S4 Capital plc ($SFOR)

So good morning, everybody. I'm joined by Radhika and Scott and Wes, and this is our S4's 2025 call. So we'll kick off with a summary of the results from Radhika, then Scott will talk a little bit about market momentum. Wes will talk about, I guess, the topic du jour if not forever, which is AI. And then I'll come back and do a brief summary on the results, and then we'll go into Q&A. So Radhika, do you want to kick off, please?

Yes. Good morning. I will start with the financial headlines for 2025. Despite global macroeconomic pressures and ongoing client caution, strong cost and working capital management improved the operational EBITDA margin and reduced year-end net debt below the targeted range. Net revenue was GBP 673 million, down 10.8% reported and 8.4% like-for-like. Operational EBITDA was GBP 81.2 million with a margin of 12.1%, up 70 basis points year-on-year. Adjusted operating profit was GBP 74 million, and adjusted EPS was 5p versus 5.2p in 2024. Free cash flow rose to GBP 86.5 million, up GBP 48.7 million year-on-year, driven by improved treasury management and tighter working capital discipline. Year-end net debt fell to GBP 86.9 million, 1.1x operational EBITDA, below the GBP 100 million to GBP 140 million target range and well below the GBP 142.9 million at the end of 2024. Subject to shareowner approval, the Board proposes to pay a final dividend of 1.1p per share, an increase of 10% compared to the prior year. Moving to the P&L. Revenue for the year came in at GBP 754.8 million, which is down 11% on a reported basis and 8.7% like-for-like. Net revenue for the year was GBP 673 million, down 10.8% reported and 8.4% like-for-like. This reflects what has been a fragmented and volatile macroeconomic environment and the resulting caution we've seen from clients. In response to these conditions, we took a disciplined approach to cost management. Personnel and operating expenses were reduced by 11.5% on a reported basis. In the second half of the year, we launched a cost restructuring program, primarily focused on non-billable roles and further back-office efficiencies. The aim was to align our personnel cost to net revenue ratio more closely with the industry averages. We exited 2025 at 74.3% versus 76.3% in 2024. By the year-end, our total number of Monks was approximately 6,350, which is 11.5% lower than December 2024. Since June 2025 alone, headcount was reduced by 7.8%, driven by the restructuring actions. Looking across our 2 practices, Marketing Services and Technology Services. My commentary here is all on a like-for-like basis. Marketing Services, our largest practice, delivered net revenue of GBP 614 million, a 5.6% decline, reflecting an improved fourth quarter versus prior expectations. Technology Services generated GBP 59 million in net revenue, down 29.9%, driven by extended sales cycles and the anticipated reduction in revenue from a major client in the first half. From a regional standpoint, the Americas declined 5.6% and represented 80% of our total net revenue. EMEA declined 19.6% and Asia Pacific declined 13.8%, contributing 15% and 5% of the mix, respectively. Turning to operational EBITDA by practice. On a like-for-like basis, Marketing Services delivered GBP 92.6 million, an increase of 1.5%. EBITDA margin strengthened to 15.1%, up 110 basis points, reflecting decisive headcount actions and continued cost discipline. Technology Services generated GBP 8.9 million, down 19.8% year-on-year. Despite the revenue pressure, margin improved by 190 basis points to 15.1%, underlining the effectiveness of our cost control measures. Central costs increased 10.3% in 