I'm Ian, the CEO. I'm joined today by Sandy Davies, our CFO. Also in the room today are Julie Southern, our Chair Designate, Jane Hyde, our Company Secretary, and Andrew Pickup, who heads up IR and strategy for us. Looking forward to talking to you today. We're going to follow a similar format to previous presentations. I'll start off with an overview. I'll hand over to Sandy to go through a financial review. I'll come back with a deeper dive into our strategic progress and our divisions. We'll touch on current trading and outlook, and then we'll go to questions. I think you're all pretty familiar with the makeup of the group, but just as a reminder, we are a unique, world-leading provider of technology-enabled language, content, and IP services.
We are very well diversified, with many leading market positions. We operate in a large and attractive market. In many ways, we are defined by the very high quality clients who choose to work with us over very long tenured relationships. We have a unique delivery platform, combining mature technologies with very experienced people and truly global reach, covering some 32,000 in-house and freelance linguists, and thousands of colleagues supporting them. We offer 24/7 delivery on a global basis, which is crucial for the sorts of clients that we serve. The business has always been run prudently, with strong cash generation, low levels of debt, maintaining the flexibility to invest both organically and in bolt-on acquisitions.
What we do is help our clients to grow by ensuring that they are understood anywhere and in any language. We do that by helping them to create, to collect, to transform, to analyze, to launch, and manage content. That content can come in any form: text, image, audio, video. Historically, we would have said that we don't spend as much time creating content as we do in the other stages that you see on this chart. Increasingly, we find that we are helping clients to help to originate content, to facilitate the development of their AI platforms, to help them with truly transforming content to be relevant in different cultural contexts. Increasingly, we are a capable organization across this content value chain. Now, I think there are probably two key exam questions we need to address today.
One is the performance on growth, why we haven't grown in the first half of the year, and why we believe that we will start to grow again in the future. Secondly, the question of AI: Is it a net risk or a net opportunity for the group? I want to hit those two topics head-on. It's been a tough 18 months. Very challenging macroeconomic background. I don't need to sort of spend time explaining the reasons for that. What that has meant for our businesses is that in Language Services, we've seen spend reductions across many of our clients in Regulated Industries. We've seen legislative changes and regulatory bottlenecks that have hindered our ability to grow in life sciences.
In IP Services, as I think you all know, the arrival of the Unitary Patent, which actually went live on the first of June, has also presented that division with headwinds. We think many of those are transitory, and that we will work our way through them. In the immediate period, what that has looked like for us was slower client decision-making, lower levels of activity, and a greater focus on cost through competitive tendering processes and through quite challenging pricing discussions in some areas. Now, of course, we haven't just sat there and watched. We've done a number of things to mitigate the impact of those trends. We've been driving an initiative on sales effectiveness, which we started in our IP Services division about a year ago, and we've recently extended that across all of our divisions.
We've been working hard on efficiency, and we are confirming today GBP 10 million of cost reductions, which will impact in this financial year, and a further GBP 25 million of cost actions that we are starting to take right now, which will deliver through the course of FY 2024. We continue to leverage technology to drive efficiencies within our operation, especially through our language platform, the LXD, and I'll touch more on that in a second. We are continuing at full speed with the transformation programs that we announced last year, which were all about delivering a more efficient operating platform for the group. If there's a change there, what we've been doing within that overall transformation program is focusing on those aspects of those projects which can deliver earlier access to margin benefits.
We'll talk a bit more about those programs as we go through. In terms of AI, we talked extensively about this at our Capital Markets Day last year, and I think we were very open that our industry has faced a long-term downward trend in cost per word as a result of leveraging technology. That trend goes back many years from the advent of translation management systems, translation memories, and more recently, neural machine translation and AI. That is not a new trend. Our strategy was very conscious of that when we launched it last year, and that's why we put so much emphasis behind our language platform and behind using AI within that platform.
We also talked about how AI is a source of growth, and today we're very clear that the net impact of AI on our business is a net opportunity. Why is that? Well, we believe we are really well positioned to help our clients through the whole AI journey, from exploring how AI can help them make their businesses more productive, to helping them build AI platforms, to helping them deploy solutions for language and content management through the software products that we acquired through the SDL merger. If you care to look today on our website, we've refreshed it recently, with a number of examples of the sorts of things that we're able to do here. rws.com/ai, you'll find some case studies and further details.
To give some color to this, you know, we are already completing projects, helping clients to look at different AI platforms and how they can deploy them within their businesses, whether that be for content management or other purposes. For many years, we've been helping clients to develop large language models, voice assistants, et cetera, and I'll talk more about how we're building on that as we go through this presentation. In many ways, AI is built into RWS, and we embrace the opportunity to work on it further. Before I hand over to Sandy, let me just touch on the headline results. Revenues of GBP 366 million in the period.
