Auto Trader Group Aktie

Good morning, everyone, and welcome to Auto Trader results for the full year ended 31st of March 2026. Today, with myself and Jamie [indiscernible] Q&A in the room.

We anticipated trading this year to be tougher as a result of profitability challenges for retailers, a shortage of stock in some age cohorts and fast speed of sale. Retailer profitability was more than we expected due to a combination of new car profitability in part due to the ZEV mandate and cost increases following the government budget last year. This led to intense scrutiny on every cost in their business with many stories of profit declines, redundancies and store portfolio restructures.

The pressure on retailer profitability is most acute in November and December, which combined with feedback on our accelerated rollout of Deal Builder. This was amplified across social media, including some factual inaccuracies, which was since clarified for customers.

Following this period, we have seen higher cancellation levels than in previous years, which has impacted both FY '26 and the run rate into FY '27. However, despite these challenges, we have continued to grow revenue, profit and earnings per share. Perhaps more importantly, retailer numbers, stock and upsells have all been growing since the end of the financial year, so we are past the low point.

Furthermore, the core metrics underpinning Auto Trader are in a strong place. Supply shortages will subside. Speed of sale has been stable all year. Retailers are returning and now years into Agentic AI, we are confident that the core of what we do will remain relevant for the future. It's a confidence that is growing as we continue to build, scale and monetize products incorporating AI. In previous technology transitions, including the Internet, mobile, native apps, hyperscalers, big data and AI, we have backed the technology and emerged to the other side better for car buyers, for retailers, for our people and shareholders, too. Now we'll turn to the core foundations of Auto Trader, which, as I mentioned, are in very good health. The most important foundation of any marketplace is its buyers. From an already very high base, our share of time spent versus competitors has increased again. There are 11x more time spent on Auto Trader compared to our nearest competitor, which is actually 4 brands added together. We're 6x greater than all our main competitors combined and have seen little change to how buyers use Auto Trader. Unique visitors have been stable at over 9 million a month and 80% of them come direct to Auto Trader, 13% through organic search and 4% through paid web traffic. AI chatbots represent less than 1%, which isn't changing significantly despite monthly LLM users reaching almost 4 billion globally. Generative chat interfaces and agents are already being used to research goods and services. However, how people use them depends very much on the category. Vehicle transactions are unique because they are pre-owned, complex, high value with a lot of relevant choice for any one buyer and a process that takes around 3 months straddling both on and offline processes.

Our experience has shown that car buying and selling is improved meaningfully only when the technology is combined with a deep specialized car buying experience, a massive range of real-time vehicles and tools that utilize data, integrations and over 50 AI models that only we have and have developed. This is true for car buyers and also for retailers who have an even higher need for trust, performance and accuracy. This is all underpinned by a sustained investment in our public cloud delivery, data and AI technology platforms with our 400-person strong product and technology team.

Now to Deal Builder. I want to start by saying this remains a big long-term focus for us because we believe it is one of the single most impactful and difficult to replicate experiences we can deliver for car buyers and retailers. This is indicated by positive feedback over the 3 years that it's been live. Car buyers like being able to go deeper into the transaction when and where it suits them. They don't want to wait for return phone calls or dealership opening hours. For retailers, they can sell out of hours and get car buyers that are at least twice as likely to convert to sale. Furthermore, over time, there will be the opportunity to get more car buyers part exchanging and using finance, which are good options for car buyers and important for retailer profitability. While we have had to make some product adjustments and taken more time onboarding some cohorts of retailers, penetration has continued to grow throughout the year. We're also updating our capital allocation policy. We have accelerated our buybacks throughout the second half of the year and intend to continue this into the new financial year. We believe this is a rare opportunity to allocate capital effectively at a price that we don't believe reflects the fundamentals of our business. Over FY '26 and FY '27 combined, we expect to return over GBP 1 billion to shareholders.

As I said earlier, we've continued to grow revenue and earnings. Average retailers for the year declined 0.5% to 13,942. We ended the year with 460 less paying retailers than where we exited the first half. Although, as I mentioned earlier, retailer numbers have been growing since the end of the financial year, which we are very focused on continuing. Average revenue per retailer, or ARPA, was GBP 2,995 per month, which was up 5% on the previous year. Group revenue and operating profit were up 4% year-on-year and operating profit margins were stable. Averages can be deceiving, so it is worth flagging that group revenue growth was 3% in the second half and lower in the final quarter, which has impacted our run rate into financial year 2027. Cash generated from operations was up 5% and earnings per share up 8%, higher than operating profit due to our continued share buybacks, which, as I mentioned, we accelerated in the second half of the year.

We have reviewed and changed our capital allocation policy, as I mentioned, reflecting our confidence in the business. And this year, we're also declaring a final dividend of 7.8p a share, which makes dividends for the full year up 9% year-on-year.

Now for our cultural KPIs, which are a subset of measures we use to track progress on a number of cultural and organizational priorities. This year, our metrics around the representation on our Board, leadership and organization are relatively stable and generally at good levels. Following Catherine's departure, our Board is now 50-50 men and women, and we're continuing to work to improve ethnicity and leadership, but it will take time for the work that we've done in early careers to make its way through to leadership positions within the business. We aim to have net zero carbon emissions across our value chain by 2040 and halve those emissions by 2030. This year, carbon emissions across Scopes 1, 2 and 3 increased 55% to 144,100 tonnes, largely as a result of capital expenditure on our new office and vehicles taken on balance sheet through Autorama. Both of these are Scope 3 or supply chain related.

Engagement has fallen from 91% last year to 72% this year. It was literally only a year ago when most cultural measures were at all-time highs. As you have heard, this year has been particularly challenging and has impacted our people, tougher trading, tighter cost control, reorganizing some areas of the business, retailer feedback and a tighter approach to working in the office. However, all other internal measures such as recruitment and retention remain largely unchanged. It's fair to say the team have performed exceptionally well in tough conditions, and I feel privileged to work with such a talented and committed group of people.

I'll now hand over to Jamie to talk us through the financials in more detail.

Thanks, Nathan, and good morning, everyone. I'll start by focusing on the core Auto Trader financials.

Starting with revenue. Total Auto Trader revenue increased 4% to GBP 585.3 million. Trade revenue also increased by 4%, with the largest component of it being retailer revenue, revenue, which also grew by 4%.

Within trade revenue, we've seen an increase in Home Trader pay-as-you-go listings and growth in other trade revenue.

Consumer Services revenue decreased by 8%. Within this, private revenue generated from individual sellers decreased by 11% due to a lower volume of listings.

Motoring services decreased by 4% due to lower revenue from our insurance product.

Revenue from Manufacturing and Agency customers increased 14% year-on-year. Much of this increase was due to manufacturers supporting their franchise network with both new and used car advertising.

As mentioned, retailer revenue grew 4% year-on-year. The average number of retailer forecourts on our platform decreased by 71 to 13,942, which was 0.5% year-on-year decrease. And average revenue per retailer increased by 5% to GBP 2,995 per month with more details given on the following slide.

The chart on the left shows the components that contribute to the movement in ARPR compared to the prior year. As you can see, ARPR growth was driven by the price and product levers with a negative contribution from stock.

We delivered our annual pricing event for all customers on the 1st of April 2025, which included additional products and a like-for-like price increase, which contributed GBP 117 to ARPR growth. Products contributed GBP 72. Most of this growth was from our Co-Driver product, which is included in retailer advertising packages in April 2025.

Prominence, which is largely the movement up and down our package staircase, was not a contributor to the product lever in the year. We continue to review our packages with changes expected in H1 with the aim of returning prominence to long-term growth.

The remaining product lever growth was driven by new car, where we increased the number of paying customers over the period.

Turning now to stock. You'll see on the right-hand side of the chart that the number of live cars advertised on Auto Trader was broadly flat year-on-year, although there was some positive impact from a stock offer that we ran at the start of the financial year, which did not impact revenue.

From November 2025, prompted by the rollout of Deal Builder and reflecting more difficult cost-related trading conditions, a number of retailers reduced the number of vehicles advertised on the platform, contributing to lower paid stock volumes.

Total costs -- Auto Trader costs increased 4% to GBP 181.4 million. People costs increased by 1% to GBP 93.6 million. There was an increase in underlying salary costs, while the average number of employees remained broadly flat and share-based payments reduced. Marketing spend decreased 11% to GBP 21.9 million, while other costs, which include data services, property-related costs and other overheads, increased by 13% -- this year-on-year uplift was mainly driven by higher cloud infrastructure expenditure and increased property costs related to our new head office. Depreciation and amortization increased by 49% to GBP 9.4 million, also due to our new office lease that commenced in July 2025.

As a reminder, we fully expensed our research and development costs, hence, our low levels of CapEx and depreciation. In addition to our investment in cloud-based services, we have 400 people in product and technology who are continuously improving our platforms and developing new products for consumers and retailers.

Operating profit increased by 4% to GBP 408 million, and operating profit margins remained at 70%. Our share of profit generated by Dealer Auction, the group's joint venture, increased 14% to GBP 4.1 million.

Having covered Auto Trader, the main part of the group, we'll briefly cover Autorama results. As a reminder, the Autorama acquisition was and remains part of our strategy to bring attractive new car offers to car buyers on Auto Trader and to make new cars a more important part of our proposition.

Autorama revenue was GBP 39 million, with vehicle and accessory sales contributing GBP 29.6 million and commission and ancillary revenue of GBP 9.4 million. Vehicle and accessory sales relates to vehicles that flow through the balance sheet, which is not our primary focus.

Total deliveries amounted to 8,056 units, around 2/3 of these being cars. And within this, deliveries from Auto Trader increased over 3x. Average commission and ancillary revenue per unit delivered was GBP 1,167. The Autorama segment made an operating loss of GBP 2 million. This is a significant reduction on last year through the accelerated integration into the main Auto Trader business and platform.

From financial year 2027, Autorama will operate and be reported as a single operating segment for the rest of the Auto Trader Group. This is due to more than half of all leasing transactions being delivered through the Auto Trader platform in the second half of financial year 2026.

In the appendix, you'll find a breakdown of financial year '25 and '26 in this new format, which will be how we report our FY '27 half year results in November.

Total group operating profit increased 4% to GBP 392.7 million, and group operating profit margins remained at 63%. As we grow, the strong cash generation of our business leaves us well placed to return surplus cash to shareholders. Cash generated from operations was up 5% to GBP 418 million.

Now to briefly review net bank debt and uses of cash in financial year 2026. During the period, the group drew down GBP 165 million of its revolving credit facility and held cash and cash equivalents of GBP 18.2 million. In the year, prompted by the lower share price, we accelerated the number of shares we acquired. To put this into context, we acquired 3.5x more shares in H2 than we did in H1. And in aggregate, acquired 58.5 million shares, 6.6% of issued share capital for a consideration of GBP 369.1 million. A further GBP 94.1 million was paid in dividends.