2025, mainly due to the centralization of procurement and IT functions, the annualized impact of 2024 hires and the change in treasury management, investments that position us to drive further efficiencies across the company. Moving to the next slide. We maintain a strong balance sheet throughout the year with solid liquidity and long-dated maturities. Our M&A obligations are also now largely complete. Stronger treasury management and tighter working capital discipline reduced year-end net debt to GBP 86.9 million. Leverage closed at 1.1x operational EBITDA, both below our target range of 1.5 to 2x and the 1.6x we reported at the end of 2024. Our EUR 375 million term loan, which matures in August 2028, continues to provide substantial covenant headroom against the 4.5x pro forma operational EBITDA threshold. Post year-end, subject to final settlement, the company also repurchased EUR 25.7 million of the EUR 375 million term loan at a discount further strengthening our capital position. Moving to the cash flow. We delivered a strong cash outcome for the year, underpinned by disciplined execution across the business. Working capital contributed an inflow of GBP 55.5 million, a substantial improvement from GBP 14.6 million in 2024. Capital expenditure was held at GBP 4.9 million, 35% below last year's GBP 7.5 million as our cost mitigation actions continue to take effect. Financing costs reduced meaningfully with the average effective interest rate improving to approximately 6% from 7.5% in 2024. Tax paid was lower, supported by the utilization of tax losses and our 2024 performance. The cash outflows related to restructuring and transformation totaled GBP 20.4 million, comprising GBP 15.7 million of restructuring costs and GBP 4.2 million tied to our finance transformation program. As a result, the group generated GBP 86.5 million of free cash flow in 2025, more than doubling the GBP 37.8 million delivered in the prior year. Momentum strengthened significantly in the second half, where we generated GBP 70.5 million of free cash flow, supported by sustained cost discipline and continued focus on working capital. Net debt at 31st of December 2024 was GBP 142.9 million or GBP 164.2 million at closing exchange rates. The company generated GBP 86.5 million of free cash flow during the year from which GBP 6.1 million was paid as a final dividend in the second half of 2025. M&A outflows totaled GBP 0.4 million. These movements contributed to a reducing closed net debt position of GBP 86.9 million, representing the 1.1x operational EBITDA. As net debt continues to improve, we have established clear capital allocation priorities, focused on delivering shareholder value through: number one, dividends; number two, targeted debt repurchases; and finally, three, share buybacks. Moving on to guidance for 2026. 2026 like-for-like net revenue is expected to be in line with current analyst consensus, slightly below 2025. Operational EBITDA margin is targeted to increase by at least 100 basis points. We expect the proportion of operational EBITDA in the first half of 2026 to increase compared to the first half of 2025 due to the annualized impact of the 2025 cost actions. We anticipate year-end net debt in the range of GBP 60 million to GBP 90 million, continuing our disciplined approach to balance sheet strength and target medium-term leverage of under 1x operational EBITDA. Our forecast net finance expense is GBP 20 million to GBP 22 million, supported by tighter cash management and increased interest income. The effective tax rate is expected to be 28% to 30%. With that, I will hand over to Scott for the market update. Thank you.

