. We count the world's most advanced companies amongst our customers, Google, Amazon, Microsoft and others have relied on the many modalities of high-quality data that we provide them for many years. Some of the world's best-known products are powered by our training data.

This is a tremendous achievement for an Australian company and one that we're very proud of. To Page 6 now and our nonfinancial impact. Our global crowd of over 1 million contributors is essential for our business and the data that we provide for our customers. It's also a responsibility. We are a member of the Global Impact Sourcing Coalition that creates opportunities for people in developing countries.

Our recent crowd survey told us that 17% of respondents were long-term unemployed before they found work with us, and 16% lived under the poverty line prior to working with us. At 63%, almost 2/3 use the money they earn from us to support their households or their education. We're supported by our talented global and diverse team of linguists, engineers, data scientists, project managers and other professionals. This talent and diversity helps us deliver unbiased data for our customers. We've also increased female representation at the senior management and Board level this year to 38% and 50%, respectively.

Our business has a low environmental footprint, but we're committed to reducing it, nonetheless. We've completed an inventory of our Scope 1 and 2 emissions and will be net zero by 2030. Page 7 brings the impact of our work to life. We worked with one of our customers to ensure that their language generator was inclusive, unbiased and that it worked for everyone. We did this by using a diverse team of contributors from our global crowd.

They were from different cultures, ethnicities, genders, ages, and they tested the product and they provided the test data for our customer to tune the products so they would ensure that it worked in the real world for everyone. And we've all heard horror stories of AI that behaves badly. So we're very pleased to be working with our customers to ensure that their AI works well for all of their users. To Page 8 -- Page 9 and the AI market. A recent survey by PwC showed that not only was the use of AI becoming more pervasive and mainstream, but half of the companies they surveyed had accelerated their AI plans because of COVID.

This is due to many reasons, including the need to alleviate staff shortages. Some of our customers are building chatbots for customer service as their customers and the need to look after them runs ahead of their ability to attract and train staff. The growth in AI is driving the need for training data, as you can see from the chart on Page 10. This chart from research firm Cognilytica is very important. It's some of the first comprehensive research on training data.

We're often then asked if techniques such as self-supervised learning will make training data redundant. Our view has always been that the need for training data will continue to grow alongside the emergence of new technologies, and this research confirms that. The chart shows the growth of different training data delivery models. Full service, where the vendor provides the technology, crowd labor and project manager -- management to deliver the required training data. Labor only, where a vendor uses a third-party platform or customer technology and provides the crowd or BPO labor, BPO, business process outsourcing.

Nonmanaged crowd, where the vendor provides the technology and access to a crowd that doesn't manage the process for its customers. Labeling tools, where the vendor just sells the tech platform and supports the customer on that platform but doesn't provide the crowd or project management. And finally, synthetic data, which is an emerging but important area to watch where data is fully synthesized and comes complete with labels. Now we are a full-service vendor, and we participate in the majority of these delivery models to vary -- to varying degrees. We provide fully managed service as well as labor only using our customer platforms, and we also have customers that use our platform and access our crowd without our management or support.

Now on the right-hand side of the page, you can see that customer needs are evolving, but we see many of the same themes year-on-year. Firstly, that scale and quality are important. The latter quality is particularly important. Poor training data leads to poor AI. Now synthetic data will play a role, but the nuance and the specificity and the quality requirements to training made us see continued reliance on humans, as you can see from the chart on the left.

Plus, not every stage of the data life cycle can be automated or synthesized, as we'll see in a few slides. And area that's been with us for a while, but it is emerging in its -- or increasing, sorry, in its importance is trust and privacy. Data privacy, ownership governance, providence are all emerging as must-haves for anybody dealing with data. Another thing we learned this year is that because AI is experimental in its nature, you don't know the outcome until you build it, our full-service customers need to be agile, and they're reluctant to commit to data volumes or annual spend. So we're putting our customer needs ahead of ours, and hence, committed revenue is not a focus for us, and we won't report on it henceforth.

We support all of these customer needs and requirements very well, and we are the market leader, as the chart on Page 11 shows. We are close to double the revenue of our nearest competitor and many times larger than the others. We've achieved this position through a close alignment with our customers and their needs. We're trusted to deliver and to look after our customers' data. We maintain our reputation for high-quality data and our platform, people and crowd combined to support our customers' needs for usability scale and speed.

Our 25 years of experience give us depth and unparalleled expertise. We maintain a strong position against our competitors. We have capacity and scale that betters the biggest players. Sorry, over the page, pardon me, to the competitor group. We maintain our position against our competitors.