The Board believes the prevailing share price does not reflect the company's fundamentals or long-term prospects. We're therefore updating our capital allocation policy. We'll continue to focus on investment in the business supporting growth while continuing to return approximately 1/3 of net income to shareholders through dividends. We expect to continue the recent acceleration of share buybacks, purchasing around GBP 500 million of shares in the year, and we'll be requesting authority to require up to 15% of issued share capital at our AGM in July. Based on current assumptions, we expect this to increase our debt levels towards 1 turn of EBITDA.

I'll now move on to talk through some of the market dynamics. Moving on to Slide 15 and looking at both new car registrations and used car transactions. From a new car perspective, supply has continued to improve following the impact of the pandemic in 2020 and '21 with 5 consecutive years of growth, although the market still remains slightly below the level seen pre-pandemic and significantly lower than the highs of 2017. The retail market or new cars sold directly to consumers continue to remain at lower levels by historic standards.

Over the past 12 months, we've seen manufacturers attempt to stimulate private demand with increasing levels of discounts and finance offers, which has had some positive impact with private retail sales growing for the first time in 4 years. As seen in the chart on the right, used car transactions have continued their steady recovery, growing at 1% over the last year.

Over the past 20 years, the size of the U.K. car park has grown steadily by just over 300,000 cars per year, reaching 37 million vehicles. We expect this growth to continue over the long term, driven by GDP growth, population growth and stable car ownership trends. We also expect the value of both new and used cars to rise over time. In 2011, the average price of a used car advertised on Auto Trader was GBP 9,000, which has almost doubled since then, reflecting average growth of over 4% per year. This increase is driven by inflation, improved product functionality and the shift towards more expensive electric vehicles.

Over the past decade, gross margin percentages have remained relatively consistent, which means that higher vehicle prices typically lead to higher absolute gross profits for retailers. This trend, combined with the growth in transaction volumes has resulted in an increased gross profit pool over time. Over the past 12 months, our audience position has remained strong. The number of cross-platform visits continued at record levels at 81.7 million per month. Engagement, which we measure as cross-platform minutes, decreased slightly to 548.3 million on average per month. The chart on the right-hand side shows the total minutes spent of an expanded set of competitors. On average, over the year, Comscore estimated that consumers spend 11x more minutes on Auto Trader than our nearest marketplace competitor. The combination of Gumtree, Motors, eBay and Cazoo, which was an increase versus the prior year. We were 22x larger than the nearest single brand and 6x larger than all of these competitors combined. 67% of our audience was unique to us, never visiting these other sites. I'll now hand you back to Nathan to talk through our latest product updates and the outlook.

Thank you, Jamie. We wanted to start by providing a timeline, including a sample of how our technology has developed over the past decade. It is this sustained investment in the way we deliver software that enables us to deliver at high velocity with stability to adapt and to scale new technologies quickly and cost effectively without big-bang replatforms, which are prone to time and cost overruns.

Rather than go through every item on the timeline, there are a few I wanted to point out. Just over 10 years ago, we began building our data science team and supporting technologies. 2017, we started to migrate to the public cloud by rearchitecting and standardizing every application, so it didn't just work in a cloud environment, but was optimized for it. We completed this without pausing development around 6 years later. Just after 2018, we acquired key resources so we could build our own proprietary vehicle taxonomy for every vehicle on U.K. roads, something that is even more relevant in a world of AI and requires a team of technologists and people to constantly update and cleanse the data as well as build integrations with manufacturer production systems and other data providers.

Our taxonomy underpins pretty much every aspect of Auto Trader. Following the launch of ChatGPT in late 2022, we employed a full-time post-doc from Manchester Met University to work on the LLMs. This capability is now productionized across our data science and engineering team and is made accessible to all our product and technology teams through an AI platform integrated with most foundation models, including hosting, scaling, monitoring and security essentially available to them off the shelf. AI will increasingly surface in our consumer experience, sometimes very obviously and sometimes in the background.

We believe the technology is transformative, but not if it is just bolted on the side of existing functionality and tools. As we've built, scaled and monetized AI, we have seen that to be effective in car buying or retailing, the technology needs to be combined with a deep user experience, highly curated real-time vehicle listings, proprietary data, specialized models and distribution to a highly fragmented customer base of car buyers, agents and retailers.

We've now built and deployed an MCP server, which enables agents to efficiently interact with Auto Trader. Our initial implementation of this is with a ChatGPT app, which you can see in the middle of the slide. However, we have architected it in such a way that it can be used to interact with other agents that arrive either today or in the future.

Another example is our new search categories available under a filter called "I'm looking for" on the left-hand side of the screen. On Auto Trader, car buyers had to choose one make or model when searching for vehicles. These filters powered by a specialized AI model enable people to search across makes and models defined by categories in their language, such as big boots, all rounders, cheap insurance, cheap to run, first cars, luxury cars and even an OMG category. These filters are currently being applied by over 100,000 people every day and will become more and more important given the explosion of new brands selling vehicles in the U.K. It's also a very good example of what's required to ensure high-quality responses. We could have used a simple LLM to do this task, which would have been less accurate and expensive. Instead, we deem machine learning to be much more effective for the task with negligible incremental costs.

On the right-hand side, you can see our current pilot using chat-based search, which is integrated with filters. This provides a different entry point to search with even more flexibility than categories. We've experimented with these interfaces using different technologies before and uptake was limited. However, if LLMs become a way in which people want to search for vehicles, then the best such experience for cars will be on Auto Trader. This is because of the specialized models I spoke about earlier that use proprietary consumer and retailer data and an odd experience to navigate enormous choice along with market-leading tools to ensure car buyers get the very best results for all their car buying questions.

Our Co-Driver suite of products will be known to most analysts and investors. For those who are unfamiliar with Co-Driver, it is the umbrella brand for a suite of AI products that save retailers' time, improve quality and therefore, their sales process. The first 2 products already available optimize the order of a retailer's images and highlight any that are missing using computer vision. The second, which is covered on this slide, writes a description for each unique vehicle, which sounds simple enough, but is a perfect example of what it takes to get good results from this technology. If you use the foundational model alone, as is shown at the top of the slide, there will be errors and many of the features of the vehicle will not be known. Furthermore, there will be no data-based understanding of which of those features are the key points of differentiation and of most value to car buyers. At the bottom of the slide, you can see how we go around this by utilizing both our models, the foundation models and proprietary data and technology. We found that every implementation of LLMs to make car buying or selling better has required us to do some version of this.

This year, we also launched Buying Signals, which is again powered by AI models. It's been added now to 800,000 inquiries since it's been launched. Buying Signals utilizes the model to predict the likelihood of someone buying a car based on signals collected throughout their journey on Auto Trader. If retailers need to prioritize their activity, this gives them the perfect method to do so quickly. It also benefits the ready-to-go car buyer who can be quickly identified, contacted and followed up. Buying signals also flag whether that buyer is local and their preferences in case the exact vehicle is no longer available. It offers a similar efficiency to Deal Builder with inquiries marked as high intent, typically converting twice as well as an average Auto Trader inquiry.

Now with over almost 15 million inquiries sent every year to retailers, many of which go unreturned, the benefit to them is reasonably obvious.

The new buying journey on Auto Trader known as Deal Builder remains a key focus of ours. It delivers a more empowered buying experience whenever people want and inquiries to retailers that convert at least twice as well. For a retailer, that's half the work and with almost 50% of deals happening outside of business hours, their stores are essentially now open 24/7 without the staff costs that would normally come with it. We always build products iteratively and with customers, which we have been doing for 3 years now with Deal Builder. However, as we accelerated the rollout to new cohorts of customers, some were not clear on what we were offering or why. We listened to this, slowed down and made changes to the product, including the ability to choose either full reservations or the ability to request a reservation. The changes were well received, and we have continued to roll out the product, retaining the goal of achieving 100% penetration during the course of FY '27.

You can see from the charts on the slide that penetration has continued to increase throughout the year with more than triple the customer numbers now on Deal Builder and almost triple the number of deals as newer customers tend to be smaller.

Now to the outlook for FY '27. We remain comfortable with our current levels of investment, so we expect group operating profit margins, excluding vehicle and accessory sales to be at least maintained. In the financial year 2027, we expect group operating profit to be between GBP 395 million and GBP 415 million, which with the continued acceleration in buybacks will result in at least high level, high single-digit EPS growth.

Auto Trader revenue was flat year-on-year in April 2026 due to a lower run rate and package increase. However, retailer forecourts, volume of paid stock and package penetration are now improving, and we expect that to flow through to growth in the second half. The contributors to this will be our packaging event, which will grow the price lever within ARPA by GBP 85 to GBP 95. Product growth is expected to contribute GBP 65 to GBP 75.

Stock will recover, resulting in improvement from current levels to minus GBP 30 to GBP 40 for the full year. While average retailer forecourts are now growing, we expect the average for the year due to the entry run rate to be down 1% to 2%. Other revenue will be broadly flat in aggregate with a decline in Consumer Services, offset by growth in Manufacturer and Agency. Our focus in Autorama is on volume growth. However, we do expect to make a small profit for the year with commission and ancillary revenue growing 8% to 12% and vehicle and accessory sales of around GBP 40 million.

As Jamie mentioned, as the majority of leasing transactions now originate on Auto Trader and due to the extensive integration we've already undertaken, we will move to one reported operating segment in 2027. That's all for the presentation. We'll now take questions from analysts in the

room.

It's Andrew from Barclays. Two for me, please. First one is on the stock and the guidance you've given for the year. And just trying to kind of unpack what you're seeing in terms of stock in the fiscal year-to-date on an underlying basis because it's quite hard to observe that on your side. There's quite a lot of noise with special offers and that type of thing. And then what's kind of baked in, in terms of underlying improvement to get to your guidance? And if you can kind of unpick that between what's fast stocks in the market and what is dealers choosing to put their stock on Auto Trader. That would be helpful.

And then the second question, I appreciate the FY '28 pricing event is still 7 months from now until you communicate it to dealers. But is there anything you can give us in terms of how you're thinking about it and what products or what type of products you're going to push to the dealer base next year and how you kind of think about the value of it? And I guess, big picture, in the last couple of years, the headline increase to dealers has been 8%. This year, it was 5.5%. What kind of gives us confidence in the value of the products that you have that 5.5% isn't the new normal and it can get better?

I'll take the first one. So I think there's a bit of a theme. You obviously asked around the stock guidance, but I think it could equally be applied to the retailer guidance and maybe to a slightly lesser extent to product and particularly prominence. So I think if you strip the FY '26 numbers into first half, second half, you can pretty quickly see that second half, particularly retailers, the stock lever are more negative than they were in the first half. And if you think about the nature of where that sort of softening happened was very much fourth quarter weighted. We talked a fair bit in the presentation around the sort of cost-related pressure that customers felt and the catalyst being the sort of Deal Builder rollout, that was sort of November. So you feel a lot of that impact in the fourth quarter. And Nathan talked about retailers sort of exiting 460 lower than where we ended the first half. Stock, if that's a minus 85 in the second half, you can assume that it's lower than that where we're sort of entering. So to get to minus 30 to 40 needs that sort of steady improvement continuing that we've seen at the beginning of this financial year.