Thanks, Radhika. Good morning, everybody. Thank you for joining the presentation today. I'm going to cover some of the dynamics we're seeing in the wider market and then share some specifics on our client relationships before handing over to Wes for an update on our artificial intelligence progress. As you can see, digital marketing spend continues to increase at significant rates, whilst overall advertising spend is growing at around 5%, meaning that Analogue spend continues to decline. The revenues at the top platforms continue to grow in the high teens, significantly outpacing the market growth. One thing to bear in mind here is that over 80% of their revenues come from small- and medium-sized businesses, and they continue to expand their market share there. So their growth is not necessarily being driven by enterprise client spend. The technology services market continues to have lower growth compared to recent historical double-digit performance. 2025 had just over 5% growth. And whilst enterprises continue to invest in areas such as cloud and AI, the outlook for 2026 is subdued. The next slide charts the comparison between agency net revenue growth at the main public holding companies and advertising spend and GDP growth. Digital spend now represents 70% of the total spend. And as I mentioned on the previous slide, is growing at high single-digit rates, meaning Analogue is in decline. Agency growth dipped to almost 0% in 2025 and is decoupled from advertising spend and GDP growth. And one explanation for this is the continued pressure from clients to maintain their media spend, but to put pressure on what they call nonworking spend, i.e., agency spend. This is particularly the case with technology clients. On the next slide, we look at the relationship between CapEx spend and sales and marketing spend at the major tech companies. Here, we've covered Amazon, Meta and Alphabet. As you know, historically, almost half our revenues come from this sector. Prior to 2022, marketing spend at the top platforms regularly grew at 20% plus rates and is now essentially being flat since then. On the other hand, CapEx spend, particularly on AI infrastructure, has ballooned in the same period, growing over 133%. And this trend is expected to continue with the hyperscalers already announcing plans to increase their CapEx spend over 70% in 2026. The tech companies are unsurprisingly leading the charge on adopting AI in their marketing workflows and leveraging it to achieve more for the same or less. Our commercial focus remains on our clients and returning the company to growth. Overall, our scaled client relationships remain strong and resilient. We've seen some spend declines, but we've also seen growth and additional scope at half of our whoppers. Those are clients of over $20 million revenue or more, of which we now have 8. We continue to have strong exposure and partnerships with the technology sector. Despite marketing spend in this sector being under pressure, it's important for us to maintain these relationships as they inform our product and AI strategy and leadership position. 2025 saw some important wins and expanded remits, many of which from existing clients such as Amazon, T-Mobile and others. We're also seeing early progress from our focus on evolving the business model away from time and materials towards a talent and machine model based on asset-based or subscription-based approaches. In terms of growth, we continue to simplify and evolve our go-to-market messaging, and we're seeing this resonate, particularly in the areas of orchestration, real-time brands and AI. We've also invested in sales operations using AI to drive collaboration on pitches. This has resulted in a stronger new business performance and a stronger pipeline. Our AI solution, Monks.Flow, continues to develop and win awards for its leadership position and more of that from Wes later. It is at the heart of all our major pitches and opportunities. Whilst we continue to focus on the tech sector, we're also making progress with vertical-specific offers in areas such as auto and FMCG. We have a very compelling client list with some of the world's leading and most innovative companies. 8 of them are what we call whoppers, that's revenues of $20 million plus, which continues to be a differentiator for a company of our scale. Most of our direct competitors have a much more fragmented client list with smaller relationships. As you can see, we continue to be skewed towards the tech industry, albeit a slightly smaller exposure than last year, driven by declining spends in that category and the ramp-up of General Motors, which has seen our auto sector exposure increase. These are strong relationships that help us attract and retain talent to work on them. Declining spends, particularly in the technology sector have had a negative effect on the average revenue size of our top 10, 20 and 50 clients. This is primarily driven by reductions in spend rather than lost business. And with that, I'll hand over to Wes for an update on our artificial intelligence approach.