Gross margin, broadly flat, which I think does demonstrate that our model is working, that we're able to match through the LXD supply to varying levels of demand across different language pairs. PBT reflects, you know, trading conditions, but also the investments we've been making in some of those growth initiatives. In terms of CapEx, 6%, exactly in line with guidance, reflecting the impact of our transformation programs ramping up. Interim dividend of 2.4 pence per share, in line with our progressive dividend policy. Cash conversion has moved year-on-year, and that is almost entirely due to the increased spend on our transformation projects. Return on capital, exactly within the range of guidance that we've given previously. Overall, we think these results show continued strategic progress, and we are maintaining the guidance we've given previously.
I'll now hand over to Sandy, and then I'll be back shortly. Thanks.
Which button is it? Thank you. Thank you, Ian. Now turning to the numbers in more detail. Revenue was up 2.5% year-on-year. This includes a small benefit relating to the Fonto acquisition, which took place in March 2022. Organic growth, excluding Fonto, is 1.6%, whilst we're reporting a revenue decline at constant currency of 6.8%. I'll provide more details, including the divisional breakdown, on the next slide. Gross margin, as Ian's mentioned, was broadly flat at 45.7%, and this reflects the softer activity levels seen amongst a number of clients in Language Services in Regulated Industries, which has impacted mix negatively, and the more competitive dynamics in other end markets, largely offset by pricing and further efficiencies being driven through the LXD.
Admin expenses, as a percent of gross profit, has increased by 410 basis points, partially due to the planned front loading of investments to accelerate growth, and partially due to the lower revenue, and thus, gross profit realized in the first half. We've recognized GBP 8.6 million of FX gains in the period, largely relating to forward FX contracts, which has helped to offset the FX impact experienced across our cost base. Please be aware that this gain has been recorded in the unallocated corporate segment, which means the divisional adjusted operating profits reflect the true underlying cost of running the business, whilst the group as a whole has benefited from some offset. The net result is profit before tax of GBP 54.4 million, running at 14.9% revenue.
It should be remembered that for RWS, profit as a % of revenue, is typically lower in the first half than second half, and with the cost actions being taken, we expect to see the full year in line with current market expectations. There are a number of adjusting items which are detailed in the accounts. These include acquisition costs of GBP one and a half million, largely relating to Fonto contingent consideration, integration and transformation costs of GBP three and a half million, amortization of acquired intangibles of GBP 19 million, relating to the past acquisitions of SDL and Moravia, and share-based payments of just over GBP 1 million. Reported tax expense for half year 2023 was GBP 7.8 million, an effective tax rate of 27.2%.
This 100 bps reduction compared to last year, is primarily due to a U.S. R&D tax benefit recorded this year, that wasn't recognized in the comparative period. Once we take an account of adjusting items, such as exceptionals and amortization of acquired intangibles, the adjusted effective tax rate is 24.4%. The increase from the comparative 23.7% is primarily due to the increase in U.K. tax rates from 19% to 25%, giving rise to a 22% U.K. blended rate for the full year 2023. This is also consistent with our earlier guidance and expectations. Adjusted EPS came in at 10.6 pence per share, down 11% on the prior year.
The board has approved an interim dividend of 2.4 pence, which is a 7% increase on last year's interim dividend, supporting the capital allocation policy of a progressive dividend. Turning to revenue in more detail. This slide shows the drivers, the revenue variance from first half last year to the first half this year. Language Services has reported growth of 1.4%, while at constant currency, this is 8% below the period last year. As previously highlighted in the half year trading statement, we continue to see reduced activities from some of our clients as they have adapted their priorities to changes in their own markets.... We continue to win new business across the division, and we have seen some pleasing results in our growth areas of eLearning and data services.
Ian will cover this in more detail later in the presentation. In Regulated Industries, reported growth was -1%, while at constant currency, this equates to a decline of 9%. This retraction is due to the previously communicated loss of a major CRO client and softer trading conditions within a number of life sciences clients, partially offset by good growth seen in both linguistic validation and the finance and legal services segment. On a reported basis, LCT has grown 13%. The organic reported growth of the division, excluding Fonto, was 7%. Once again, we achieved excellent growth in the proportion of SaaS revenues, with 29% of new revenues in the division being SaaS. This dynamic suppressed the constant currency revenue growth of -2%. Although, of course, for the longer term, SaaS revenue is more valuable given its recurring revenue.
In IP Services, trading remains on track for the year, with a positive impact from our sales improvement initiative. The decline in revenues of 6% on a constant currency basis was in line with our expectations, as clients continued to delay the granting of patent applications in order to benefit from protection under the Unitary Patent. As you can see, the FX impact in the period has had circa 8% impact on our top-line results in the period, as the dollar strengthened against the pound. Ian will discuss business performance across all the divisions in more detail later. Turning to the cash bridge, operating cash flow before working capital movements came in at GBP 60.2 million. Cash conversion in the first half was 85%, in line with expectations, and once again, demonstrating the strong cash generation of the group.