I think we're pretty pleased with those signs, particularly retailers and stock that we've seen this year. Obviously, there's offers that we've run alongside the price increase, which has maybe created a little bit of a catalyst, particularly for the stock and prominence.

I think there's also people that cover us will know that we had a stock boost that ran last year that converted at the beginning of June. So there is a tougher comp in the first half when we lap that. So again, and it gets easier in the second half. So I think to hit that guidance, particularly on stock, bearing in mind where we're coming in, does need that sort of continued improvement. But pleasingly, we are seeing some of that at this particular point in time. And I think -- as regards to kind of trading, where it was certainly felt tougher for customers in November and December. I think generally, the sentiment feels as though it was a better January to March, certainly, which is also helpful from the stock lever perspective.

And On the pricing event for next year, I mean, we haven't made a firm decision. I mean, I think you know that both the products and the actual size of the event is something that we do -- the product we tend to choose quite a bit earlier, but the actual size of the event, we take into account things that are going on at the time. On the question of how do we feel about our product road map and things that could be included in that event or indeed stand-alone, I think the answer is that we feel good. And whether that's the advertising products, Jamie spoke about packages, they've actually been the piece that have been missing over the last few years, and that's been driven by a very, very fast stock turn that's actually not only been fast, but been accelerating at the same time as well as dealers having pretty acute pressures on their own profitability.

So those products still do what they've always done. You get more, you pay more. We know how to do those. We've been doing those for literally decades. And we do think we are actually looking at those products, relaunching or recalibrating those products at the moment. So I think, as Jamie said, we think they should be contributing to growth. It's kind of a core part of the business model and goes along with the packaging event.

As for the packaging event, I think there's probably 3 areas that we think about where there's plenty to go up. One is insight data, pricing, inventory management. That is something that any retailer that you speak to will probably talk as much around that when it comes to Auto Trader as they would talk about the advertising. Normally, actually, we talk to them more about that than we do the advertising itself. That feels like a rich theme of product development.

You've got the Co-Driver suite of products, but think of that more generally as things that we can do to help them be more productive and have an easier life. And then there's digital retailing. Now first 2, insight and productivity, it's very, very clear that AI extends the runway on that. It allows us to do things that we just weren't able to do before. And we're already showing that, sorting out images, writing descriptions for them. There's a lot further we can go, helping them have conversations, helping them qualify leads. We're sending 15 million their way. That's a lot of human work to be done as an example. So it feels like there's plenty for us to go. In digital retailing, we're really only at the very early stages of that rolling it out and getting into the customer base. So there's a lot more potentially for us to do around finance, part exchange. Now don't ask me next year, why haven't you done all those things because that's kind of 10 years of work, but there is plenty for us to go at.

So that feels good and feels like product ought to be able to support package events and also stand-alone products in the future. One thing that we did see through some of the feedback we had subsequent to November where we've engaged pretty constructively with retailers is some of them like that thing, they don't need this thing. So actually thinking about how we might provide a bit more choice will probably be something that we do moving forward.

I would say for next year, we've got quite a few opportunities. If you look at what customers were actually -- a lot of the talk is around Deal Builder and they weren't really happy with Deal Builder. If you look at the feedback from the customer advisory groups, which I go to, I host, actually, it was about localized pricing metrics. It was less friction in portal that they wanted to stand out and be more individualized, better reporting around leads and us to communicate a bit more effectively through channels that they use. Those don't sound like the most sophisticated products, but they really matter actually to our customer base.

So doing some of that stuff is just built into our plan for the whole year. Not so for the question around. We did 5.5% this year. It's been 8.5% in years before. I think the reality is this year, we're sensible people. There was feedback. It's not the year to kind of be tend to those sorts of things. And also retailers are under acute profitability pressures. If asked, if retailers are under acute profitability pressures at some time in the future, will you take that into account? The answer is unequivocally, yes, we will. We think that's the right thing to do. But that does swing both ways because the 8%, 8.5% that you spoke about over the past few years post-COVID were partly due to the fact that the products were good, but profitability was also very good at the time as well.

I think rewinding back into ancient history to the IPO, we talked about events being pricing product combined should be between 5% to 7%. I kind of feel comfortable that that's still true. And I think we've got some interesting product opportunities that might be able to be a bit more

stand-alone.

It's Will Packer from BNP Paribas. A couple for me. Firstly, the forecourt metric is something the market is very focused on. It's encouraging to hear that there's been some improvement in April and May. The numbers you communicate are averages of averages of averages. So it can sometimes be quite difficult for us to distinguish what's really going on. Can you tell us what the trough number of forecourts was at the bottom, which I assume was Feb or March? And how many you've gained in April -- maybe March, April and May to help us understand the cadence there?

Secondly, a common question I get from investors is, can you explain where those who've churned off Auto Trader are getting their digital marketing needs served? I'm sure you've been in dialogue with some of your customers and -- or former customers -- is it CarGurus, Motors? Is it Google? Is it ChatGPT? Just some color there would help us understand, as I suppose your guidance implies that some of those partners will stay off the platform for as long as 12-plus months.

And then lastly, alongside Deal Builder AI disruption risk has been a kind of key focus. Could you talk through how LLM-originated traffic has developed on your platform? I suppose what we're hearing from most of your peers is it's pretty limited in terms of its volume and scale. Is that the same for you? And how do you think about that?

So yes, I totally take the focus on the forecourt number. I mean if you take that sort of -- appreciating the averages and the exits. So we take that sort of first half sort of just over 14,000. Nathan has then talked about the 460. So that's from the end of the first half to the end of -- the end of September to the end of March. That's how many net has come off. Some of that does have the kind of seasonal churn, but also clearly some reaction. So you're coming down there to 13,500, just a bit over 13,500. Then we have improved through April and May. The reason why I'm not going to put a number in terms of how much is we always run new business offers, and we have had a new business offer running from probably the middle back end of February into March, and that's driving some of the uptick. What we don't know yet is how that offer converts. But I would say I'm very encouraged with how it's gone. And as much as it's winning back customers, we actually seem to be acquiring new customers. Some of it out of our Home Trader line. But Home Traders, the guide is to be relatively flat. So it doesn't feel like that's impacting the revenue. And obviously, we're only 7 weeks into the financial year. So we don't want to get ahead of ourselves. But yes, overall, I think we're encouraged at the trend.

And On where have the dealers gone, I think I mean, we kind of said this in the presentation, but it shouldn't be missed. The challenge to retailers, some retailers' profitability is pretty acute. So some of those retailers just don't exist anymore. They've come off -- come out of business, sorry, altogether. If they own their property, maybe residential property development seems to be a common one for many retailers may be sold out and being absorbed into other businesses as well or combined. So there's a chunk that have done that. There's a chunk that have come back. So where they went is kind of a bit less of a concern because they're kind of flowing back to Auto Trader.

I think there's also a chunk that are just using a whole combination of competitors, and I wouldn't say that we really honestly -- I think we hear a lot less about some, but probably wouldn't be right to comment on individual competitors and more about some others, but it is literally a whole mix. You hear, and you can see this in the social channels yourself, they're trying to cobble together a whole bunch of the others put them together to try and make up that response. Now our 11x, which is independent data from Comscore would suggest that you could probably add it all up and get some leads, but whether you get all is yet to be seen. And a prominent trade journalist did this experiment, had cars on Auto Trader, had cars on other platforms, you'll tend to get bleed across. If you advertise on Auto Trader, you pick up leads on other platforms and found that in terms of cost effectiveness, Auto Trader was the most, but that doesn't mean the others just don't work at all. They tend to be technically expensive, which is why we've always had more retailers than others.

I think I'd echo given that we spend a lot of time with our peers, we're all seeing pretty much the same thing. It's a line that kind of goes like that, not like that. And at the moment, it's kind of somewhere between 0.5% and 1% doesn't -- yes, it doesn't look like it's really increasing. And that doesn't actually -- pardon me, that doesn't seem to be -- I don't think that's going to change massively because we've got an app embedded into ChatGPT for us was more about building the Agentic infrastructure than it was trying to increase that traffic share.

And just as a quick follow-up. It seems -- the next focus seems to be Agentic and how classifieds kind of interact with the future of Agent there's lots of uncertainties as we don't quite know how it will look. We've heard quite divergent things from classifieds, some front-footed the future is Agentic, others a bit more skeptical.

Any kind of quick comments on your initial perspectives on that?

I'll try and give you a third view. So my view is we don't mind, whether a person sends -- comes to us directly or whether they send their agent, we're going to make sure that both of those people when considering where they go or agents, sorry, so not both people, whether it's the person, whether it's the agent that they choose Auto Trader. And we have confidence that, that will be the case because they'll end up making that decision based on the same parameters. Is it quick? Is it efficient? Does it provide me the most choice? Will it -- if I'm an agent, will this thing give me the chance to answer the question as well as I possibly can for the user, which is the way any agent is going to be measured and how they optimize their platforms, we think the answer today is yes, without doing anything other than building an MCP server that they can use, but we'll go even further to make sure that is easier and easier and easier. And if the alternative for the agent is to try and scan 14,000 of our retailer websites, none of who will have -- well, very few of who will have an MCP protocol, you're relying on scraping and unstructured data. We don't charge for our service. So we think it will be okay, and it will be some mix of the 2. Our CTO being particularly open to working with machines as opposed to human beings would say that he'd probably be quite happy if it was more agents. I suspect we'll still have lots of people coming to Auto Trader and then interacting with the agents in lots of different ways. To the extent that they're doing things like within chat interfaces, I suspect Auto Trader will appear in a similar kind of philosophical way as we do in Google today. At some point, we'll appear, will help. They'll come off to Auto Trader to come and do the real detailed stuff because the interface doesn't work for that.

Sorry, just to clarify and answer just on the LLM traffic. I think it's actually a little bit less than 0.5%. Rather than between 0.5% and 1%, less than 1%. But it's actually less than half just to clarify.

It's Lara Simpson from JPMorgan.

I just wanted to come back to the guidance on Autorama, which feels quite bullish. So just trying to understand what's driving those 2 line items. Obviously, on the commission revenue, you're talking about 8% to 12%. It was still down at least mid-single digit in H2. So what's underpinning that inflection? Is it sort of pricing or volume?

And then similarly, on the vehicle line, I think your guidance underpins probably double-digit increase in vehicles. What's driving the confidence in the outlook there? And are you taking more stock on balance sheet as a result of it? And I suppose the last question on Autorama is, are we seeing any change in strategy there on the long-term view? Because I thought longer term, you've been running down that vehicle line. So just trying to understand the moving parts.

And then just one question on the private revenue line. I know getting smaller, but it obviously remains in decline. Can you just talk about what's driving that pressure from a private individual side? Is it competition? Is it just slower stock turn? And maybe just a bit of color on the competitive dynamics and where that traffic is going?