Thanks, Scott. Hi, everyone. I will run us through the Monks AI update. If we go to the next slide, I think key takeaway is that AI transformation isn't experimental work anymore. It's an increasingly mature and defined offering with a clear structure. We assess, we build, we help clients change the way they work. We're seeing ongoing enterprise adoption of our AI transformation services. We have large-scale AI efforts in place for the bulk of our top 10 clients. And for many, we are a key AI transformation partner. Perhaps more importantly, we're seeing this make us a more strategically significant partner in general. I think that plays into us getting the timing right in this space. So Martin just referred to it as the topic du jour, maybe the topic forever, clients know AI matters to marketing. It's probably the key strategic effort in every global marketing organization, but they don't necessarily have a plan on how to reorganize around that change. They're looking for credible road maps. They're looking for credible partners to help them execute that road map. That's the gap we've been filling. I think in 2026, the focus really is on extending that effort across our broader client portfolio. We focused our initial efforts on our top 10 clients, as you would be expected to do. Now that we're past what I would say is a little bit of an experimental phase, it's more scalable, which we think is quite exciting. And you just got to know from Scott as well, it's quite a key part when it comes to driving the pipeline for us, which is looking strong. If we go to the next slide, I think our ability to be a credible partner in this space comes from having done it ourselves first, we ran another one of our 25 minutes of AI sessions at South by Southwest last week. It's standing room only. People understand that we've been talking about this very consistently for quite a long time. You've heard us talk about this in the context of our own restructuring and our headcount reductions, something that hasn't been easy, maybe it's not completely done yet, but many companies are only now beginning the restructuring. We started about 18 months ago. So we believe we're further along than most. I think clients can tell the difference between partners who are figuring this out alongside them versus the one that's already been through it. And then if we go to the next slide, what I think is quite interesting for our team. I think this gets officially announced later today. We picked up an award for our own internal AI transformation. The business intelligence group has given our team an AI excellence award for the human and machine interaction, which brings me to Monks.Flow. Scott mentioned it earlier, core part of our own transformation efforts, the transformation efforts for our clients. We've put real effort and energy into developing Monks.Flow over the last few years. It's where we package both our software and services into a single offering. That's, I think, a trend and a pattern that we're seeing now play out in the broader market. I think it's quite an interesting moment in time because the cost and complexity of building software is collapsing. Code really is nearing zero as a barrier to entry, which means that, I would almost say all distinctions between software and services companies is collapsing as well. Everything is converging on jobs to be done. I think in that space, there is a case to make that services companies are actually better positioned for those jobs to be done because the experience and expertise is going to be harder to replace than interfaces. The reality is that only holds true if you manage to change your commercial model. So if we go to the next slide, Scott mentioned the launch of our subscription model. It got quite a lot of attention. Really, this is us focusing on outputs and outcomes for our clients, moving away from reverse incentives. Simply put, fixed monthly fees, annual contracts. We combine great talent with our AI workflows, our machines. We don't focus on headcount pricing. And what it means for clients is they see less friction in procurement, less friction in operations, and that subscription gets better as our deployment of AI improves. So if you have output that wants delivered 50 assets a month as the pipeline improves, as the pipeline gets smarter and it goes to 70; as a client, your costs don't go up, which I think is quite a key part of this is to make sure we have aligned incentives. From an analyst perspective, from an investor perspective, it starts looking a little more like ARR than traditional project work. We believe there's more defensible margins on this type of revenue as well. It sort of goes back to something that we have been talking about for a while as well. If we go to the next slide. How do you start using AI to decouple hours from output. It's a large driver behind our own reorganization and restructuring to make sure we can take full advantage of that opportunity. But really, it's the existential ask, right? If a task that used to take 8 hours, because of AI gets reduced to 45 minutes. If you're charging hours, it punishes that innovation. And because of that, there's a lot of perverse incentive in the current agency client relationship. We're committing to output and outcomes within a fixed fee, and we're also committing to our clients getting more faster and better as AI models improve. It's an advantage that compounds for our clients with every model release, and it's driving the ability for our clients to operate closer to real time, right, the ability to operate at a scale and speed that really allows them to be hypercompetitive in their categories. And of course, that's our main goal, help our clients win their category. We have one enterprise client signed up to this model. We have 3 more in discussion. We're planning to steadily change our revenue mix in this direction. The real takeaway is the billable hour has had a great run, of course, but it's not a long-term viable model as you see software and service collapse into one. Let's talk about some work. If we go to the next slide. So we brought a few cases here. I think important to note, this has gone from strength to strength. It's very normalized for us to use AI across pretty much every part of our organization, and there's very few pieces of work where AI has not played a role in the creation of that work. It's helping our clients save money, save time. I think more importantly, it's helping them really push into this real-time space where they can be more competitive and get more value from their media spend. Our ability to do this at scale was down to our early adoption of agents in our content supply chain, which we started doing at the end of 2024. 2025 has really been about scaling that across more and more of our teams. So we'll show some work here for Mr. Muscle. I think this is a great example, very important piece of brand equity. We use the technology initially to refresh the character in a way that would have been a lot slower and more costly if we would have done it in traditional manners. But the real takeaway if we go to the next slide is we're not just doing that refresh and then putting brand guidelines in place. We're creating playbooks that really act as a knowledge base for agents. And then if we go to the next slide, that means we then have agents on top of our infrastructure on top of that knowledge base, allowing you to start using that very important brand asset in real time. We go to the next slide, we can see how that plays out for our clients. In this case, we're using the new Mr. Muscle as a real-time content creator on the local version of TikTok, which is a really interesting way to think about how these brand assets are suddenly a lot more valuable, right? The cost of production and the time of production has really collapsed because of these pipelines and these systems that we've set up. And we're going to show how that plays out in practice with a quick video. [Presentation]