Significant outflows relate to dividends, tax, the final purchase consideration for Iconic, and the increased investments as outlined at the CMD, in line with our strategy for investment. Please note that our definition of cash conversion is free cash flows before exceptional cash flows, divided by adjusted net income. Any additional payments made relating to the newly announced restructuring, or indeed any acquisitions that may arise in the second half, will impact our net cash balance, but will not impact the cash conversion metric, as this focuses purely on the underlying business activities and not those considered one-off in nature. Total cash at the end of March 2023 was GBP 76.3 million, and net cash after borrowings was GBP 57.8 million.
Even after the incremental investments and the final dividend paid, this is a net increase of GBP 19.6 million from the net cash position at half year 2022. For information, as is typical, lease liabilities of GBP 38 and a half million are not included in these figures. As a reminder, we currently have a committed facility of $220 million, of which $194 million remain undrawn at the 31st of March. Moving to the balance sheet, which remains strong, and when comparing to September 2022, there's only limited change other than the FX revaluation. Some points to note are the goodwill decreased by GBP 39 million due to the movements in the dollar relative to the pound.
The closing rate of the US dollar at 30th September was 1.117, whereas at March it was 1.239. Intangible assets also decreased by GBP 25 million. The additions of just over GBP 20 million being more than offset by amortization of GBP 27 million, and the FX revaluation impact of GBP 19 million. Net working capital decreased in the first half, primarily driven by lower revenues and FX revaluation. Now to our key KPIs, which Ian introduced in the year-end presentation in December, and calling out a few of the key ones we've not already touched on in the financials. As you can see, both our NPS and our repeat revenue metrics remain high and fairly stable, reinforcing the strong client relationships and satisfaction that RWS maintains.
The cumulative incremental revenue from our defined growth initiatives now stands at GBP 11 million, with continued progress being seen in linguistic validation, eLearning, and data services. We continue to see a shift in our license models to SaaS, linked to the increased R&D investments in our products. SaaS revenues grew 29% over the half year and now represent 33% of the site license revenue reported in LCT, compared to 28% this time last year. Development spend of 12.5% is in line with our strategy to invest behind our organic growth and continue to meet our clients' evolving needs through the right range of solutions with increasing AI-based functionality. CapEx, 6.2% of revenue, is also in line with the investment profile indicated last year at CMD.
The spend year to date is a combination of the continued investments being made in our software development, as well as the transformation programs, of which I will provide a more detailed update in a moment. Finally, from a people perspective, you can see our current level of voluntary attrition has reduced since September to 12.5%, and the percentage of women in senior leadership positions has increased, bounced back a little bit to 39%. A very brief update on our transformation programs. The main takeaway here is that we are progressing in line with expectations. We're pleased to report that the first program, Highlander, which was the global transition to a single collaboration platform, has been completed on time and on budget, and I'm pleased to say that we're now all benefiting from easier collaboration and communication.
The LXD program continues to progress well, with a number of staged migrations, both of tools and services, taking place until the end of 2025. As you can read on the slide, we're making progress migrating volumes into the LXD for life sciences and Language Services clients. The planning for the migration for IP Services is well underway. The IP Services program focusing on delivering efficiencies, both internally and for our clients, is due to be delivered towards the back end of 2024. As previously communicated, the finance and HR programs are focused on simplifying and optimizing the group-wide operating models and platforms. The new target operating models were announced in April, the associated restructuring program has been initiated. Delivery of the new platforms are scheduled to take place in a number of releases throughout 2024. Now a reminder of our capital allocation policy.
On top of our business as usual maintenance, we look to invest to grow. We continue with our progressive dividend policy, and we look to acquire for further growth, and this creates a virtuous circle for building future value. You have heard that the board has approved an interim dividend of GBP 2.4 pence, reflecting the 7% increase over prior year, and Ian will talk to M&A in a few slides' time. Finally, from me, due to the group's strong cash generation and balance sheet and current share price, the board has decided to initiate a share buyback program of up to GBP 50 million. The group has substantial headroom under its existing facilities, even after taking into consideration this proposed buyback, payment of dividends in line with policy, and the capital to fund its organic growth and acquisition strategy. With that, I'll hand back to Ian.
Thank you, Sandy. Right, a quick reminder of the structure of the group, which basically hasn't changed. We have four operating divisions. I'm going to go through each of them in turn. You can see there the revenue shifts. Not much to say there. Slightly more from a Language & Content Technology. Just a reminder, our Language Experience Delivery platform is our language platform, and we are in the process, as Sandy has said, of migrating more work from the services divisions into that platform. That's where we combine the use of some 1,700 in-house linguists with just under 30,000 freelancers and the use of our own translation management and AI tools to drive productivity. We think that platform is unique in our industry. Going through the divisions one by one, starting first with Language Services.
At a reported level, as Sandy said, revenues were up 1%, but on an organic, constant currency basis, down 8%. As we've already commented, client retention and satisfaction levels remain very high, but we have seen softer activity across a range of verticals. We've also seen, in this division, in particular, more competitive procurement-driven tender processes, particularly in the technology sector. Having said that, we're very pleased to report that we've had a very successful renewal with one of our largest clients, a 3-year renewal there. We've also seen some positive outcomes since our last trading statement on a couple of important tender processes. We're now feeling much calmer about that than we were a few weeks ago, although you can never be completely calm.