I can take both. So the Autorama -- so start with the commission and ancillary revenue. What's sort of held it back this year is the decline in vans, like cars has actually grown pretty strongly. Volumes on Auto Trader have grown relatively strongly. And I think that's down to -- it very much is focused on the Auto Trader platform, the car journey. And there is naturally less focus on vanarama.com and the volumes that are getting driven through that channel.

So I think we're of the belief that, that van volume is likely to stabilize. And so that's not going to be a headwind, and we're going to continue to drive more car growth. So that's the driver behind the 8% to 12% revenue growth from a commission and ancillary revenue perspective. The vehicle and accessory sales is really just growing in line with the sort of volume of units that we think we're going to do. So I think it's more that the share of vehicles that wash through the balance sheet is more likely to be consistent in this year.

I think it is still a longer-term goal that we're not reliant on that volume. But ultimately, we're still sub 10,000 deliveries in the year and getting access to that inventory is still an important part of the volume. But it's still longer term is something that we want to and believe that we will move away from. It's just not at this particular juncture. And I wouldn't read too much into, obviously, if it's going up or down. It's obviously just washing through and doesn't have a real impact on profitability.

From a private revenue perspective, I think we have seen over the last probably 2 years, it's been quite a competitive environment. It used to just be car buying service. We buy cars, obviously, the biggest we were competing within that space. And then you've had in this intervening 2-year window. Motorway have clearly grown their volumes, Carwow created a similar proposition to Motorway, and that's given people more options between the proposition of selling privately, where generally we'll say it takes longer, but you'll get the best value for your vehicle versus a car buying service where it's very quick and efficient, but you're probably getting a lower price. This is something in the middle.

And I think it's is something that we have to navigate our way through competing with, like we do now have a proposition that's live where we're leveraging the dealer auction joint venture to do a similar car buying service, similar to what Motorway and Carwow do.

It's small volumes, but is growing. So obviously, implied in the guidance is the decline is going to be less in fiscal '27 and bit like probably how it feels all of the guidance. Where we're entering is probably a bit weaker, the implied sort of exit position for the year is certainly closer to flat, if not positive. And that will be a combination, I think, of -- we think there's a real power in putting to the seller that choice of what you might get for the vehicle over time and hours and effort and then the people making the decision all in one place on one platform.

So I think we're a little bit more optimistic that we're kind of getting to grips with a competitive challenge. Clearly, we're very fortunate to have the brand, the audience, most people looking for their next car or have a car to dispose of. We also want to make sure we're supporting retailer part exchange because that's a big part of their profit pool, but we can offer the kind of holistic options to a seller. And I think we're just starting to get some traction in terms of that proposition versus maybe where we were 12 months ago.

The only thing I would add to that, agree with everything that Jamie said, but I would not put down the decline this year to competitive pressures. I think we have got WeBuyAnyCar, Motorway, Carwow have been out there for years and years and years. The thing I would flag is that we do see private revenue move around. I mean it was only 2 years ago where it hit the highest levels that it's ever hit, and there was still the same competitive environment. So there is a bit of an element that it seems to move around with macros. We've not really been able to work out what that correlates with, whether it's interest costs and people getting out of finance deals and feeling that they should get the extra GBP 600 out of the car. So that's the only thing I would add to what Jamie said there.

Will Larwood from Berenberg. Firstly, just in terms of new capital allocation policy in terms of the leverage now going to be at 1x by year-end. How should we think about that? Is that the new normal, new level?

And then also you talked about evolving the packaged staircase in H1. Just wondering if you could share more detail about that and particularly in relation to prominence as well.

Second Yes. So I think I mean we're taking a view on the new capital allocation policy. I mean, I think -- yes, I mean, I think I would expect based on current assumptions. I don't think we feel as though that is sort of over the top. We even say by some company measures, not where we've been historically, but that's still at the sort of prudent end. So very comfortable with the turn of leverage.

So would you go higher?

It's not out of the question. Yes, we're certainly comfortable at a turn based on current assumptions.

And the package staircase change, so on the surface, the packages are still named the same and still essentially include the same functional benefits with most of those being in the base package. And the only difference between the packages above that is how much response that you get. The way that we deliver that now is heavily -- well, it's heavily driven by the relevance of the vehicle, which without saying the 2 letters too many times, is a model that sits behind that, that predicts the likelihood of someone to want to interact with that car. So that's kind of the most purest level that we do that. We then overlay that with a boost. So within a set of relevant results, you can appear higher or lower depending on what package level you have. There's a relationship between that, which is algorithmically driven and the price that we charge for that. So we're looking both at the levels of step-up between the staircase without going into too much details, there was one package in particular that wasn't quite a big enough step-up for you to be able to unequivocally see it as a retailer when you came on. It's part of the gig of running marketplaces that a retailer say, well, I spent more, but I didn't sell any more cars. You don't want that. You want it to be a bit clearer than that for packages.

Some of the other packages work really well. So they tended to jump right over the top of all of them. So evening up the staircase, the height of the stairs is part of that. And then tweaking the price that you pay for the uplift in the response. So that's kind of the thing that we've been feeding in over time to make sure that people are getting good value wherever you are on the staircase.

It's Jo Barnet from UBS. Two questions left with me, please. First one is actually a follow-up on Andrew's original first question relating to understanding the factors that are impacting stock. So clearly, one of the major driving forces on your stock lever is the supply and demand dynamics at play in the market. But I'm wondering to what degree other dynamics impacted stock as well. So for example, the churn in forecourt flowing through to stock, which is something we haven't really seen before. And also secondarily, are you seeing retailers listing a lower proportion of their cars on Auto Trader than we've seen before? So that's question one, just understanding what's impacting stock. And second one, probably for you, Nathan, you spoke a fair bit about AI product innovation, but less about how you're using it internally. So Scout24 last week showcased at their CMD that they're already implementing using it quite significantly and seeing headcount reductions as a result. So interested if you see other opportunities, how you expect headcount to move over the next 12 to 24 months, but also perhaps a comment on gross versus net headcount movements. Have you seeing an evolution of your workforce in light of AI styles of people that you're employing?

Yes. I'll take the first one. And sorry, I wasn't being deliberately evasive on Andrew's question. So I think it's fair to say the market dynamics are like speed of sales that was more of a headwind in fiscal '25 has been pretty flat in fiscal '26. It is running at historically fast levels. But when you think about the kind of year-on-year impact, like the slot utilization is more consistent over those 2 financial years. So it's not -- it's certainly not a headwind. Equally, I wouldn't go so positively say we're starting to see it slow down and there's a buildup of inventory.

Supply, I'd say, is probably getting marginally better. If you look at the kind of transaction volumes, used car transactions are about 1% better. The number of unique cars we've seen on site is around 1%, maybe marginally lower, the amount of live cars advertised on site, appreciating there were offers involved is up about 1%. So a little bit of better supply.

You get this funny dynamic though, where the kind of supply tightness that existed in the sort of 3- to 5-year-old category is now just sitting in a slightly older age cohort. So it sort of works its way works its way through the profile of customers. I would like to say that I would feel like there should be some very slight underlying positive dynamic towards stock getting better. So then the reverse of that is the negative stock lever, the worst run rate is down to the fact that, as you say, we've lost some customers who are clearly lifting cars, but then there's also some moderation. And I think we said it in the kind of presentation, there were some customers that obviously opted to leave in November and December. There were some that opted to moderate the volume of vehicles and some opted to moderate package. So that's ultimately the driver to the entry rate into the year, where the second half stock lever was and then like I say, improving from this point, but that's sort of where we've entered and the reasons why.

Yes. On AI internally, it was definitely not being evasive. I mean we've got a couple of Board members in the room, and this is something that we spoke about at a Board meeting 2 days ago. So yes, we are using it internally. I'll start with that. I'll come on to the headcount question.

We're focusing initially, partly is just the way that we work on engineering side of things. We take the general view in life is that if you build really good technology, the benefits of that will then flow through to the rest of the organization as opposed to having a free for all go and adopt an AI tool and knock yourself out, please make yourself more efficient.

I think there is a point in that, that is -- and this is no comment on Scout's business at all. It's very much for us. But people do forget that AI is not free. I mean the latest model that's just come out from one of the big providers is now 3x more expensive, I think, than the previous model. So there are productivity gains, but there are also AI costs that I think are not always being factored in.

It actually reminds me of the pitch around moving to the cloud, which I think is a brilliant thing to do, by the way. But the pitch was always move to the cloud, you'll save money, get rid of those old data centers. We're spending exactly the same money that we were spending when we're in those old data centers, but we can do a lot more, and we do it with better quality. So I suspect I wouldn't be surprised if from an enterprise perspective, that's how this plays out.

I think there is clearly big productivity gains. And you can choose to take those either as headcount reduction or greater execution around product innovation.

Now my preference and my belief is that we should be able to do the latter, but you've got to prove that over time. You never lose the opportunity to say, okay, well, actually, let's just take a bit of a balance of both those things, do more stuff, but also become more efficient over time. We're most -- we're furthest ahead on engineering. We have an agent development team that you can ask to do jobs. Still, those jobs are checked off by an engineer first because not every problem should be solved by an LLM, not least because token is very expensive.

So we've always been cost conscious. But then once the engineer says, yes, go, then the agent will go off build, it will do the pull. It will get signed off and then an engineer will just check in before they send it live. So the agents are doing jobs as we speak. It's very scalable, can utilize multiple models.

It understands because we did the cloud work, I talked about standardizing our applications, we're able to very easily tell the agents or make sure the agents have the right guardrails and understanding of our estate so they can solve very rarely can you solve a bug or a problem in one application. You tend to need to straddle multiple applications, and that's where these things can stumble. So that's some of the hard work that we've done that's really good, and we're very, very proud of it. We're a bit behind in some of the other areas of the organization because we're doing the engineering first, but we're starting to see product managers be able to post jobs, release code, make simple changes that used to take -- used to never generally get done actually. The small product requests normally get put on the backlog and no one ever gets around to them.

We're using it in marketing, both for performance marketing. We've used it a bit for creative development as well, but it's fair to say our people weren't overly supportive of AI executed creative, but we'll continue to experiment with that.

We use a massive number of -- we basically throw thousands of creatives when it comes to performance marketing. We're able to very quickly work out which ones are working, which ones aren't and hone those further.

The other areas where we're focusing is operations and partnerships, which are probably the equivalent size of product and technology, so around 400 people. The jobs that they're doing, the systems are not as good as the systems that our engineers use. We're cleaning up the data models underneath. We're putting in a new CRM system, which will then enable us to put artificial intelligence on top of that to make their jobs way more efficient, whether that's diagnosing a customer's problems, whether that's answering the really easy queries like check my invoice. I think there is quite a bit of opportunity, but there's a bit more work for us to get to before we can really make that matter. So yes, from a headcount perspective, to be really explicit, I think at the moment, we're very comfortable with the margin profile. There might be some reinvestment of that productivity over time, you might see it change, but we wouldn't want to make a call on that yet. Measuring engineering productivity is a controversial topic, I think.

Sean Kealy from Panmure Liberum. Just a couple to finish off. I think what you've tried to do with Deal Builder over the last couple of years is get a bit more into the transaction, maybe start to monetize that a bit more directly.