We have the capacity and scale that betters the biggest players and the breadth of technology that keeps us ahead of the tech-forward competitors. Both of these elements, capacity and technology, are necessary in our market, and the depth of our capabilities is an effective moat vis-à-vis our competitors. To Page 14 in our business on a page. We are and will continue to be the #1 provider of data for the AI life cycle. We provide what our

customers need: trust, the quality, usability, scale and speed.

Our products support the data-heavy stages of the AI life cycle. We collect and originate training data for our customers in many ways and in many data modalities. We've done this for years, and we continue to enhance our capabilities, for example, with the recent acquisition of Quadrant and our ability now to collect point-of-interest data. We're also working on some synthetic data. It's early, but this will play a role.

We prepare data for ingestion and models by labeling and/or organizing it, subject to the requirements of the use case. We do this with our crowd and using machine learning to automate these processes to improve our scale, quality, speed and the value we provide to our customers. We leave the model training, that is the development of the model, and deployment to others. Building models is compute heavy, and that's not our core competence. There are others that are far more expert than us.

So we're building partnerships with them to be part of their ecosystem and to help them support their customers. Finally, we provide essential evaluation and testing services. Much of the relevance we do -- much of the relevance work we do falls into this category. We provide specific crowd cohorts that respond to the demographic needs of our customers in order to test their search and social media applications at scale. We deliver our products with a combination of our technology platform, our crowd and our expertise.

All are essential, and the extent of each varies depending upon the project and the use case, and we cater for all

data modalities: text, image, audio video, 3-dimensional LiDAR, multimodal data and point of interest. Our ability to cover all of the data-heavy stages of the AI life cycle, all customer requirements, all delivery modes and all data types make us a truly unique and powerful full-service partner for our customers. We invested in a refresh of our growth strategy last year, and I'd like to take you through that now, if you could turn to Page 16. Our growth strategy has 4 core pillars. We'll continue to grow our revenue and diversify our customer base.

We have 5 customer-facing business units to ensure a focus on our customers as well as growth outside of our large Global customers. Our BUs Global, supporting the 5 tech giants; Enterprise, who support commercial customers in the U.S., Europe and Asia Pacific; China; Government; and Quadrant are all highly focused on their customers and have the resources they need to win and deliver in their end markets. The next pillar is automation, and we're seeing early success in our efforts to automate crowd and labeling processes. We'll continue to rely heavily on the crowd, of which more later, but our automation will reduce the cost of the crowd and enhance speed, scale, quality and value of the data that we provide for our customers. Our data science team is an important enabler of our automation strategy.

We're expanding our product offerings. We want to be a one-stop shop for training data needs. So the more use cases we cover, the less reasons our customers have to go to our competitors. We have a point-of-interest data, and we're exploring synthetic data. We'll continue to invest in product and engineering to enable this expansion.

We're also working to evolve and transform our internal processes. We're investing -- we have invested in a transformation office to enable this change. We have many complex and manual processes such as recruiting our many crowd workers, and we're working to streamline these to improve the service that we offer to our customers in crowd and enable our teams to focus on higher order work. Implementing this strategy is a journey. So we've set ourselves 3 long-term 5-year goals.

The first is to grow and at least double last year's revenue by 2026. Secondly, to diversify. We want to have more than 1/3 of our revenue from our non-Global customers. That is customers outside of the 5 U.S. tech giants.

And of course, we're a profitable business and will maintain that and set ourselves an EBITDA target, a margin target of 20% by 2026. It is a journey, but we're on our way, and you can see some progress on Page 17. To support the growth pillar, we expanded operations in China in 2021, hired a new leadership team for Enterprise headed by Jen Cole, and hired Sujatha Sagiraju as our Head of Product. Jen and Sujatha are highly skilled and impressive executives and have made an immediate impact. Their bios are on our website.

For our automate pillar, we pulled together a world-class data science team and have developed models to automate speech data preparation. We're active on customer projects with these models. We'll deliver these projects, improve the models and move on to the next project and the next use case. We've added new products this year, point-of-interest data via the Quadrant acquisition, and we had great success with our autonomous vehicle product in China, of which more later. Finally, we've built our transformation team headed by Eric de Cavaignac, who is an experienced change executive.

He and his team are deep into the process analysis and improvement stage of that transformation project. We lay out some of the 2022 steps and investments on the page as well, all of which contribute to the goals on the right. And we'll update this slide as we progress along the journey. We'll now dive a little deeper into some elements of the strategy, starting with automation on Page 18. As a full-service provider, we need our crowd, our expertise and our technology in different measures across the AI life cycle.

This chart shows the relative human involvement and automation potential of the different stages. Data collection, for example, is highly manual. You need humans to speak into their phones if you're collecting speech data or to take photos of buildings if you're collecting point-of-interest data in the field. Now steps can be automated for efficiency, but humans are essential as they are for model evaluation and relevance, which is on the right-hand side of the chart. Contrast is, for example, with synthetic data, which is compute intensive and highly automatable.