We've also seen very encouraging progress, as Sandy's already mentioned briefly, with data services, which I'll talk more about in a second, and also with our eLearning initiative. I think with data services in particular, we've won a couple of really quite exciting new pieces of work with some of our existing clients, which are relatively small in their first packages of work, but are a foot in the door for what we hope will be further growth as we move forward. We are seeing significant investment amongst our existing client base in the areas of developing AI and large language models. Operating profits down. Here, gross margin was down slightly, reflecting our ability to balance, you know, with the LXD supporting increasing volumes in this division.
Overheads, you know, drove a big chunk of the deterioration here, as we weren't able to fully recover inflation, and we've also been making investments in those growth initiatives that I've mentioned. If I turn now to AI, and a bit more of a deep dive into data services, which was one of the initiatives we talked about last year. What's the need here? Well, organizations are increasingly deploying AI to better engage with their customers and users. That's social media, voice assistance, face identification, predictive outcomes, et cetera. Those AI models require huge quantities of data to learn and improve quality, in every form that data can come. Those successful models require accurate data. What's our view of our position here? Well, this is a large and growing market.
We estimate about GBP 2 billion today. We have long-standing experience in this segment. We've been working with a number of the innovators in this space since about 2016, providing them with data services. We have an established capability, a very large community that support the work of gathering and annotating the data that goes into this activity. It's a familiar route to market because a lot of the people who are innovating this space are existing clients of ours. We relaunched this proposition under the brand TrainAI back in February, in line with the plans we set out last year. The focus here is on data that is sourced ethically, with a focus on accuracy, avoiding bias, and ensuring inclusivity, with privacy and security remaining central to the approach.
As we've already mentioned, we've made some encouraging progress with this initiative since we announced it in February. We feel that launch was particularly well-timed, given the investments we're all seeing in this space. I turn now to Regulated Industries, at a reported level, -1%, on an organic constant currency level, -9%. Here, as we've already previously talked about extensively, we chose to stop working with a large CRO client about, you know, this time last year. That has been a drag on growth over that 12-month period, but obviously, that effect starts to unwind now, we've reached the anniversary of that.
That has been offset by good performance in linguistic validation, where we've seen very high single-digit growth, and also in our finance and legal segment, where, again, we've seen high single-digit growth. In that case, assisted by the PRIIPs regulations, which have caused a number of our clients to have to do new disclosures in multiple languages. We have seen, though, lower client activity than we expected this time last year in the life sciences segment, and I'll come to another slide to explain some of the reasons for that in just a second.
Here, gross margin was a bit more affected by the loss of that CRO client and the headwinds we've seen on growth. We had similar effect here in terms of the impact of inflation and investments on overheads, which both contributed to the decline in operating profit. If I turn now to life sciences, this chart aims to try and explain why we felt faced some short-term headwinds in the life sciences portion of this division. Along the top there, you see the R&D process for bringing new drugs to market, the four stages there. Along the middle, you see how our revenues are typically aligned to those four stages.
What you see is we tend to do more work at the regulatory and launch stages of this process than we do in the clinical and preclinical stages. The one big exception to that is linguistic validation, which was the initiative we have been doing for many years that we again decided to invest more in last year. What we're seeing in the market is very buoyant demand for clinical-related work. We think the market is up about 9% over the period 2021-2022 in that portion, which is why linguistic validation, we think, is doing very well. At the same time, we've seen some impacts, I think, from the Inflation Reduction Act in the U.S., which had some, you know, material changes on drug pricing, contained within it, which has caused, we think, clients to reevaluate their product, launch plans.
We've also seen some bottlenecks in the regulatory process, in particular at the FDA. We do think those things will reverse, the fact that we're seeing very strong activity at the clinical stage is a very good indicator that that work will come through to the regulatory and launch phases subsequently. Little bit hard to predict when, we remain very confident. I think the fact that we have very strong relationships with many of the large players in this space, and the retention levels remain very strong, means that we do believe we are well positioned to benefit from that work as the market recovers. Turning to language and content technology. Here, revenue's down slightly on an organic constant currency basis, up 13% on a reported basis. Generally, we're pleased with progress here.
SaaS revenues continuing to grow very strongly, ahead of the plans that we had for them. That does, as you all know, I think, bring the headline reported growth rate down, but it does point to longer-term value creation. We've seen new logo wins across a range of sectors, including defense, government, infrastructure, and software and verticals. We've seen an increase in the bookings rate for Language Weaver, our neural machine translation platform, in the latter stages of the first half, and we have new releases of our Tridion content management platform planned for the second half, which we expect to support growth in the months ahead. Margin erosion here, similar themes.
Little bit impacted by the SaaS mix, in terms of the gross margin, and then, we are also making investments in this division, as you all know. Last, but never least, IP Services, where the group originally started. Here, as can be said, very much in line with expectations, and seeing the impact of the Unitary Patent. Organic revenue down 6%, flat on a reported basis. The Unitary Patent was finally launched on the 1st of June, and we're actually quite relieved to be able to say that. We do expect to see a backlog of work coming through the business, as patent grants move through the process now.