Given the backlash, is that still something you want to do longer term, maybe start to think about adding financing and other bits into that journey, especially given I think you took a few parts out when you put it into the packages. And then sort of linked to that, how do you think about the trade-off for sort of remaining a lead volume platform versus sort of pushing into lead quality more deeply going forward?

Good final questions. So on the first one, -- we needed to listen to dealers, don't let me say anything other than that. But the Deal Builder feedback was actually relatively small part of that. It was a pretty challenged profitability.

You can go on to our own website, type in Auto Trader Customer Advisory Group, you can see what they actually raised. And I did check it this morning. We didn't even have Deal Builder down as one of the points, and that was something that was curated by the group themselves.

So I think we needed to change some things. We needed to ease up on the implementation of it because all of us humans were all right to impose change on others, but none of us really like change ourselves, and we experienced that in a big way.

So I don't think our fundamental views of anything relating to Deal Builder have changed as a result of what happened on November.

Our implementation of it just needed to be different to what it was.

In terms of monetization, finance, I think generally, we've put it in the packages because we see that it is much better for consumers. They really like it and retailers once they're on it, they really like it, too. That feels like it makes the moat, the business that is Auto Trader so much stronger, and that's worth more of our investment than trying to monetize a stand-alone revenue line, which was a decision that we took last year.

I think there are still opportunities within like the bucket of digital retailing finance, as you've called out, as being the obvious one, potential different executions of part exchange also being another one, 2 of the most profitable elements of a transaction for any retailer. And the truth is back to the answer on private, where most of the volume that Carwow, Motorway, most of the volume they're getting has been from part exchanges, from retailers. The average rate has probably gone from 45% to 50% down towards 35% to 30%. So they could really use help there. Are any of those things at the top of our list at the moment? No, not just at the minute.

We're focused on getting dealer embedded, getting those reservations and deals up and running beyond the -- at the moment, it's about 10% of their sales. We want to get that up and up and up once we've got it available across everywhere. That's really our focus for now. We've got plenty of other stuff to do around the Co-Driver products, the advertising products. So we've kind of prioritized that stuff and focus more on rollout.

And your -- yes, your second question is actually a really, really good one. And it is a debate that we have both internally and with customers as well. It is clear that it would be better not -- we sell probably 4 million, 4.5 million, I think, million trade vehicles a year or go through the platform and we influence the sale. But that number is not too important, but we send 15 million inquiries and we send a lot of walk-ins in as well. There is clearly an inefficiency there. And anyone with a mathematical or economic minded as all of you are would say, well, hang on, if I can not get 4 things, I can get and 1 and I get the same number of sales, and that feels like a really good place to be.

Getting our retailers into that place. Some retailers are definitely there, but getting the broad base of retailers into that place is a journey that we have to kind of take them on kind of sequentially because some will say, well, it's only me and I've only got 5 cars, and I only get 20 leads and I can make 2 phone calls a week. I'd rather have a go at all of them than I would you to send me the 5 that buy a car. So it's just working through that over time. And I think our approach to Deal Builder is really helpful like that because they start with getting maybe 1 in 10 sales coming through Deal builder. They start to think, oh, I like that. I'm going to pay more attention to these and they get to 2 out of 10 or 20% of their deals are coming through, and they tend to find their way to that conclusion themselves. That's basically the way that we're rolling out Deal Builder, but it's a long one.

Great. I think that's all the questions that we have from the floor. So thank you very much for joining us.

Good morning, everyone, and welcome to Auto Trader's results for the 6 months ending 30th of September 2025. I'm joined by our COO, Catherine; and our CFO, Jamie, who will both be presenting and joining me for Q&A. You will have seen the announcement regarding Catherine moving on to become CEO at Moonpig. We are very pleased with Catherine and wish her the best as she embarks on the next chapter of her career. Catherine has been a pleasure to work with and has had a real impact at Auto Trader. She will be missed, but she leaves behind a strong team with bench strength that is both broad and deep. So we have a little worry that we will carry on uninterrupted.

Our disproportionate focus on internal development over external hiring serves us well at times like this. Catherine is still with us for some time, and we will announce her leaving day in due course once we have worked through a smooth transition plan.

Now on to the results. Overall, we are pleased with the progress that we've made through the period. Our profit was marginally ahead of expectations, and we have made significant progress against our strategic priorities.

As expected, we have been impacted by the fast speed of sale of vehicles, but the team has delivered well on those areas that are within our control. I'm confident that the actions we are taking today will underpin growth for many years to come.

Our market position remains strong with record levels of buyers and retailers using Auto Trader. As we saw last year, there has been a further increase in the number of unique vehicles advertised on our platform and high levels of engagement with those vehicles. This underpins our core proposition for buyers as we provide full choice and transparency and clearly, retailers have benefited from more vehicles moving through a similar number of advertising slots.

Our annual product and pricing event in April this year went well, underpinned by the first features under our Co-Driver AI umbrella. There is significant future potential in this area as we make more AI-powered solutions available and accessible to both retailers and car buyers. We are uniquely placed to do this due to our strong brand, deep integrations with the industry, and our real time proprietary data. The first features of Co-Driver have seen strong engagement with over 10,000 retailers already using the tools.

It's important to note that AI doesn't just enable us to increase the richness of the car buying experience on Auto Trader, but it does extend the opportunity across all our retailer product areas which includes advertising, data, digital retailing and now Co-Driver.

We've also dramatically accelerated the adoption of Deal Builder, growing customers, stock and deals since our change in approach earlier in the year. Deal Builder will no longer be an optional add-on product available on a selection of cars, it will be the default experience for retailers and car buyers on Auto Trader. Catherine will cover this in more detail later.

With a car market that will grow over the long term, our strong market position, we're comfortable that we can continue to grow through delivering meaningful improvements to the buying and retailing of cars in the U.K.

Auto Trader has never been short of growth opportunities, which remains as true today as it has ever been. I wanted to thank everyone at Auto Trader and our customers, shareholders and wider stakeholders for their continued trust and support.

We'll start with some of the highlights during the period. Group revenue grew 5%, operating profit grew 6% and basic EPS grew double digit at 11%. Our largest revenue area, retailer revenue, grew at 6%. This is made up of strong forecourt numbers and a 5% increase in ARPR, mostly driven by the price and product event in April 2025.

AI has been a big focus for many investors recently, given its potential to significantly alter consumers' online behaviors. I'll speak to this in more detail later. However, we are confident that on any platform, we are well placed to provide the best car buying experience for users.

The transaction is high value, complex and occurs over a 3-month period, not one session. The reasons car buyers choose Auto Trader come from our singular focus on the U.K., our category-defining brand, well-invested technology and the rich tools we provide both car buyers and retailers, which are all made possible by retail -- by real-time data, pardon me, at a vehicle level that only we have access to. These unique characteristics are why our market position has been not only maintained but strengthened when new platforms have emerged such as Google, iOS and Android.

Now to Deal Builder. I am very proud and pleased with the progress that we've made since we changed our approach midway through the year on Deal Builder. Adoption has dramatically accelerated as we make this journey the default experience on Auto Trader. We've added 4x as many retailers this half than we did in the previous 6 months.

We know from years of development and live testing that Deal Builder deepens our engagement in car buying and selling. It delivers better conversion for retailers and a more connected and empowered journey for time-poor car buyers who want to do more online when it suits them with a brand they trust.

At our full year results in May, we presented this slide for the first time to better show how market dynamics not previously seen before had impacted our financial results. We have 4 charts here, which I don't intend on going through in great detail, but it is a picture of a more stable market. The key points to note are, last year, demand or visits were strong. Supply was constrained, used car prices had come down, which all resulted in an acceleration in speed of sale. This meant more unique vehicles sold through roughly the same number of slots, which doesn't benefit our business model.

This year, however, demand does remain healthy. Supply is gradually coming back and used car prices have been robust, even increasing from the levels seen last year. As a result, speed of sale has not accelerated as it did last year. So the headwinds we were facing have subsided somewhat. However, we would caution too much optimism in the near term as speed of sale was still one day quicker in October.

I'll briefly cover the financial results, which Jamie will cover in more detail next. Group revenue increased by 5%, with core Auto Trader revenue also increasing by 5%. Group operating profit increased by 6%. Auto Trader operating profit increased by 5% to GBP 208 million. And Autorama halved its operating losses to GBP 1.4 million.

Noncash acquisition-related costs was GBP 6.5 million. Group operating profit margin increased to 63% and Auto Trader's operating profit margin remained at 70%. Basic EPS, as mentioned earlier, grew double digit at 11% and cash generated from operations was up 7%. We returned GBP 162.2 million of cash to shareholders through GBP 100.2 million in share buybacks and GBP 62 million in dividends. Finally, we are declaring an interim dividend of 3.8p per share.

Now on to our operational results. The average number of cross-platform visits was up 1% to 83.3 million per month, and we continue to account for over 75% of all time spent across our main competitor set. The average number of retailer forecourts advertising with us was up 1% to 14,080.

Average revenue per retailer was up 5% to GBP 2,994, mainly due to our product and pricing event implemented on the 1st of April 2025, which was underpinned by Co-Driver.

Live car stock was up 2% to GBP 457,000, with this increase being due to an offer, which ran at the beginning of the 6-month period, and we delivered 3,687 new lease vehicles. Finally, the average number of full-time equivalent employees in the group decreased slightly to 1,249.

In previous years, we would have covered our cultural KPIs in half year results. However, we've decided to do this now at each full year results. The KPIs and initiatives that sit behind them remain as important as they've always been. However, covering them annually better aligns with our annual employee survey. The KPIs on gender, ethnicity and GHG emissions are all included in the press release and the appendix of this presentation.

As it relates to culture, it is worth noting that we have entered a new lease and are halfway through investing significant capital in a new home campus for Auto Trader, which will provide a great environment for our people to do their very best work in a space that facilitates both how we work and the other elements of our culture.

I'll now hand over to Jamie to talk us through the financials in more detail.

Thanks, Nathan, and good morning, everyone. I'll start by focusing on the core Auto Trader financials. Starting with revenue. Total Auto Trader revenue increased 5% to GBP 296.3 million. Trade revenue increased by 6%, with the largest component of this being retailer revenue, which also grew by 6%.

Also within trade revenue, we've seen an increase in both Home Trader and other trade revenue. Consumer Services revenue decreased by 9%. Within this, private revenue generated from individual sellers was down year-on-year due to a lower number of listings compared to a strong prior year, and Motoring Services revenue was flat.

Revenue from Manufacturer and Agency customers increased 13% year-on-year due to manufacturers supporting their franchise networks on both new and used car advertising.

As mentioned, Retailer revenue increased 6% year-on-year. The average number of Retailer forecourts on our platform increased to 14,080, a 1% year-on-year increase and average revenue per retailer increased by 5% to GBP 2,994 per month, with more detail given on the following slide.

Here, the chart on the left shows the components that contribute to the movement in ARPR compared to the prior year. As you can see, ARPR growth was driven by the price and product levers with a small headwind from stock.