So good example of how we are not just making an ad, we are making system to allow [indiscernible] advertising. We go to the next slide. Another quick example, award-winning piece of work for GMC where we're really using technology to make a very high quality social spot to drive further engagement after a big TV spot aired around the NFL playoffs. But this is a great example of a piece of work that really probably wouldn't have been possible to do within traditional either time lines or budgets, but now was made possible because of the use of AI production. [Presentation]

We go to the next slide. I won't belabor the details here, but I think important note while the technology is clearly transformational to get to this level of level of craft and quality still takes quite a lot of sort of bespoke talent work. So our ability to have these types of hybrid production pipelines, mixing 3D with generative output is quite a key part also of our category offering for automotive. And then the last piece of work, if we go to [ Picton, ] this is, I think, another fun example where for Picton Investments, we did a rebrand. And then as part of that rebrand, we're building a system that allows us to operate more closely to real time. In this case, if we go to the next slide, we have the Bear. The Bear is the new [ Massport Bear Sport Murphy ] for Picton Investments. And it is, again, a brand asset that is usable because of the AI pipeline and agents that we've built. We go to the next slide. Those pipelines allow us to be always on, always relevant. Let's go to a quick video to see what that looks like. [Presentation]

And that is a small example of the work we're doing with AI. With that, I'm going to hand it back to Martin.

Thanks, Wes. Thanks, Scott and Radhika. Can you put the slides up, guys, please? So just to summarize what Radhika and Scott said, 2025 results were in line with revised guidance, which we gave in November. And that showed improved margin and significant improvement in liquidity and net debt. The second point is net revenue at GBP 673 million was down almost 11% reported and down just over 8% like-for-like due to continued client caution and challenging global macroeconomic conditions, which continue into this year. Operational EBITDA at GBP 81.2 million with margin improved by 70 basis points like-for-like to 12.1%. Number of Monks at the year-end was down 11.5% versus the previous year '24 to around 6,000 -- currently to 6,350. Net debt of GBP 86.9 million, which represents leverage of 1.1x EBITDA was down from GBP 142.9 million. That's leverage of 1.6x previous year at December 2024 and well below the targeted range of GBP 100 million to GBP 140 million due to strong working capital management. Subject to shareowners' approval, the Board is proposing to pay a final dividend of 1.1p per share, which is 10% up on last year's final dividend. Post year-end, the company repurchased just under EUR 26 million of its EUR 375 million Term Loan B at a discount, and that includes EUR 1 million, which remains to be settled. Post 2026, like-for-like net revenue is expected to be in line with current analyst consensus. That's slightly below 2025. The increasing global macroeconomic volatility and client caution is expected. But our AI capabilities, as Wes has outlined and our strong relationships, as Scott alluded to, both create significant new opportunities. We're targeting this year an increase of at least 100 basis points or 1 percentage point in operational EBITDA margins, and that reflects in part the annualized impact of 2025 cost-out initiatives. Net revenue is expected to be down in the first quarter, in part due to the ongoing conflict in the Middle East. However, our cost management initiatives will enable us to partially mitigate the full impact of any revenue shortfall. The 2026 net debt target range is GBP 60 million to GBP 90 million with a medium-term leverage target of under 1x operational EBITDA. The company's capital allocation policy is to prioritize dividends first, then further debt repurchases and finally, share repurchases as net debt falls further. We saw momentum in new business wins in 2025 with new or expanded relationships across all of creative, media, technology and AI-driven transformation work and that reflects the group's AI and data-centric positioning as a driver of future growth. And finally, we remain confident in our talent, in our business model, in our strategy and in our scaled client relationships to position us to drive and deliver sustainable long-term growth. So with that, as a summary, we can move operator to questions.