We've seen in parallel, actually encouraging growth in the WorldFile segment of IP Services, and we are seeing the impacts of the sales improvement initiative that we started last summer, as well as the impact of some of the leadership changes that we've made. The team have been very effective in managing cost mitigation against the declining revenue line here, very much sort of in line with expectations. Turning back to AI, just to sort of pull it all together, I think a few key points to emphasize. We have well-established products in this space. Language Weaver, one of the original neural machine translation products, dating back to the University of California some 20 years ago and acquired by SDL, some 13 years ago.
Data services, we've been in that business since 2016, really innovating alongside some of the pioneers in the space, and we are increasingly adding AI functionality to our other technology platforms. For example, we now have a semantic AI search capability within the Tridion content management platform. We have real expertise in the space. We have over 40 AI-related patents. We have over 100 peer-reviewed AI papers. As we've said already, 60% of the words going through our LXD are translated by Language Weaver first. And we use a Content Analyzer tool within Language Weaver, which allows us to scan content, understand what it's about, and that helps us direct the work to the most relevant linguists to work on it. All of our products are moving forward with developments in this space.
Language Weaver continues to evolve its quality and the range of languages. We have an enhanced data services proposition, which we've already talked about. We've already implemented a connector within Trados Studio, the world's leading translation productivity tool, which allows linguists to connect to GPT-4 if they wish to. I would emphasize, we do not allow our linguists to use that capability due to security concerns, but our philosophy is to allow consumers of Trados Studio, who buy that product, to connect to products like GPT-4 if they wish to, and they feel it's secure for the content that they are using. We continue to look at how we can incorporate large language models into our localization process. There will be secure ways to do that.
To do that, we need a hosting environment, we need to partner with a provider of a large language model, we are currently evaluating several, then we need data to train those engines. On data, we have no shortage of capability, both to originate data from our linguists and our community of data AI engineers. The message here is we are embracing the opportunities, we are mindful of the risks, we really do think we're well-placed to develop with this market. Touching on M&A briefly, nothing really has changed in terms of our focus areas or our screening criteria.
I think the key message today is we are seeing a much better pipeline of opportunities than we saw 12 months ago, we are at an advanced stage with several bolt-on acquisitions, we'd obviously hope to complete one or two of those over the coming months. On ESG, just really to continue to emphasize that this is important to our investors, it's important to the sorts of clients that we do business with, it's important to all of our people, we continue to make progress on the commitments that we've made against ES and G. On governance, I'd just point out, we've made particular focus in the last few months on health and safety, on whistleblowing through a speak-up program. I should just touch on information security.
As I think you're all aware, we had a cyber incident a couple of months ago, and we were pleased that the Information Commissioner here in the UK has confirmed they'll be taking no further action off the back of that incident. Final slide before we go to questions. In terms of outlook, as we've talked about already, we expect to deliver an improved second half due to the benefit of some of the client wins that we've talked about and some projects that were delayed in H1. We expect to see further incremental revenues from our growth initiatives. We expect to see that release of backlog of work in IP Services now that the Unitary patent has gone live.
We are making good progress with our infrastructure programs as expected, with the first major program completed. We have a strong belief around AI being both a driver of efficiency and growth. Our outlook, sorry, remains in line with market expectations, with the second half weighting to performance. We have seen a confluence of headwinds impacting the group's ability to grow in the short term. However, the encouraging outcomes we've seen on some tender processes are giving us confidence around our guidance moving forward. We're taking cost actions to ensure our cost base remains aligned with our top line to support our margins. We remain confident in the long-term growth drivers that we talked about last year.
We've mentioned the buyback of GBP 50 million, which we expect to initiate shortly, and we continue to see exciting opportunities to invest in our business. With that, I'll end there. Thank you all, and go to questions. I think we're gonna take questions in the room first. I think there's a microphone, so if you can wait and just say who you are. James hand up first.
Morning, James Beard at Deutsche Numis. Two questions from me, please. Firstly, on the, on the sort of competitive tendering and procurement-driven processes that you sort of alluded to back in the Q2 update a few weeks ago, why do you feel that you've made good progress in sort of securing your position on those tenders? Have you had to give away a bit more on price for example, and does that have a sort of consequent impact on sort of margin expectations with those customers going forward? Then second question, thinking about the sort of data services proposition and TrainAI, can you sort of talk to the competitive landscape for that business?
Who are the sort of principal competitors that you're up against, and what sort of proposition do you see from other large language service providers in that in that arena?
Fine. Thank you. Let me maybe take that second question first. I think there, we think our focus within that space should be around language, where we have real strength and expertise. There are, as you know, one or two players who, whose focus on this data services space. We don't really see them as direct competitors in the space that we're working in, and we think we are almost certainly ahead of most of our big competitors, given the work we've been doing in this space for some time. It's an evolving picture. I think we got to it early. We spotted the opportunity in our, in our strategy work last year.