We delivered our annual pricing event for all customers on the 1st of April 2025, which included additional products and a like-for-like price increase, which contributed GBP 89 to ARPR growth.

Product contributed GBP 64. Most of this growth was from our Co-Driver product, which is included in retailer advertising packages in April 2025. Prominence, which includes upsell to our higher-level packages, was not a contributor to the product lever in the first half.

We continue to review our package staircase and have recently created an offer to incentivize customers on to higher levels, which has had good levels of uptake. This offer converts throughout the second half and will inform how we evolve these packages in H1 of next financial year with the aim of returning prominence to long-term growth.

Turning now to stock. You'll see on the right-hand side of the chart that the number of live cars advertised on Auto Trader increased 2% year-on-year. Used car stock also increased by 2%, although much of this was driven by a stock offer, which we ran at the beginning of the financial year. Excluding this stock offer and private listings, which do not impact ARPR, the live stock increase was just under 1%. The stock lever was marginally lower due to a slight reduction in underutilized slots, which typically occurs when we run this type of offer.

Total Auto Trader costs increased 3% to GBP 90.4 million. Salary costs increased by 6% to GBP 42 million due to higher average salaries and a small increase in the number of Auto Trader FTEs. Share-based payments increased by 1% to GBP 6.9 million.

Marketing spend decreased by 21% to GBP 8.9 million due to the timing of campaigns, and we expect a greater level of marketing in the second half of the year. Other costs, which include data services, property-related costs and other overheads, increased 6% to GBP 22.9 million, primarily due to property costs for our new head office and other IT-related expenses.

Depreciation and amortization increased by 31%, again, related to the cost of our new head office. As a reminder, we fully expense our technology, research and development costs, hence, our low levels of CapEx and depreciation.

In addition to our investment in cloud-based and AI services, we have around 400 people in product and technology who are continuously improving our platforms and developing new products for consumers and retailers.

Operating profit increased by 5% to GBP 208 million during the period, and operating profit margins remained consistent at 70%. Our share of profit generated by Dealer Auction, the group's joint venture, increased 17% to GBP 2.1 million.

Having covered Auto Trader, the main part of the group, we'll briefly cover Autorama results. As a reminder, the Autorama acquisition is part of our strategy to bring attractive new car offers to car buyers on Auto Trader and to make new cars a more important part of our proposition.

Autorama revenue was GBP 21.4 million, with vehicle and accessory sales contributing GBP 16.5 million and commission and ancillary revenue contributing GBP 4.9 million. Vehicle and accessory revenue relates to vehicles that flow through our balance sheet, which is not our focus for future growth.

Total deliveries grew 16% to 3,687 units. As can be seen from the chart, this growth was driven by cars and importantly, more of that growth was driven by the Auto Trader platform, which saw a 6x increase in delivery volumes.

Average commission and ancillary revenue per delivery decreased to GBP 1,329, reflecting the changing vehicle mix during the period. We delivered around 750 vehicles, which were temporarily taken on balance sheet, the cost of which was taken through cost of goods sold. This was a year-on-year increase driven by just over 300 extra vans, which were taken to support van volumes as they were slightly lower in the first half.

Excluding the cost of goods sold, cost of GBP 6.2 million represented a 25% year-on-year reduction with all lines seeing a decrease. The Autorama segment made an operating loss of GBP 1.4 million, which is a significant reduction on last year as a result of the accelerated integration into the main Auto Trader business and platform.

With group revenue up 5% and a reduced Autorama loss, we saw group operating profit increased 6% to GBP 200.1 million and group operating profit margins increased to 63%. As we grow, the strong cash generation of our business leaves us well placed to return surplus cash to shareholders. Cash generated from operations was at GBP 215.4 million.

Now to briefly review net bank debt and capital policy. During the period, the group drew down GBP 15 million of its revolving credit facility and held cash and cash equivalents of GBP 20.2 million. Cash generated from operations was largely used to pay tax or return to shareholders through a combination of dividends and share buybacks.

The group's long-term capital allocation policy remains unchanged, continuing to invest in the business, enabling it to grow, while returning around 1/3 of net income to shareholders in the form of dividends. Following these activities, any surplus cash will be used to continue our share buyback program and to steadily reduce gross indebtedness.

That concludes the financials. I'll now hand over to Catherine to talk through progress against our strategic priorities.

Thank you, Jamie, and good morning, everyone. We have made good progress against each of our 3 strategic focus areas. These areas are closely interconnected. Our platform and our digital retailing capabilities build on the strength of our marketplace and deepen our relationships with both retailers and car buyers.

Our marketplace continues to grow, and we have seen a record number of car buyers and retailers using Auto Trader. This means we are also building our unique data advantages through the growth in observations and actions that we capture. Whether it is consumer behavior and interactions or retailer actions and pricing movements, we continue to extend our data lead in this area.

We have successfully executed our annual pricing and product event, which included the Co-Driver product, a set of AI-enabled features designed to drive retailer performance and efficiencies in the advertising journey. This product has seen strong engagement from retailers and the features that surface on the Auto Trader app and website have been well used by buyers.

We continue to scale Deal Builder, enabling consumers to do more of the car buying journey online. At the same time, we are launching the Buying Signals product for retailers, which will provide a greater level of actionable insight to drive their performance.

This year, we launched Co-Driver, a suite of transformational AI tools that utilize our unparalleled vehicle data and consumer insights to significantly improve the consumer and retailer experience. Our first Co-Driver suite is available to all retailers and includes Smart Image Management, AI-generated descriptions and vehicle highlights.

As of September, over 100 retailers have used Co-Drivers to create 1 million high-performing used car and van adverts to optimize over 12 million vehicle images, and we've seen over 85 million buyer interactions with the vehicles highlights on Auto Trader. Whilst Auto Trader has been working with and delivering AI products for over 10 years, Co-Driver is the first retailer product, which has leveraged generative AI.

As detailed in our FY 2025 full year results at the end of May, we decided to make Deal Builder part of our core proposition to retailers and the consumer experience for car buyers. This will enable us to increase the speed of retailer onboarding, accelerate the level of buyer adoption, materially increase the number of deals being delivered through Auto Trader and strengthen the competitive moat for our core business.

As can be seen on the right-hand side chart, this decision has enabled us to scale the product faster from June onwards than in previous periods. Retailer acquisition during the period was 4x greater than the preceding 6 months, resulting in 4,000 retailers live with the product at the end of September.

We also saw a significant increase in the volume of listings with Deal Builder, ending September with 128,000 adverts live. This was over 160,000 live adverts at the end of October, a 25% increase in the month.

Consumer engagement has also grown considerably with 52,000 deals in the period compared to 23,000 in the previous year. The feedback on the product continues to be positive from both retailers and car buyers with deals converting twice as effectively as a regular Auto Trader lead and over half of all deals being submitted outside of traditional working hours.

We are also launching a new product called Buying Signals, which leverages our unique consumer data to surface both high-intent buyers and their preferences to our retailers. Across multiple inquiry types, we have used an AI-powered buyer propensity model to apply a flag for the retailer, indicating how likely the buyer is to buy the vehicle, how local the buyer is and the type of vehicles that they are interested in. This will enable retailers to prioritize the next best action with different car buyers.

The goal of this product is to drive improved conversion for retailers and to close the gap between the journey on Auto Trader and the consumer experience that the retailer forecourt, complementing the Deal Builder journey. Over time, these buy propensity models will also inform our own marketing, remarketing and optimization activities for our products and experiences.

I'll now hand back to Nathan to discuss the broader AI trends we are seeing and our outlook for the remainder of the year.

Thank you, Catherine. The popularity of LLMs, chat-style interfaces and agents is at the forefront of investors' minds as it relates to most businesses, and that includes marketplaces. For years now, we have used AI technology in our in-house products and platform, which informs our perspectives on this technology shift.

Before we talk about Auto Trader, it is worth saying that we believe top-of-funnel research and discovery for all products, including cars, will be disrupted by these new interfaces. AI platforms reduce the need to visit multiple websites and summarize what is essentially static content very effectively.

Top-of-funnel content has never been a big focus for us. And when we have experimented with it, the direct benefit to our core was unclear. For this reason, it is not a risk that concerns us as we wholly focus on the point where people want to browse and purchase real available inventory.

In terms of Auto Trader then, we have 4 observations to make. Firstly, the opportunity to use AI to enhance the car buying experience and the tools we provide retailers on Auto Trader is clear and something that we have been doing for a long time now.

The recent developments in AI technology, particularly LLMs, provides even greater runway for this across our advertising, data, digital retailing and Co-Driver product streams as well as our consumer experience.

Secondly, brands really matter. Car buying is an incredibly complex, high-value purchase. It is almost never online only, typically involves multiple transactions, is regulated and usually takes place over 3 months, during which the selection of vehicles changes constantly.

To navigate this, people use Auto Trader. Over 75% of marketplace activity happens on our site and most people come directly to us. 49% from apps, 29% from our URL or searches for Auto Trader, 18% from organic search with only 4% being paid for web traffic.

This brand position doesn't just come from a trademark, it comes from the deep and rich experience we provide car buyers. It's not just about a wide selection of vehicles, it's about making sure that selection is real, available, described to a high quality, easy to navigate, comparable and not fraudulent. But this is just the listings. People need a lot more than just listings.

They need a lot of high-quality real images, comprehensive and accurate descriptions, price flags, dealer ratings, valuations, checks on the vehicle, part exchange quotes, finance and so on. This is why people seek out Auto Trader, and we believe these tools will continue to be important to a car buying transaction moving forward.

Thirdly, as these platforms grow, we will ensure new and existing users of Auto Trader can reach us there. We've taken this high-level approach in similar situations in the past, and it has served us well, whether it was the rise of Google when fears of disintermediation were raised, iOS or Android. In all these cases, our market position strengthened and our audience grew. This is because these platforms grow by providing the best experience to their users. For the reasons mentioned earlier, when it comes to cars in the U.K., that is what we do. There will be a myriad of technical, commercial and strategic decisions along the way, including the depth of experience and protection of our data, but these are not new decisions to us.

Finally, we are confident that our deep real-time vehicle data and rich tools for car buyers and retailers will remain essential to what is a large and complex transaction. The vast majority of that data is not available on the public Internet. Agents may, over time, provide users with automated or semi-automated assistance for carrying out varying tasks on the Internet, but they too will require sources of high-quality real-time data, where they'll face similar constraints to search engines. In fact, most of these interfaces for that sort of data use the search engines we know today.

For that reason, we expect that AI agents, like other client technologies, will either remain top of funnel and generalized or they'll look to provide direct integrations using standard web technologies customized for their environments. Interestingly, this is exactly what ChatGPT recently announced with their Apps SDK, and we expect others will be soon to follow. These integrations allow greater control over the experience and data that we provide such that it can bring the best of what LLMs do together with what we do best, providing another way for users to find and engage with Auto Trader. Again, there will be many decisions to make along the way, but we feel very well placed to make those.