[Operator Instructions] Our first question this morning is from Laura Metayer of Morgan Stanley.

Two questions, please. First one is on the revenue model change that you've been talking about. How quickly do you think you can move away from the traditional time and materials model? And in the meantime, how do you protect your revenue and margin until you achieve the majority of subscription and asset-based revenue model? And then second question is on your 2026 revenue guidance. What's the impact on revenue from net new business that you expect in 2026? And also what macro scenario is the guidance based on?

Yes. I think why don't you deal with the second one, Radhika, first. I mean, basically, there's little net new business in there, but do you want to comment just on the makeup?

Yes. So I mean, as you can see, there's little net new business in the '26 guidance. But what we are confident on is the impact of our annualized cost actions from last year to protect our margins this year and drive it by at least 100 basis points, which is what we just shared with you earlier. So I hope that answers that second point.

Yes. I think we'd just add, we tried to be as practical and conservative as possible given previous experience, which is not what we wanted to achieve. So I think we focused really on existing relationships and then the impact of new relationships from last year and their impact on this year. On the model, Scott, do you want to just comment a little bit on how we see -- I mean, on protection, that's -- there's not an easy answer to that other than -- what we do see in personalization at scale or orchestration as we call it, very significant volume opportunities. So whilst it might be true that visualization and copywriting, we're seeing compression, you're seeing the cost of ads and the time taken reduced. And if you charge on time, that's it. That is compressed. On personalization at scale, I would say there's more opportunity. But Scott, do you want to talk about sort of timing and the 2 models?

Sure. I mean I think Wes can fill in probably more on the actual models. But in terms of the timing, I think it never happens as quick as we would want it. We're certainly ready to do this. We're pushing it. As Wes said, we have -- one of the major clients we won last year is fully on this model, and we're certainly pushing it with our other larger clients. One of our large technology clients, we've agreed a different rate card approach, which is asset-based rather than people-based. So we're certainly pushing it and keen for this transition to happen as quickly as possible. I think as most things, when you're trying to change something really significant like this with big enterprise clients with multiple stakeholders, decision-makers, et cetera, procurement who love to compare apple-to-apple and don't like anyone throwing pairs in the mix. I think it takes a little longer than we would like. And I think it's -- as you pointed out in the question, this is something that's really at the heart of our go-to-market, our product approach, our new business approach. It's something every pitch we're discussing has an element of this built into it, if not fully on this model. It's a little tougher with existing clients because you have existing contracts in place. So that's where it's probably slower. But maybe, Wes, you want to give a bit of an...

I think that's the right way to look at it. It's our de facto offering in new business. New business is looking quite good. If you think about that offering that's driving a lot of pipeline with existing clients, we have opened this conversation with pretty much all of our large clients. I don't think anybody in marketing procurement teams is against a change. Everybody understands that the traditional model really isn't sustainable in any meaningful way. But to Scott's point, doing this at scale in a running business is not the easiest thing to do. So steady does it. But I think this also plays into our consistency. We've been having these conversations for a long time with our clients. So we're, I think, broadly seen as an advisory partner in these types of changes.