We got our TrainAI offering out, you know, really at the right time in February, and we're seeing the pipeline building. It's clearly a space that's going to evolve, but I think we've got a good start into it. In terms of the tender process, it's always difficult for me to talk about specific clients, but I think compared to when we put our trading statement out, a few weeks ago, we have signed a material, a three-year contract with one of our largest clients. That was something that we were quite focused on when we put that trading statement out, and we had a couple of other important tenders running in parallel, which both seem to be moving to a positive outcome.
They have been more price-focused, than previously, but we've also been looking at how we can deliver those services more cost effectively, to preserve margins. Whilst none of those contracts have ever given us commitments to volume, we do think we will retain meaningful shares with all of the clients involved in those tenders, and at margins that we think will be respectable. Overall, I think, you know, feeling, as I said, calmer about that piece than we were a few weeks ago. I would say that, you know, we have, you know, clients who remain under commercial pressure in their end market, so we can never say that, you know, that process ends. We did see a particular peak a few months ago, and we seem to be coming out of that.
At the same time, some of the same clients are awarding us business in the data services space, which I think is very encouraging in terms of the strength of those relationships. Calum?
Morning, guys. Calum Battersby from Berenberg here. Three questions, if I may, please. Firstly, I'm not sure if I actually understand how the gross margin was effectively held flat in the first half year-on-year. Is that all LXD? Is that partially mixed? Just any color there would be helpful. Secondly, just following up on James' question there, it'd be helpful to get a bit more color on pricing and expectations going forward. If we're now in a period of inflation being higher for longer, does that create an ongoing headwind for the business, or do you think it's a case of, in time, you'll be able to increase pricing in line with inflation? Lastly, just following up on the AI point, would you mind talking about how that has impacted the tender processes you've been going through?
Is it a case of clients talking about wanting to do more work in-house compared to how they have historically, or just generally, how would you characterize kind of the conversations with customers and how they're talking about the use of these AI engines?
Right. Why don't I take the AI and start on the pricing point, and that give Sandy a moment to get ready on the gross margin question, and hopefully, they'll all link together. I think in terms of those tender processes, AI is not really a factor. I mean, there's a separate conversation about data services opportunities, but what we're not seeing is clients saying, "We're taking work in-house and doing it somehow on generative AI." That just isn't a thing. It isn't what we do. So that, I think, you know, is important to sort of just set that straight. What we are seeing is clients more willing than they have been in the past for us to use machine translation in our production processes.
Historically, clients, for example, in the life sciences space, would have been more cautious about us using machine translation. I should emphasize, we never use machine translation unless a client is happy for us to do so. If they don't want us to, we'll just use, you know, our talented linguists. What we are seeing is, as we are able to demonstrate the quality of outcome from a machine-translated first proposition, we're seeing it is becoming more acceptable across even the more sensitive areas of our business, and I think broadly, that's fine. We're, we're happy to put more work through that model.
I think in terms of pricing, I think we are seeing, you know, there has been a long-term sort of downward trend on the cost per word element of what we do, but that is a component of the range of services that we offer. Again, our strategy has always been, for many years, you know, through RWS, to leverage technology to keep up with that downward trend as technology becomes more capable. I think if you look across the group, we've had more success with pricing, for example, in our technology division, where we've done pretty well, actually, on all software products and related services. We've done reasonably well, actually, in IP Services, where, again, slightly different competitive environment.
It's been much tougher, toughest of all in the technology area of language services, and in some portions of life sciences as well. Quite a mixed picture, but I think overall, compared to a year ago, when we, I think, we're quite honest, that the group had really not been doing very much on pricing for a number of years, we have made progress in every division on pricing, but it is a tough market out there. There's no ducking that. I think there is a need for us to continue to drive further efficiencies, which is why we've announced the further cost actions that we're taking. Sandy, do you want to pick up on GM?
Yeah. Without getting into, you know, the weeds, right? I mean, I think, you know, we have taken some pricing across the board, which, you know, overall is a positive piece. There is the divisional mix, as well as some language and client mix that plays out within the divisions. There is clearly a reasonable amount of efficiency being driven through the divisions and the LXD kind of transcript. Yeah. But.
No, that's helpful. Thank you.
Okay.
Hi. Katie Cousins, Shore Capital. Just with the life sciences revenue mix on that slide, I'm just interested how the profitability changes within each of these, those segments. Is there much difference between the early stages and the later stages?
I don't think there is actually, Sandy. I'm just trying to think.
No.
I don't think so.
It's not-
I don't think so. No, I don't think that's I, it's not such a big thing. The bigger impact on the operating profit there, the CRO client loss was quite a hit to the operating profit in the period. Again, that will sort of, you know, resolve itself over time. No, I think across those other services, not much of a difference.
Okay. If you're working more on those earlier stages, is there any guarantee that you'll also be working with those same clients on the later stages? Is it a new conversation every stage?