Finally, we know the landscape will evolve, and you can have confidence that we will stay abreast of these changes. We'll continue to be fully engaged in the technology and we'll maintain the ability to move both strategically and quickly when required.

Now on to the outlook or not. Right, on the outlook, there is actually not a great deal to say as those of you who have read the announcement are aware, other than that the first half has pretty much played out as we expected. So our outlook for the remainder of the financial year 2026 remains the same as it was at our full year results. All that for that short sentence.

Right. We'll now move to the Q&A, which Jamie will manage, and we'll take questions from analysts in the room.

Yes. So we'll wait the mic and start down the front and then work our way back as usual.

It's Will Packer from BNP Paribas. Three questions, please. Firstly, thank you for the very useful comments on AI and how you're positioning yourselves. Could you help us think through investment requirements in the next 12 to 24 months. Should we interpret your comments as you're well invested and you -- the kind of formula we've seen in recent times of flattish margins or slight expansion depending on the top line is the right formula? That's question one.

Secondly, could you help us think through the dynamics around the integration risks and opportunities with ChatGPT? Am I right in thinking that you have a choice in that you do have a significant inventory lead versus your nearest competitor. Some of the classifieds don't, so you'd think the hand is more forced, you can choose. And in the event that you do decide to integrate, how should we think about the split of economics versus the current status quo? My take would be you're not paying very much to Google compared to some other segments. So is there a risk that the economics deteriorate in that environment?

And then finally, you've got a prominence offer. That's something which is -- we haven't heard too much about in the past. Could you think us through -- could you help us think through what that means for the upside on prominence and how that will flow through in due course?

I can take the first one. So as both Nathan and Catherine mentioned, we've been investing in much of this technology for a long period of time, previous product iterations and most recently, Co-Driver. So I think there's still much work that we can do from both retailer products, consumer experiences, tools internally to help us find greater levels of efficiency and productivity. But you're absolutely right, I think we feel like because these investments have been happening for a long period of time, there's no -- certainly in that 18- to 24-month window you mentioned, no change from a guidance on margin perspective, consistent margins in the Auto Trader segment. I still think we believe at a group level with Autorama, profitability or losses improving and hopefully into profitability, the group margins can actually continue to expand.

And on the second question around the integration risks and opportunities, I think that there's probably a few things that I'd say. As I said, we think that they'll go for reasonably structured integrations. The idea we've heard and seen some notes about talk of them scraping the Internet to get real-time data. We just don't think that's going to happen. I mean that technology is very, very old and nonperforming. And indeed, ChatGPT's SDK suggests that they're going to go for something more structured.

What comes with that structure is a pretty great deal of control and transparency about how your data is used, what depth of experience that you provide. So you're able to manage the risks and opportunities as it turned out when we've been with Google, we've made decisions to open up our site to a certain level, but not necessarily fully. There will be those sorts of decisions. When it comes to iOS, we open up everything, but it is within a native app. So when I talked about the strategic commercial tactical decisions, they tend to sound very, very technical, but they are quite important.

The SDK has not launched in Europe yet. So we haven't had a detailed look at exactly what that looks like, but we would go for something more structured, and we suspect they would as well because those platforms, and I think this is where the opportunity is, and it relates to your economics point is Google only became really, really successful because it provided and prioritized the most relevant results for users. That's going to be the true for any interface. So we think we can do that for car buying. And we think for that reason, they'll want to work with us.

If they were to compromise something like that for the sake of some form of rent that they collect, that would seem to be a bit inconsistent with the pattern that has played out with these platforms. But does that mean around an app that we do, there might not be -- might be paid positions or there will clearly need to be some economics. Yes, we would expect that to be the case. But again, that is no different to Google. And by and large, most people come directly to Auto Trader. I think that is also the point that we're not getting much of our traffic at all about what, 18% plus 4% paid. So around 20% of our traffic is coming through those more generalized search engines.

Most people wanting to buy a car kind of know where they need to do that. And I suspect that will still be true in the future. What we hope though is as more and more users start to use these interfaces, actually the use case for Auto Trader can appeal to people that perhaps might not have otherwise found us, that might have found the search by make and model a little bit intimidating and ChatGPT and those other kind of tools can hand off into a structured search like us, which feels like an opportunity that we don't necessarily have today. Of course, we can do it on our own site, but that is only for people that are coming to Auto Trader.

Dominance and offer. So I think we talked at the full year results about how we were doing a lot of work and imagining that in the next year or so, we would look to evolve the packaged staircase. Again, typically, we've done that every 3 to 4 years. And we're coming to a time again where we think that is the right thing to do. So you're right, we are in market with a bigger and more attractive prominence offer than we would typically have run in the past. It's had good uptake from retailers. And over the coming months, we'll be in the process of converting those retailers through to fully paid.

One of the reasons for making the offer a bit bigger and more attractive is really to learn and to test the value response uplift that we're seeing and the different levels of retailer adoption, all with the view that it will inform the structure and the makeup of the packages that we look to try and then roll retailers into at some point during next financial year.

Gareth Davies from Deutsche Numis. Two from me. The first with a couple of parts on Deal Builder. 2,000 onboards since, I think Catherine emphasized, June. But just kind of understanding how that's built up, should we assume it was kind of pretty straight line through that period? Or were there any sort of stumbling blocks initially that you've got through? And has that ramped into August, September?

And then I think you said 25% increase in adverts in October. I mean, can we be as simple as thinking that means we're up to 5,000 by the end of October? Or is that being too simplistic?

And how are you feeling sort of overall in terms of getting everyone you need on Deal Builder by the sort of March, April time line you need? So that's question one.

And then the second one, you confirmed guidance for the year just in terms of the minutia. Can you talk a little bit on stock and a little bit on dealer forecourts because I think stock feels that it's sort of stubbornly at 28, 29 days. How are you feeling on that at the moment? And then dealer forecourts feels like it's running a bit stronger than I certainly expected.

Yes, sure. So on Deal Builder, we have been talking on webinars and in the trade press about being around 6,000 retailers now and about 160,000 or so live adverts. So those numbers are -- they're out there. You talked about whether the growth has been linear or not, I think we've talked before about looking, particularly for the independent retailers that work through our portal system, we've been onboarding them in waves. So it's definitely not been a linear line from June through to October. There's been waves of retailers. We've defined cohort segments that have similar attributes or similar ways of working with us and then have been onboarding them in a more scaled way than we were able to do prior to June.

We're getting to the point where we are a good way through all of the independent retailers that work through our portal system. And so growth from this point onwards will be more influenced by the technology API integrations that we're putting in place with the tech partners out there in the automotive industry. So we'll continue to see, I think, a slightly inconsistent patterns of waves when we complete a tech integration with a dealer management system partner, suddenly a new cohort of retailers will become addressable, and we'll look to get those onboarded pretty quickly.

So I imagine it will continue to be quite lumpy between now and March. We're hopeful that we will have made really good progress by March. I imagine, as is typically the case with the integration work that we do, I imagine we will have a tail of retailers that will need to work to get over the line beyond March, driven principally by that tech integration work, not work on our side, but work for the third parties integrating with the API that they will need to do.

And on the more detailed kind of guidance for stock and forecourts, I mean I think we've been pleased the stock has improved through the half. I think at full year results, we gave the April number for the stock lever, which is sort of materially down and then obviously only marginally down for the first half. We haven't quite got into positive territory. So September was still marginally negative. And I think we are -- don't -- are probably a little bit cautious on just what the outlook looks like for these remaining 5 months or so. The fourth quarter, the first quarter of the calendar year is always slightly volatile. January is generally a very strong sales month, and it's not always easy to source stock. So I think that's why we're sort of holding that guidance at marginally down for the year.

Similarly, forecourts almost sort of shown an opposite trend where obviously, the stocks got better. And if you look at the growth rates on forecourts, it is -- I think we're pleased that it was as positive as it was in the first half, but the growth rates are trending down. So we've exited the half slightly lower than the 1% growth we delivered in the first half. And again, I think in the round, holding that guidance of flat forecourts seems reasonable.

It's Jo Barnet-Lamb from UBS. Firstly, a couple on sort of product-driven ARPR into next year. So firstly, on Deal Builder, you've obviously giving it away for free at the moment. I think you'd articulated previously, you're then going to sort of do an upsell sort of thing through next year, and that sort of, therefore, becomes a tailwind for product. So could you talk a little bit about Deal Builder into '27? You've probably got a more formulated views as to how you're going to do that. So any more color you can give us there would be great.

Then secondly, on Buying Signals, which you're sort of -- is sort of being rolled out at the moment. And I think you said you're going to start commercializing that in H2. Any more color you can give us on sort of the scale of tailwind that, that's going to give product in H2 would be great?

And then a final one. There's something in the release relating to a property -- Autorama property sale. Is that right? Can you give us some color on what that's about? I presume it's just getting rid of an old Autorama building, but any color you can give us there would be great.

So Deal Builder and Buying Signals and how and when we'll look to monetize both of them. You will have seen how we've positioned and talked about the product is that the 2 very much come hand in hand. So as part of Deal Builder, we are evolving, I guess, the value currency that we use to talk to retailers and evolving that to very much be anchored around the deal. And the positioning for Buying Signals is that you get all of this insight, rich insight about the buyer, their intent to purchase that vehicle, their preferences that they've been looking at and engaging with on our platform. You get all of that rich data as part of the deal. So they have become really how -- the combination of the 2 products has become how Auto Trader works for retailers.

We're looking to monetize certainly the first wave of both of them because I think they're both products that have multiple iterations and life cycles for the business as part of the rate event next year. And we have -- we've been pretty open when asked by retailers in forums and webinars and have been talking about that being the case. So first wave of monetization likely to be from April next year for both combined as a package for retailers.

And that will be as part of the pricing, but it won't be tiered. It will be a sort of bundled.

Yes, very likely to be part of the overall rate event for all with no tiering.

Yes. And just to add to that, because you're asking about the second half, whether there's any second half product. The second half product I think where consensus is slightly higher is really all coming from prominence and that conversion of the offer is where that kind of product lever growth comes from.

So just on the building in Hemel Hempstead, I'm delighted someone's made it to the notes in the back of the account, which might be your first questions for me. So when we acquired Autorama, they owned the building in Hemel Hempstead. You will have noticed from the accounts and the FTEs that we report that as we've kind of integrated into the main Auto Trader business and platform, the FTE numbers come down.

We weren't actually -- weren't actively looking to market the building at the time. Opportunistically, someone came and said through an agent that they were looking for property space and just made sense because the building is probably bigger than we require. So we've taken a space almost next door that's smaller and fits better for us. It's just on a lease basis, and we're obviously then disposing of that asset, which I think is likely to go through in the next couple of weeks.

It's [ Kieran Darling ] from Citi. Firstly, maybe just on -- could you give us your thoughts on kind of the pros and cons of the stock-based offering you guys have at the moment? Has there been any internal debate around rather moving to an all-you-can-eat model makes a lot of sense, particularly in the context of, I guess, underlying retailers are becoming more technologically efficient and innovative and therefore, maybe that's a headwind permanently?