Humans play a limited role. So automation is important, and we're investing to deliver its benefits for our customers, but not all elements of the AI life cycle can be automated to the same extent. So our crowd and those related capabilities give us an effective moat against competitors in these areas. Relevance, the bulk of our revenue is crowd dependent, and hence, the risk of disruption by automation is very low to nonexistent. Page 19 highlights one of our new products.

We're having great success in China in the autonomous vehicle market. Autonomous vehicles require vast amounts of training data, and we're working with 11 auto companies in China as well as over 20 other tech companies such as drone providers who are investing in autonomous mobility. Slide 20 highlights point-of-interest data. Point of interest or POI is important to keep maps up to date, especially now given COVID. For example, small businesses that have closed may still appear open on the maps, and it's important that, that's all updated.

The Quadrant product enables very efficient collection of point-of-interest data, and the data supports mapping e-commerce and marketing applications. There's also an interesting intersection between point-of-interest or geolocational data and augmented reality. I'm sure there will be apps in the future that detect where you are and automatically provide directions or translations and other information via wearable devices like glasses. It's a pretty interesting area of technology and one that we're glad to be part of. Let's turn now to the financial performance starting on Page 22 and an outline of our reporting segments.

So we have 2 reporting segments and 5 customer-facing business units, and the reporting segments reflect our product-led strategy. So for example, we have Global Services, which are the services that we provide to leading U.S. tech companies on their platforms, and we have New Markets, which is all the customers that we support, including some of our Global Services projects on our platform, on our technology. Our 5 customer-facing units include Global, which are the 5 largest U.S. tech companies; Enterprise, which cover other companies in North America, EMEA and Asia Pacific; Government, federal agencies; China, which now encompasses China, Japan and Korea; and Quadrant, a provider of location data.

So over to Page 23 and the financial highlights. So we had another record full year revenue performance. Revenue was up to USD 447.3 million, and that was up 8% on last year. Key drivers were a sharp uptick in Global Services revenue in the second half that grew 32% on the first half of FY '21. New Markets revenue was also up sharply, 21% to $102.5 million, and that was driven by a very high increase in revenue in China.

Underlying EBITDA before foreign exchange grew 12% to $78.9 million, and that was driven by the revenue growth and some gross margin expansion in the second half. We maintained a strong balance sheet, $48 million in cash and no debt as of the 31st of December '21. And we're pleased to provide a final dividend of AUD 0.055 per share, and the total dividend is flat on 2021. So digging into the segments now to Page 24 and Global Services. So the chart on the right shows the uptick in Global Services revenue for the first half and the second half, and that came on the back of a number of the expansion of existing projects and a number of new projects in the second half, and this was consistent with the skew that we called out at the first half -- at our first half results.

Overall, revenue was up 5% to $344.7 million and EBITDA up 3% to $91.2 million. To our next segment of New Markets, and you can see a sharp year-on-year growth of 21% to $102.5 million and a pleasing half-on-half trend. China played a big role in this growth, as we'll see in later slides. Over the Page 26 and the Global customers. So Global revenue overall was up 3.4% to $386.3 million, and you could see again that return to growth in the second half on the right-hand side.

Importantly, you can also see the percentage of projects that our customers are working on -- or sorry, the percentage of projects that we're working on with our customers that are not related to ad products continues to climb. This is very important and shows that our customers are investing in AI products outside of advertising. To Page 27 and just a little bit on the digital advertising market, which is important. First of all, the advertising market, which underpins a lot of the revenue of our biggest customers, continues to grow, very healthy upward trend for digital ad spending worldwide. It is a dynamic market.

As you're all aware, there are many changes in the market, including, for example, the recent change to the iOS our operating system that impacts the data that search and social media companies can collect from Apple phones. What this means -- the combination of those dynamics and the chart on the left mean that the spend is still happening, but it's shifting between vendors. What it also means is that the major search and social media customers -- companies, of which many are our customers, are working hard to solve this problem to return to those highly personalized ads. The impact of these trends on us is threefold. First of all, we're working with our customers to build products that are not related to ads, which is to our advantage.

We also work across many of these companies, both in the U.S. and China. So as ad spend, which is growing, shifts from one to the other, then that can be advantageous to us as well. And finally, as our customers look to solve this problem, it may provide opportunities for us as they seek new sources of data to create highly personal ads. Over the page now -- Page 28 and our other BUs, Enterprise, China, Government, Quadrant.

So strong growth, '21 revenue of $60.8 million, up 55% on FY '20. This has a lot to do with success in China, but the other business units played their part. And you can see on the right-hand side, tremendous growth in those business units. And very important for us to improve the amount of diversity that we have in our customer base and the revenue that we derive from these customers is now 14% of our total revenue, which is up from 9% last year. Over the page to Page 29.