I mean, Look, I think we're quite, we have quite mature relationships with a number of the clients in this space, so I think it's. There's not a guarantee that you go from one stage to another, but the fact that you've been involved in an earlier stage does help. You know, with our largest client, for example, you know, we have, you know, I guess, more detailed conversations about their sort of product plans, what's at which stage. They are indicating, for example, that they expect to see a pickup in that regulatory stage over the next few months. It depends a little bit on each client's situation.
I think the more important point actually is just the fact that there is that level of activity at that clinical stage, which means one way or another, we should be reasonably well placed to pick up as things move through the pipeline.
Just one final one, please. Obviously, the long-term growth drivers are in place, and you're working well on those strategic initiatives. Thinking back to the financial guidance we got at the Capital Markets Day, and some of those headwinds that we spoke about today, are we still happy with some of those long-term targets? Perhaps, is there a risk now of that moving a year out to more 2027 and 2026?
I think we're being quite cautious about guidance beyond this next half today. Certainly, our aim is to see organic growth improving as those headwinds that we've discussed improves. Certainly, the cost actions that we're taking and the efficiency investments that we're making through the transformation programs and the increasing volumes through the LXD, should all help us get the margin moving back up in the right direction. I think we're a little bit cautious about sort of giving specific guidance at this point. We haven't done our budgets yet for next year, if you can bear with us, we'll probably come back to that later in the year.
Brilliant. Thank you.
James.
Good morning. Just on cost, could you provide a bit more detail on the GBP 35 million of targeted savings, kind of, you know, why that was the right number? Maybe the wrong question after your comment at the end, but does that support margin expansion in FY 2024? Does it sort of challenge the view that there are scale advantages in this business, noting you're taking a huge amount of cost out, whilst the plan is to kind of invest in a platform. How are those two reconcile? A separate question on, you know, if you just estimate the revenue headwind from mix of bringing cost per word down by increasing machine translation usage. Thank you.
No. I think, you know, where do those numbers come from? I guess, you know, you take a look at the volume decline that you're seeing in the year. I guess, if you start with FY23, I guess we've been, you know, updating our forecasts through the year as we've adapted to the new trading environment, and I think those were cost actions that we felt struck the right balance between, you know, preserving profitability in this year and not killing the investments that we're making in growth. I should add, with all of the cost reductions, we have preserved the investment in all of the growth initiatives that we talked about, both from an OpEx and a CapEx perspective. Obviously, with reduced volumes, we have project managers who are less busy.
We have support functions that are less busy, so we've tried to scale appropriately. I think the same approach really applies to the GBP 25 million that we've now announced for next year. I think to add to that, though, I mentioned that we were sort of, you know, thinking about the delivery on some of our transformation programs a little bit differently. For example, the LXD program, which was on the chart that Sandy presented earlier. There, we have been prioritizing parts of that program, which allow us to release cost early. For example, we had multiple supply chain teams managing freelance linguists across the group. We're now going to have a single procurement team for the whole group managing those relationships.
It's the right thing to do anyway, in terms of consistency and quality, but it also means we can take headcount out of that area where we had unnecessary duplication. What we're trying to do with the cost actions is do things that are sensible, so to absolutely prioritize customer service, and to do things that are in the direction of travel that we would have wanted to go anyway. Greater use of lower cost countries, and greater use of automation where we can as well internally. Then the transformation programs, things like the IP Services transformation, that will deliver efficiencies within IP Services, and within finance and HR. Again, as we move to a common platform, that will allow us to have shared service models that we don't have today.
All of those things help. In terms of the makeup of the cost, I think there's about 500 heads or FTEs, and I think the makeup of the GBP 25 million is roughly 75% headcount and 25% other things, which includes offices and discretionary spend, et cetera. To answer your first question, I think that we do believe that the LXD allows us to leverage our scale effectively. As we explained last year, you know, it was a fairly unique capability that we acquired with SDL. Most of our competitors do not have an LXD, they just rely 100% on freelancers.
We think it is much more sensible to centralize that activity, to balance the use of permanent employees, where we can keep them fully occupied, and then leverage freelancers to manage the peaks and troughs. We think that's worked really well through the last six months. We look every month very closely at the language mix and how we've balanced resourcing to demand. We think that is a platform that will give us scale economies. You overlay on top of that, the use of technology to drive down the amount of time we're having to spend on every word that gives us further operating leverage. That's the key thing. More standard things like finance and HR, we'll add to that. Does that answer your question fully? Yeah, great.
there was
Sorry.
The last question was just about, you know, what the mix headwind was from more machine translation this year through the LXD.
Yeah, it's interesting. I think if we looked at our usage of machine translation, it surged through the pandemic, and it sort of settled a bit around that 60% level. We still think it will go up, but I think that also does demonstrate that machine translation AI is not the full solution. You know, there are languages where it's much less capable. There are all the cultural nuances, the technical and knowledge, the brand relevant content that the machines just aren't capable of. I think as we look at generative AI, and from the trials we've been running internally, we do see a role for large language models in that solution.