Yes. I'd just add one thing, Laura, to it. I mean the market is painting everybody in the sector irrespective of who they are, big or small or medium size, they're painting them with the same brush. They're assuming that they will be -- the agency businesses will be AI losers. I don't think that's necessarily the case. I think in our own case, we do have AI model, which we're implementing or trying to implement at the scale. And the only verticals to date where we've seen significant AI adoption by clients has been autos and financial services. And I would single out autos in particular. Maybe a little bit of packaged goods coming, just go into it in a little bit more detail. On the first autos and -- or first 2 financial services, there are existential threats to autos from Chinese EVs and AVs. In the case of financial services, there are existential threats from fintech platforms. And as a result of that, the pressures on those 2 verticals to adopt AI at scale for increased efficiency because somebody is coming at you with lower -- with good models, which are lower priced with consumer advantages, which are lower priced are prodigious. So I would say instead of thinking about how fast we can move clients to an output model or a subscription model. The far more important thing is what is going to be the pace of AI adoption because the simple fact is consumers have adopted AI faster than companies. And companies aren't doing it because it's not just about technology or workflow, which Wes went through with Monks.Flow. It's about change, change management. And companies find it difficult to do that, particularly complex bureaucratic companies. I don't mean that in a negative sense, but in an organizational sense, it's difficult for them to do it. So I think the answer to your question about new model adoption is really dependent upon the speed with which clients adopt. On packaged goods, we are starting to see a little bit of adoption and the existential issue there is pricing that clients during COVID, particularly in packaged goods, drove -- passed on price increases. They haven't done it with tariffs interestingly, at least not yet, but they did do it with commodity price increases during COVID where they were moving prices up by 10%, 15% or 20% per annum whilst their supply chains were under pressure. So we are starting to see a little bit there, but I would say I would call out autos in particular, probably the order of magnitude is autos first. Autos is quite incredible the pace of change there and the adoption at scale. I would say it's also significant on financial services, but a little bit less so. And then I think packaged goods is starting to start. Just one final point. It may be -- and I don't mean this to draw positives from the conflict in the Middle East or the war in the Middle East, but it may be that, that acts as a catalyst. If global growth slows, inflation rises and interest rates are stickier, which seems to be the scenario already being built in by some of the analysts and the investment banking firms. If that is the case, that might be the engine for increased tech adoption and AI adoption. So anyway, long answer to what you said, but I think it's an important point because the market is certainly painting everybody with the same brush at the minute.

Next question will be from Steve Liechti of Deutsche Numis.

Yes, I've got 2 as well. One, on Marketing Services, say that was down 5% to 6% last year. Can you try and cut out or carve out there how much of that was existing client spend reductions and how much of it was net client losses or wins? And then I guess, question 1B would be, can you give us a kind of pro forma adjustments to think about for 2026 in terms of business wins in '25 that roll into '26? That's the first question. And do you want to do that first?

I don't know how far we're going to get on that. I mean, do you want to try and answer that question for those 1 and 1B there, Radhika? Or is it [indiscernible].

I can do the first bit. I think it's quite difficult, Steve. If you look at our client base, quite a significant amount of our business is project-based business. So it has a defined beginning and end. And so it's quite hard to -- if it ends and it's successful, it's hard -- you shouldn't be considering that a loss, obviously. It was never expected to continue beyond the project stage. So for our large-scale client relationships, it's really the dominant factor in the decline in performance is decline in spend. And you can see that in our annual report, we give detail on our largest client relationship, and that's pretty indicative of what's happened, particularly in the technology sector. You can see it on the slide I presented as well on the sort of average size of clients. So it's really, I think, for large retained ongoing client relationships, the bulk of the decline has been that decline in spend. On the second part, I'm not sure.

I think I'll have to go.

The second part is on clients. I mean, we've named -- if you look at the release, we talk about the relationships in '25 that grew, the significant ones. We called them out specifically. So I think that names it. But the prints -- it's difficult to give you a precise figure other than to say that will be the major part of what's included in '26. Radhika, anything further you want to add on that?

No, not really. I think I can -- I'll try and do a pro forma, but I haven't got anything at the moment to share that will give you an accurate figure at the moment.

And going back to the sort of the 1A of your question, I think most of -- the project point that Scott made. But in addition to that, I would say most of it tends to be reduction of spend rather than total loss. That's what I would call out in '25. It tends to be more about reductions of spend or increases of spend. And again, I just draw your attention to who we've specifically mentioned in the release, not the new business wins that are referred to, but the enlarged relationships that we referred to. That was one. What else?

Yes. I mean, I guess I'd lump together the sort of let's say, scope of work increases plus new clients together that would just be useful. So almost like a pro forming gross and then even down to a net. But yes, we can see what you can come up with. So that's great. So my second question, on tech services, I'm presuming now your client loss or his sort of stop spending has now washed out the figures.