And the chart, as you can see on the left-hand side, quarter-on-quarter revenue in China is growing extremely healthy. Our customers include the tech giants, social media companies, mobile providers and autonomous vehicle companies. As I mentioned earlier, we have 11 of the leading AV companies or autonomous vehicle companies as customers, and we have more than 20 other tech companies such as drones and robotics that are using our data for autonomous mobility. China operation is highly focused on growth. That includes growing projects with existing customers as well as new customer acquisition, and we are on track to be the market leader in China.

So we're really pleased with progress there, and you'll see further growth -- further high growth in China. To Page 30 and an update on our other business units. The Enterprise team had a solid year but yet to fully reach their potential. They're winning a lot of projects and winning a lot of customers. And of course, we have a new leadership team there, highly focused on accelerating growth.

Our Government team was successful recently in being selected in the partnership for the Joint Artificial Intelligence Center Blanket Purchase Agreement to support AI -- the development of AI capabilities. That positions us to win projects with a number of agencies. And Quadrant is integrating with the business and winning work with the Appen customer base, including many projects or a number of live projects with some of our largest customers. I'll now hand it over to Kevin, to Page 31, to take you through the rest of the financials.

Kevin Levine: Yes.

Thank you, Mark. So on to -- on Slide 31, our revenue and other income increased 8% to $447.3 million. This reflects a strong half-on-half performance with the second half up 28% on the first half, reversing the first half decline of 2%, and that's comparing to the prior corresponding period. There were 2 major drivers for the record second half performance. Firstly, Global Services, which delivered a significant turnaround from the first half revenue reduction of 9% with second half revenue up 19% versus the prior corresponding period and up 32% on the first half of FY '21.

This record second half revenue performance was driven by continued increase in non-ads and with ad-related projects returning to growth, as forecast. This highlights the value that our Global customers place in our ability to deliver high-quality data at scale across all data modalities. The second driver was China with 422% growth coming from both increases in market share, i.e., from new customers as well as in customer share, i.e., expansion from existing customers in new and existing projects. Underlying EBITDA, excluding an FX loss of $1.2 million, increased 12% to $78.9 million. This result in the associated margin was positively impacted by strong second half drivers, namely revenue growth, gross margin expansion and moderate expense increase to support the growth.

Underlying NPAT reduced 10%, impacted by increased amortization of product development investments. The effective tax rate of 20.5% is in line with the prior year. The effective tax rate is subject to fluctuations from the tax effect of movements from expensing, investing of employee performance shares and differences in overseas tax rates. Excluding these fluctuations, we have reduced the normalized tax rate to 25%. Over the page and to the balance sheet on Slide 32.

Balance sheet remains strong and resilient with no debt. Trade receivables increased by $38.6 million to $89.2 million, increase in trading volumes approaching year-end. Invoices were raised on 30th December for work completed in December as the billing milestones were satisfied. This resulted in an increase in trade receivables and a corresponding decrease in contract assets. Noncurrent assets comprise mainly goodwill and identifiable intangible assets, mostly arising through acquisition.

Following a detailed impairment review, we report adequate headroom in the carrying value of these intangibles. Noncurrent assets increased mainly due to $45.4 million being recognized as goodwill relating to the Quadrant acquisition. Total liabilities increased by $19.1 million to $107 million, mainly due to the earn-out liability of $18.4 million associated with the Quadrant acquisition. The final dividend of AUD 0.055 per share has been declared. This is in line with last year and is 50% franked.

This takes the full year dividend to AUD 0.10 per share in line with FY '20. Over the page on to Slide 33. In 2021, we invested $30.2 million in product developments, representing 6.8% of revenue. This focus is important to drive customer growth and repeatability as well as quality improvements and margin expansion. Since FY '19, we have strategically invested in engineering resources to develop new products.

68% of our product spend was capitalized in FY '21, up from 64% in the prior year, reflecting our commitment to development of new products and tools. We will continue to invest in product development of up to 10% of annual revenue. Moving on to Slide 34 on the cash flow. The cash on hand at year-end decreased by $12.6 million. However, this was due to the upfront payment for Quadrant of $25.3 million.

Cash balance and cash conversion were impacted by timing issues, primarily due to the working capital cycle impact from the higher volumes in November and December. Cash flow from operations reduced by $17.9 million due to the aforementioned working capital cycle impact. However, this was somewhat offset by lower tax payments. Cash has been effectively deployed for product development, tax, dividends, OpEx and growth investments. Notwithstanding the cash flow cycle impacts, the cash flow conversion from EBITDA was still solid at 77%.

Thank you, and I'll hand you back to Mark.