For example, large language models are actually quite good at tidying up poor quality input content before we put them into machine translation. We've already trialed a connector between Language Weaver and GPT-4, where we would use GPT-4 to tidy up the source and then put it through Language Weaver, and actually, the outcomes were pretty encouraging. Clearly, we can't use that solution for client data for security reasons, but if we can, as I described earlier, then found a secure LLM solution to link to Language Weaver, then that actually has some quite powerful implications, I think, in terms of further productivity and quality enhancements. One thing I would say is, the whole noise around AI, I don't think we've ever had a more innovative business.
you can see colleagues around the business really seeing and thinking how in big ways and small, the innovations can help us deliver better service to our clients. There's a degree of excitement around it, I would say, internally. Other questions? Have we got any online or Do people can ask them directly, can they?
Yes, certainly. As a reminder, dial star one to ask questions over the telephone. We have a question from Kai Korschelt of Canaccord Genuity. Please go ahead.
Morning, Kai.
Yeah, good morning, both. glad with the royal title of Canaccord Genuity. just two questions, really, if it's okay. The first one was around the TrainAI model that you highlighted. My understanding was that, you know, when you laid out the opportunity there, that you were going after the sort of, you know, data annotation and labeling business where, you know, Appen and TELUS International, I think are already pretty established players. Just wanted to confirm whether that is the case or the comments you made before. You said, you know, you're trying to do something different. If you could perhaps just explain in simple terms, what, you know, what it is?
Sure
... you are planning to do in that area? That was the first question. The second was for Sandy. Just very quickly, if you could be so kind and repeat the non-cash FX gain within PBT. You mentioned it on the presentation, I just wasn't sure I heard the right number. Thank you very much.
Thanks, Kai. Look, in terms of the data services, broadly, yes, same space as Appen and Telus, I guess, but specifically, what we're trying to focus on in things that are very language related. If I look at the projects we've completed already, helping to train voice assistants for a couple of clients in the technology space, helping to train a voice assistant for a large automotive manufacturer. Helping to source voice and image data and annotate that correctly, helping with tools that support language accessibility. We're trying to focus in the space where we are known, where we think the barriers to entry are a little bit higher, and the value added is a little bit more.
I think the fact that we've completed projects of those nature over many years, means that we have credibility to play in that space, and a fairly mature network of or community of people who are working with us to do those projects, which often require, you know, many thousands of small bytes of content. That's basically where we're trying to play. Over to Sandy?
Yeah. Hi, Kai. It was GBP eight and a half million that I mentioned, that's moved through the P&L, and that's primarily due to the forward FX contracts that are not designated as cash flow hedges, so the fair value movement is taken in full through the P&L.
Okay. The adjusted PBT you've provided, that includes the GBP 8.5 million gain?
Yes.
Okay, great. Then in the main P&L, does that sit in is that income in cost of goods sold, or would that be part of the OpEx line? Sorry, just proper clarification. Thank you.
It's below gross margin, and it's sat in the unallocated corporate segment.
Yeah, that's great. Thank you.
Thanks, Kai. Any other questions online, or...?
Yeah, we have no further telephone questions, so I'd like to hand the call back for questions from the webcast.
Thank you. Any webcast questions? No.
Thank you. We had two questions come through on the webcast.
Oh.
They were answered in the room.
Oh, right. Okay.
Ian, I'll hand back over to you for closing remarks.
Thank you. Any more questions in the room? James?
Uh-
Sorry, we'll get you the mic.
Just to follow up on the, on the FX, I mean, if spot stays where it is today, how does that gain as we go into next year? Does some of that reverse or?
In the second half, it will move accordingly to the spot rate.
I more into the FY 2024, I guess, if the spot rate stays where they are.
So we, the-
Year-on-year.
The FX contracts are 12 months, we'll take out new contracts in September.
I know.
Yeah.
If the spot rates are where they are when you're taking out those new contracts.
Yeah.
What would that imply versus what you've taken out this year?
I'd have to get back to you on that.
Oh, we've got one more question, it looks like. Is that a webcast one?
Yeah, we've just had one come through from the webcast. Andrew Ripper from Liberum. "What return on investment hurdle do you have on acquisitions, and how did you decide upon the size of the buyback?
Sandy, you all right to take that one?
How did we decide on the size of the buyback?
Yeah, on the return on capital hurdle on our acquisitions.
Yes. Gosh, actually, no, actually, could you take that one?
Okay, sure. No, I mean, I think look, we've, we'd always look at alternative uses of capital and look at the level of accretion on acquisitions, you know, compared to other things we could do. I guess right now, we've looked at the buyback alongside the acquisitions that we're considering. We think at the valuation that we're currently trading at, the use of cash on the buyback is a very good use of capital right now. Very conscious that that sets a, you know, a comparator for any M&A that we might do. What I would just say is that we're focused very much on bolt-ons at this point, so they're not huge acquisitions.
We think they're strategically important, but we think they will compare favorably with other uses of capital.
Thank you. That's all from the webcast.
Great, thank you. Any other questions? Okay. Well, thank you all very much. We're here for a few minutes if there's any questions after we finish the call. Thank you very much for joining.