That's right.

So we've kind of got a clean number now.

Yes.

And the market is growing around 5%, we think.

Well, I think -- Scott, what do you think?

Yes, that's what -- that's based on looking at the projections for some of the quoted companies that specialize in that area. So I think it's on the slide that I have.

So is that a fair number to think about for next year?

I wouldn't say it's unfair. Let me put it like this. All I would do is go back to what we said in answer to a previous question that we've looked -- because of historic forecasting inadequacies, I go as far as that. We've tended to be a little bit more conservative. And I think Radhika underlined there is some new business in our budgets for this year, but we've sort of taken that down. So the to-be found column is less. So you've got to think about that, Steve, in relation to what we're projecting.

Yes. Okay. And can I have sort of 2B on tech services again? Just help me out in terms of thinking about that particular business as a possible AI loser because rightly or wrongly...

No, no, that's not an AI loser. That's an AI win. I mean that...

So help me out then because I'm thinking of it as a sort of nearshore, offshore player historically.

No, that's the wrong way to think about it. Think about it as a full part of the transformation process. I mean the type of projects that they are -- have historically won and are winning are not just cheaper because it's offshore. It's not a sort of TCS, Infosys total model. This is trying to implement transformation. And it's broader -- it's enterprise transformation as well as marketing transformation at scale. So they would be at least in theory, AI winners as well. I mean, Scott, do you want to...

Even the Accenture and people like that are under pressure now from an AI and they're obviously doing...

Yes. If you are a -- it's true. If you're an Infosys or a TCS and stuff was moved offshore to -- because it was cheaper. It is true that manufacturing might go back to America as a result of what's happening, wouldn't create employment, would create employment for robots or bots. I get that. But the sort of work that we're doing is AI transformational work. And I think it's a little bit more fundamental in terms of digital transformation and not tied to marketing transformation or the type of work that you're referring to with Accenture. I mean you say Accenture, if I look at sort of the work that Accenture Interactive or Droga are doing, that's not the case.

[Operator Instructions] We'll now move to [indiscernible] of Barclays.

You just highlighted predominantly spending cuts rather than client losses. But if I look at your Page 19 in the presentation, your sort of second biggest client category at GBP 5 million to GBP 10 million in revenue has more than halved. Can you maybe just give a little bit more color on what drove this? And then without giving like a specific guidance, but what do you think your cruising altitude can be in terms of net revenue organic without sort of any impact of client losses and if AI CapEx goes back to marketing OpEx? And do you think sort of related to that, do you think you can go back to growing in line with your underlying markets again?

Yes. Just go back to -- just deal with the second one first. I think it's extremely difficult to answer. We are seeing signs of sort of stabilization of marketing spend. But I think it will be -- it's a virtual impossibility to call out now what you referred to as our cruising altitude. I think what we want to do, obviously, to get into positive territory. That's what we are really focused on. And I think that goes back to a previous question. That depends, I think, if you think about the 5 areas that we think about in terms of AI transformation, the first 3 are probably the most meaningful with digitalization and copywriting, personalization and scale of media planning and buying. And it's really a question of how fast that adoption takes place. We're referring to the verticals. On the first part of the question, Scott, do you want to comment on the segmentation of the client count sizes?

Yes. So as I said before, I think a lot of our business, particularly at the lower end of that chart is project-based business, and there's been less of that. So that explains a lot of the sort of lower numbers of clients. In the higher brackets, what we've seen is that declines in spend have pushed clients down a bracket. So that's really what's happened there again. We haven't had beyond First American and sort of clients in tech services, we haven't had large losses in marketing services. We've had some clients that have had quite large declines in spend, but they remain clients.

As we have no further questions at this time, sir Martin, I'd like to turn the call back over to you for any additional or closing remarks. Thank you.

Okay. All right. Thanks very much for joining us. Anybody has any further questions, so we're all here to answer them. We'll see you shortly, hopefully, to talk about quarter 1. Thank you very much.