Mark Brayan: Thank you very much, Kevin. So to conclude and to move now to Page 35. We're very pleased with our performance in 2021. We also refreshed our growth strategy to transform the business with a focus on revenue growth, customer diversification, automation and product expansion.

Implementing this strategy is a journey, so we're taking a long-term view, and we've set ourselves 5-year targets as our North Star. Those targets are to at least double our FY '21 revenue by 2026. This demonstrates our focus on long-term revenue growth. We're off to a good start. Our revenue order book for this year, which includes year-to-date revenue, plus orders in hand, stands at USD 190 million.

We also expect the FY '22 half-on-half revenue skew to be similar to prior years, excluding FY '20. The second target is to improve the mix of customers with 1/3 of revenue from our non-Global customers. That is 1/3 of revenue from customers other than the 5 U.S. tech giants. And we'll achieve this with investments for growth in new products, sales and marketing partnerships.

We'll explore M&A opportunities, and we're targeting higher than 35% compound annual growth revenue from non-Global customers, and this is in line with the market growth rate of the chart earlier in the deck. Finally, we are a profitable business, and we'll maintain that. We've set ourselves a long-term goal of 20% EBITDA margins. But our focus on revenue may impact near-term EBITDA margins and future dividend payout ratios. This long-term focus also means that we will no longer provide short-term EBITDA guidance.

A few things, higher costs in the first half of '22, including the transformation office, investment in product and technology and share-based payment expenses will impact first half earnings this year. And there will be an earnings skew to the second half, which could be larger when compared to FY '21. In closing, I'd like to thank our customers for their support and especially to thank all of our talented and hard-working teammates around the world. We appreciate everything they do for us. This result is theirs, and I'd like to thank them.

And now I'll hand it back to the moderator to take time for questions. Thanks very much.

Operator: [Operator Instructions]. Your first question comes from Josh Kannourakis with Barrenjoey.

Josh Kannourakis: Mark and Kevin, can you hear me okay?

Mark Brayan: Yes.

Josh Kannourakis: Yes. Just first question, just with regard to some of your core customers. You mentioned that the non-ad-related projects obviously saw some good growth in the period, and that was sort of on a project basis. Would you be able to give a bit more context on how that's come through on a revenue basis and just how -- I guess, your comfort sort of, I guess, in the future outlook of some of those non-ad-related projects?

Mark Brayan: Sorry. That -- the line on Page 26 that gets to 77%, that's revenue.

So the non-ads revenue is 77%. Sorry if I misspoke.

Josh Kannourakis: Yes. No, that's fine. And so I guess in terms of those projects, though, in terms of -- because, obviously, they're still at early stages, is it too early also for you to say in terms of the potential materiality long term from those projects?

Mark Brayan: Yes.

Hard on a project-by-project basis, but some of them are to do with augmented and virtual reality, for example, which we know is a growth focus for some of our customers. So it's hard to know how any individual project will play out, but they're in areas that are growth-focused for the customers. Another one is e-commerce, another growth area. Another area is mapping. So these are all in forward-looking areas, but hard to tell how an individual project may play out at this point.

Josh Kannourakis: Got it. And I guess some of our industry feedback has suggested data collection works have been quite a significant part of the market or a growing part of the market. Would you be able to comment on, I guess, how you're sort of seeing those trends there and Appen's ability to capture share of that market on a go-forward basis?

Mark Brayan: We have seen an uptick in data collection, and it is a competence of the business with our crowd-based model because, as I explained earlier in the presentation, a lot of the data collection needs humans; hence also the acquisition of Quadrant, which also goes to data collection. So yes, I agree with the industry feedback you're getting. It's very important, and we're very much in the thick of that.

The reason why -- sorry, just to add one more thing. The reason why data collection's important is because the best-performing AI is typically narrow AI, something that doesn't reach a specific path. So correspondingly, you need data that's representative of that use case. Now not all the data may be available from current sources or by the customer. So if they need a specific piece of AI that needs specific data, they have to go find their data and collect it, and that's what we're doing.

So yes, a lot of different data collection projects currently.

Josh Kannourakis: Got it. And just a final question. I'm sure there will be a few around margins, so I won't spend too much time on it. But in terms of, I guess, your longer-term commitment versus the shorter-term investment, should we be thinking about some of these shorter-term investments as transitory in nature or more in terms of a step change in terms of how the business needs to operate and the cost that needs to put in to fulfill those growth objectives?

Mark Brayan: Yes.

So I think in the sort of the -- what we foresee immediately is more incremental, but we just need the flexibility to make those investments as and when they come up. It could be, for example, lining up a team of people to build a particular model to automate part of the business. So we don't just see any step change-type investments in the near term, but it's not to say they may not come along, but it's more likely to be sort of incremental at this point.

Josh Kannourakis: Got it. So is there any way or any comments you can make in terms of how you're thinking about, I guess -- obviously, you talked it's going to be below 20, but is there any numbers you're sort of comfortable talking about as being above in the period or on a go-forward basis to give people a bit of comfort?

Mark Brayan: Yes.

We wouldn't want the percentages to go backwards. The objective is to keep going forward with a long-term view of getting to that 20% target.

Josh Kannourakis: Okay. That's great context.

Kevin Levine: Yes.

Sorry, Josh, if I can just add to that as well. And that is, yes, obviously, we're going to manage the cost base and prioritize the spend to align with those long-term growth objectives, so around product development and expenses to support the growth and then, obviously, calling up to -- in the short term, as we called out just some impacts from some of the investments that we've been making in the transformation office and things like that. But we're very much aligning to the future growth and prioritizing the spend in order to achieve that.

Operator: Your next question comes from Siraj Ahmed with Citi.

Siraj Ahmed: A few questions.

Just first one on the work in hand and conversion to revenue. Any change in how we should think about it? Last year, obviously, there's second half skewed because of the major customers. How are the discussions tracking this year? It's up 15% year-on-year, that's pretty good. But just keen to understand how we should think about conversion.

Mark Brayan: Yes, I think -- hey, Siraj, I would -- on a full year basis side, I wouldn't think any different to prior years.

There's always subtle differences year-on-year, but in general, I think about it in a similar way.

Siraj Ahmed: Okay. And second thing, you're going 6-year -- a 5-year target of more than doubling revenue, sort of a -- so I was just trying to understand, when you think about the growth trajectory over that 5 years, are you expecting growth to be faster in the first few years, slower later? Or is it just consistent revenue growth?

Mark Brayan: We've obviously got a bunch of models that get us to those figures, and there's different trajectories. I think the shape of the curve will depend upon the non-Global customers more so than the Globals. As you can see with the China chart, that had a slower start and then got some momentum.

So I'd expect a little bit of a build that -- probably more so from the non-Global customers than the Globals. Globals will be a little steadier.

Siraj Ahmed: Sure. And actually, just on the Globals. Again, the target sort of implies, I think, Global is growing at 7% year-on-year.

It grew 3% this year. So the -- how -- just keen to understand how you're thinking about that? And what gives you confidence that growth will be at -- around those high single digits?

Mark Brayan: Yes. We -- there are 5 customers in that cohort. They all have their own characteristics and challenges. And if we look at some of the things they faced last year versus this year, we're pretty optimistic given the book of projects we've got with them compared to where we were at this point last year.

Siraj Ahmed: Got it. That's helpful. Last one, just maybe for Kevin. Just regarding the margin comment. Just confirming that we shouldn't be expecting margin to go backwards next year.

That's a thing. And would you be able to quantify the level of investment? I mean given the project office, but you also mean to take costs out. Just keen to understand the level of investment that you're thinking about.

Kevin Levine: Yes. So we keep the guidance of up to 10% in any one year.

In terms of for this next year, probably a range of 8% to 9% is how we're thinking about our investment. But, as to Mark's point, the key thing for us is to build the foundation and the pathway for that future growth, which means we may accelerate or decelerate at any point in time in order to do that. But within that overall framework of the 10%, probably around 8% to 9% for the '22 in terms of how we're seeing it this year. And then in terms of the margins, obviously, we have some which we've called out. We've obviously got some -- something that's going on, particularly just in '22 when you compare to '21 unless they're like cost of the transformation office.

This will give us a positive benefit probably from '23. We've also -- when you look at the share-based payment costs, '22 versus was '21, '21 had an adjustment in the downward adjustment. So obviously, expect normalized, share-based payment costs coming through 2022 and note the comparison with '21, which is going to impact things as well. So once again, these are -- these aren't anything that is really impacting us or banning us from our long-term positioning, but just letting the market know in terms of how we think we're seeing things just in the immediate short term.

Siraj Ahmed: Okay.

So the earnings skew is more about those costs coming in the first half rather than the second half compared to last year. That's why the earnings skews -- more skew to the second half.

Kevin Levine: Yes, yes. There's a couple of comparative pressures that impact that I've called out. And obviously, that impacts that skew, as we've called out.

Operator: Your next question comes from Michael Aspinall with Jefferies.

Michael Aspinall: I might just start on the order book. I mean Siraj kind of touched on this, but it's plus 15% year-on-year so far. Would you expect the first half or the full year to reflect that at the top line based on what you're seeing so far?

Mark Brayan: Michael, so we do expect a half-on-half revenue skew, as we call out there if you look at the year's prior to 2020 as a guide.

Michael Aspinall: Yes.

Just thinking about the 15%, though, growth you're seeing kind of in the first 2 months, call it, of the year, would you expect at least the first half to, say, reflect that given it's kind of revenue already received for the first 2 months of the year?

Mark Brayan: So this certainly includes revenue for January, to be clear. And I'm not sure I understand the question, given we're calling out that half-on-half skew. I think that's your answer there.

Kevin Levine: Yes. Actually, Michael, just to take away.

I mean basically, the methodology here is that this is -- these orders are for the whole year. So you can't really draw any necessary comparisons as to a certain period. It's just -- most represented over the full year period is the first data point and then take the other one where we talk about the skew. So that -- those data points will help you with that.

Michael Aspinall: Okay.

So the full year should be 15% growth, but the first half might not be?

Kevin Levine: Yes. You've got the data points that help you with that skew.

Michael Aspinall: And it sounds like you saw a return to growth in the large legacy projects in the second half. Just confirming that, that's right. And what drove the customers to return to investing in that area?

Mark Brayan: So this is -- I don't know if legacy programs is the right way to put it.

So there are a number of large programs that we support with our customers, and there's inbuilt seasonality in some of those programs. For example, those that deal with advertising tend to pick up in the second half because of the retail season. So I think that was sort of as the season would have it. Really, what we saw was newer projects picking up in the second half that were consistent with customer strategy to build non-ad-related projects, for example. So it's a blend of both.

But I think on the large core programs that we support, there's always a degree of seasonality there.

Michael Aspinall: Okay. And you showed us some numbers at the half. I think it was projects started prior to calendar year '21 and projects started in calendar year '21. I mean if we just have that in the back of our minds, it sounds like we should expect those projects that started this calendar year have contributed more than what they had in the first half on a proportional basis.

Mark Brayan: Yes, yes. So we get new project starts over time, and sometimes they take a little while to ramp up, sometimes they ramp up very quickly. And yes, we had some projects start in the first half that ramped up into the second and then some that started fresh in the second half that delivered material revenue as well.

Michael Aspinall: Okay. Yes.

That's interesting. Last one for me, just kind of a follow-up on Siraj's question. You mentioned that the target in '26 is for 1/3 of revenue to be from non-Global customers, which kind of implies that 2/3 will be from Global customers or revenue going from $340 million-odd last year to $600 million in '26. So I'm just interested in how much consultation you've had with your Global customers kind of on that medium-term perspective.

Mark Brayan: So we obviously work very closely with our customers on their MAUs, and that gives us a view of the demand for data.

It's a combination of existing projects and some assumptions for new projects. So yes, there's some customer input there, but there's also -- we're backing our ability to find new projects and continue to deliver into existing ones.

Michael Aspinall: Okay. But there is some input there from customers in terms of that kind of 3-, 4-, 5-year outlook?

Mark Brayan: Yes. The customers just want to make sure that we have -- in addition to the quality of data that we provide for them, the customers are always keen to know that we've got sort of new ideas to bring to them.

So we do have forward-looking conversations that inevitably go to some discussion around capacity and volume. Having said that, I don't lay out specifically what their requirements are because their businesses are very dynamic. And they rely on us to be dynamic and agile to support them as their needs arise.

Operator: Your next question comes from Xavier Waterstone with QuayStreet Asset Management.

Xavier Waterstone: Just a couple from me.

So first one, positive to note that there's a clear disclosure of the development capitalization and amortization on Slides 33 and 38. I'm just curious, though, given the enduring nature of a lot of these costs, especially the development spend, whether EBIT margin should be the focus rather than EBITDA as a better reflection of economic profitability.

Kevin Levine: Yes. I'll take this one, Mark. Yes.

We certainly consider this. What we do as well is we look at ourselves compared to our peers as well, and we look at those levels of development as a percentage of revenue. And essentially, what we've derived from that research and comparison is that a lot of companies -- similar top-tier companies and other companies all disclose on an EBITDA basis, even though they have higher levels of development as a percentage of revenue. So at this point in time, we would consider ourselves an outlier just going to the EBIT -- relative to everyone else that's on the EBITDA, but it is something that we consider. But given our levels of spend lower than a lot of others, we're reflecting and we're looking to report at this level for the time being.

Xavier Waterstone: All right. Cool. And the second one was, I guess, I understand your -- the comment about no longer providing short-term guidance. It's understandable given the volatility of some of these revenue and contracts. So second question, I guess, is it looks like given the surprise magnitude of today's share price reaction probably reflects a bit of a lack of confidence in timeliness and continuous disclosure.

Can the market expect anything in terms of more frequent or substantial just kind of business operational updates to help address this trust and information gap?

Mark Brayan: I'm sorry, I was off-line talking to the moderator. Could you repeat the question?

Xavier Waterstone: The question was, I guess, I understand your comment about no longer providing short-term guidance, given the surprise magnitude of today's share price movement and some of that being a lack of confidence in timeliness and continuous disclosure, whether the market can expect anything in terms of a more frequent or substantial business updates to help address this information gap.

Mark Brayan: So we've always provided information as we've been required to do so, and we'll continue with that policy. And if there's anything that we feel is in the interest of the shareholders, then we'll provide that.

Operator: Ladies and gentlemen, we have five minutes remaining with us.

Your next question comes from Garry Sherriff with RBC.

Garry Sherriff: Mark and Kevin, a couple of questions. First, the FY '26 goals, they're looking ambitious. Does that include any acquisitions for those goals, if you could clarify for us?

Mark Brayan: Not material acquisitions, Garry.

Garry Sherriff: Understood.

And thank you for that graph on the competitive landscape, it's very helpful. You flagged after Lionbridge's scale, the next largest competitor. Who is that? And the follow-up question is do you find any of the assets that you've flagged or companies you flagged, does that have anything attractive that you think could be beneficial under the Appen umbrella?

Mark Brayan: So the fourth one is sort of a blend of information that we've picked up in the market. There are some competitors in the tens of millions of revenue in the sort of the $30 million to $50 million range. So it's just a sort of a blend of people that the point of the chart is to show that Scale has put out a figure about their revenue.

The others can pick up things from time to time, but they're going to be around that size. So it's not a specific company. It's just sort of representative of where the fourth place competitor is. In terms of other companies on the slide and do they have things that would be beneficial, look, we look all the time, Garry, at companies in the space and whether or not we could combine with them, bringing them under the Appen umbrella, as you say. Oftentimes, though, they're fairly early and loftily valued, to be candid.

So we just have to be very careful and strategic about some of the things that we look at. But I'd also say that most of the things -- most of the companies on the chart have -- sort of tech forward. And when we look at the technology and the capabilities we've got, there's a fair chance we can build it ourselves at a better return to shareholders than buying them.

Garry Sherriff: Understood. Just going back to the ad-related revenue being, call it, 25% of group revenue.

Have there been any insights in terms of your discussions with Facebook in terms of volumes for calendar year '22 versus last year, given that they're clearly a big contributor to your ad-related revenue?

Mark Brayan: Yes. So the order book number is obviously inclusive of a lot of work with our major customer, and I think that's probably the best answer I can give you there, Garry. You can draw some inference from the size of the order book number.

Garry Sherriff: Yes. No trouble.

Last one on China. Maybe let us know who some of those customers are. And if you could provide us any detail on the margins, gross margin or EBITDA margin, that would be beneficial.

Mark Brayan: Yes. So many of our customers, as you know, are building future-facing products, and they're not keen to -- for us to talk about who they are.

We've managed to get permission for a number of logos, as you can see throughout the deck. None from China, though. But I can tell you that all of the China tech giants are our customers. I can also tell you that the major mobile companies are our customers as are the major autonomous vehicle companies. So you can probably find a bunch of them pretty quickly just with some searching.

In terms of gross margins and EBITDA margins, we haven't disclosed those, but I can say that we're still investing into China, and I can also say that the gross margins are improving. That said, the focus in China is very firmly on growth, and you can see that, that's paying dividends for us.

Garry Sherriff: And so when you talk about some of those large Chinese tech giants, is there any reason why then you'd classify China as being non-Global in terms of when we segment them as customers? I just noticed that in terms of that non-Global customer base, you've flagged China as being non-Global, but I would have thought some of those Chinese tech giants would be.

Mark Brayan: Yes. I guess it's in the name.

I mean we -- our Global customer base is the 5 big U.S. giants because that's where we derive the bulk of our revenue, and everything outside of that is non-Global. Some of our customers outside of that are sizable, but just by our definition, the Global is those 5 U.S. tech giants. I assume that's it, Garry?

Garry Sherriff: Yes.

Sorry. Yes. Fine for me. I thought the moderator had cut me off, but yes, that's great.

Operator: Ladies and gentlemen, that was the last question for today.

I'll now hand back to Mr. Brayan for closing remarks.

Mark Brayan: Yes. Thank you very much, and thank you, everybody, for joining the call today. As I said earlier, we are pleased with our results this year and firmly focused on those 5-year goals to reiterate or to at least double our '21 revenue by 2026 to have 1/3 of our revenue from our non-Global customers.

And as per the response to Garry there, the Globals are the big 5 U.S. tech giants, and we have 1/3 of our revenue from outside of that. And finally, to continue to be a profitable company with an EBITDA margin target of 20%. So thank you once again. And I'm looking forward to me and you in one-on-one meetings that we may have.

Thanks very much.

Operator: Thank you. That does conclude our conference for today. Thank you for participating. You may now disconnect.