And two, I guess, just in terms of OpenAI and a potential launch of a competitive app or I mean, could you just break down in terms of your visits, how much comes through the app versus desktop and just how much of a moat that is for you guys?

And then thirdly, I guess, just in terms of speed of sale, how should we think about it going into next year in terms of comps as it gets easier, how much of a potential tailwind could that be for you guys?

Yes. I'll take the first one, and Nathan can manage the second one. So I mean I think we said this at the last set of results. Obviously, the slot-based model where speed of sales has been running quicker has generated this small sort of headwind. And I think we have been doing an exercise and looking at other charging models. And obviously, we're fortunate enough to have a number of peers and everyone seems to have slightly different variations and nuances.

All you can eat is always a slightly more challenging one because the nature of retailer customers is you have some customers with 4 to 5 cars and up to the biggest customer on an individual site will have 4,000. So that -- not completely insurmountable, but that makes it a little bit more complex just to run pure all you can eat.

But I think we have been doing an exercise of looking at unique listings and there are many different kind of variants that you can do. And so we have been doing that work. I think at the moment, especially as the kind of speed of sale and market headwinds are not as prominent right now as they were this time last year, we still think that if you get supply easing up a little bit or speed of sale staying flat year-on-year or slightly decelerating, that should be positive with the charging model that we've got. But it's not lost on us that, you don't just want to sit there and say, well, everything will be fine, it will come back. So we are doing the piece of work. And I think it's not something that we would never consider. But I think at this point in time, especially where we are in the sort of cycle, we're reasonably comfortable with the model that we've got.

Kieran, I got the second bit of your question. The first bit around OpenAI and competitors, can you just, sorry, go through that one again.

I think it's a more general point around potential competitors coming in terms of utilizing OpenAI technology.

Yes. Right. No problems at all. So if I take your first question, I think OpenAI is another window to the Internet that uses kind of a highly efficient text prediction to kind of summarize answers and give people the next step that they might want to go to. And when it was Google, it's all around a page rank algorithm.

What we found with Google is that their desire is to provide high conversion rates to their users to satisfy them to give them relevant. So Auto Trader tends and over time as SEO and those new releases have gone in, Auto Trader has tended to just do better and better and better. And that's not really down to our own SEO activities, although that's clearly part of it. It's because they want to get around people being able to gain those systems to just give users what's the best answer for the task.

So our biggest protection, I think, is the fact that with that depth of data, our brand, people look for us, but it's not -- it isn't about the trademark, it's actually around what people know that they can get there, and it's all founded on data. So I'd say that's probably our biggest defense is we don't see a world where we really feel like we'd be threatened in terms of providing the very best car buying experience. You've got to believe that people will be willing for some degradation to never ever come to Auto Trader.

I think apps have always been a really big strength to us. And I think we've always said that most of our traffic, as I laid out, about 80% of it is coming direct to us. The difference between traffic coming direct to our URL and apps is it cannot be intercepted. It is literally a direct connection. So it is about half of our visits, probably an even bigger percentage of the activity that you see when you go a bit deeper into the funnel looking at vehicles. And yes, I mean, it's an area we're always going to invest in.

Some people might say we always talked about 10% of marketing being -- 10% of our audience coming from marketing. And I mentioned before, the difference is actually marketing with apps. So we do that very actively because it's a different sort of marketing. So yes, we think it is a big strength. But at the end of the day, we've just got to be the best place to buy a car, and that's hard to do in the U.K. because it requires loads and loads of data, and we've got a lot of that and other people don't.

We have a question down here, Giles.

It's Giles Thorne from Jefferies. First question, I guess, for Catherine, Buying Signals, was that always part of the product road map for Deal Builder? Or is it something that was introduced or accelerated when you changed your commercial approach?

Secondly, coming back to this idea of the April 2026 pricing event, how transformational would you describe Deal Builder and Buying Signals is for your customers, for retailers?

And thirdly, maybe back to Nathan and perhaps you're going to reference again some of your prepared materials. But as you've seen this agentic AI debate suddenly materialize in a very quick and aggressive fashion, which elements of it do you think are most misrepresented, misunderstood? I don't know, you tell me.

I take the first one. So on Deal Builder and Buying Signals, Buying Signals is built, the product was enabled because we've, for many years, been investing in building a buyer propensity model, which takes all of the sales observations data that we get from retailers, takes all of the consumer interactions and observations that we see on Auto Trader and then looks at how you connect those 2 sets of observations to know what types of behaviors or patterns do you need to see from a consumer to mean that they're very likely to convert to a sale on a retailer's forecourt. So that model and that logic has been years in the building and creating.

So I think definitely Buying Signals was always a product that we had in mind that we were planning to build and launch. The timing of us testing, piloting, really, really robustly testing that model, the connection with deals anyone that submitted a deal, any buyer is clearly very likely to be pretty high intent when we talk about levels of intent. So you've immediately got a very identifiable cohort of consumers that you know are going to be very high intent.

What Buying Signals does is then for consumers that might just have submitted an e-mail lead or in time buyers that we might just have seen interacting on our platform, but that haven't left any digital footprint with a retailer, we're able to predict for retailers which of their stock units are more or less likely to sell and how fast they're likely to sell. So step one is the connection to Deal Builder and delivering up, serving up a level of intent and a greater level of understanding, which we're already seeing from retailers will change like the next best action they then take with that buyer.

But in the future, the evolution of this product should be to enable retailers much more actively to manage their forecourt based on all of the observed leads and deals that they would have been getting in the old world, but also every interaction that's happening on our marketplace and giving them some sense of what that really means for their forecourt. So it makes sense, I think, and the timing is right to bring it together as part of Deal Builder and to make it part of that proposition for launch. But in the future, there's lots more we can do with the buyer propensity model and that buyer signal thinking and logic to deliver more value to retailers.

On the AI, I mean, we do use the technology quite a bit, but I'm going to pretend that we're right at the center of OpenAI. We do work very closely with Google and Gemini. I think as it relates to -- the first thing I would say is that when you speak to -- listen to the people that are building this technology, you tend to get quite a balanced view. I personally think and whether it's the founder of OpenAI or one of the founders of OpenAI, they do tend to give a pretty balanced view about, this is about token prediction and text prediction. There's not really a semantic understanding of the content that the models are doing, but they work very effectively because they basically say, we don't really care how you get to it. But if you can predict an output very accurately, then that's a good thing. It doesn't matter so much how you get to it.

I think like 2 observations of things that I think have been a bit oversimplified is, the first I would say in relation to agentic is that there is quite a big difference between general models and what general models can do and what agents might be able to do. And those agents need to be quite specialized in order to do jobs and someone needs to do the specialization. I'll give you a really simple example. We could not use an open model to do something as simple as categorize images and write descriptions on Auto Trader. We had to augment the model, train it ourselves. And that's not even really particularly agentic. That is still a generalized model. But even that task itself, using a general model wouldn't work for it.

Now do we think agents can do lots of stuff for users over time? Yes, absolutely, but they'll need to be more and more specialized. And the idea of you just being -- these general models are going to solve the whole world's problem. I don't think anyone really believes that's possible. And agentic AI itself technically is still a bit of a way to go before you see that playing out, although in some fields, it is. The second thing I would say is actually all around the real-time aspect of things. The models are trained every 6 to 9 months. Maybe that increases and they do kind of hoover up the Internet, all the information it can get to on the Internet and use that to create better and better predictions. That is very compute heavy as we can see and NVIDIA's share price suggests is true.

What they don't do and don't necessarily need to solve is accessing real-time information because in order to do that, well, what they do there is they partner with search engines, ChatGPT, with Bing, Google and Claude, with -- sorry, Gemini and Claude with Google. And that's because that's a job that has been done well and replicating that, you'll run into exactly the same issues there.

So for real-time data, what they do is they use their big model that's very, very intelligent to make better queries of a search engine and then bring it back and summarize it. When you come to really granular data like even listings and the data that's on Auto Trader, you probably need a level deeper than that because you can't get that through a traditional search engine. So that is why we think ending up working with one of those platforms and allowing people to access Auto Trader there is probably the way that it will go. But we can't offer any guarantees around these things. But I think it's very easy to see it as a big blob and extrapolate out. But technically, a lot of the engineers will say, well, no, that scenario is just not going to play out. And there are a few examples of it.

I want to get the transformation...

Transformation, do you want to take that, Deal Builder and Buying Signals.

Are you going to do bigger than normal?

So it is in relation to the event. Yes, something like that. I mean, I think 1st of April pricing event, it's obviously a very live conversation internally. As Catherine explained, I think we feel like the product set of Deal Builder in itself and buying signals should particularly over a longer period of time, generate a lot of value for customers. But we still haven't quite landed at what the -- what percentage we're not going to communicate to customers until January. I mean, historically or certainly in the last 3 or 4 years, we've done 3% to 4% on price, 2% to 3% contribution to the product lever. There's nothing here that suggests would be outside of those ranges. And generally, if you say the last 3 or 4 events that we've done have been -- I think we feel like they've been good ones, then hopefully, this is another good one.

It's Lara Simpson from JPMorgan. Sorry, I just wanted to come back to stock. I know it's been a big talking point. Firstly, I suppose, on the speed of sale, you said it was still 1 day faster in October. Were you surprised by that acceleration? Because it feels like a lot of the forward indicators, it should start to stabilize because we're talking about slower demand, supply coming back, but then the speed of sale keeps disappointing. So were you surprised?

And then you've obviously reiterated the guidance. Interested on the stock lever guidance. What are your assumptions of speed of sale? Because I feel like in October, we should be getting to easier comps. Are you assuming that speed of sale stabilizes or slows or the status quo maintains?

And then just a quick question on Autorama actually. A bit of the top line beat was actually from the vehicle and accessory sales, up 20%, I think it was. Has there been any positive surprise there? Because I thought longer term, we should be scaling down that line from a P&L perspective. So just interested on that. And if what you've seen in H1 has changed any of your strategy for Autorama, particularly in terms of the top line moving parts and then the profitability of that business?

Yes. I mean I can take all of them. So look, I think stock -- I mean, if you think about the stock lever specifically. So we've guided it to be marginally down for the full year and it really is marginally down in the first half, that's implying similar in the second half. I think generally, the assumption around speed of sale is certainly what was set out at the full year is that when you hit this point, it gets to be more consistent, and a day quicker. I think in the round, it does feel more stable generally.

So I think we're not expecting speed of sale to accelerate in the second half. That would probably be contrary or a downside to that guidance, yes. And I think there still is slightly -- if you look more medium term or into next year, a hope and belief that supply does start to improve as you get better flow of vehicles, better registrations coming out the back of the pandemic. I think we're just being a little bit cautious on whether we're going to see that in the second half or not. And as I mentioned, that fourth quarter is always a slightly unpredictable one.

From an Autorama perspective, I think you're absolutely right that the vehicle and accessory sales is not part of the long-term strategy, and we still have a belief that over time, that will reduce or disappear. It was really a tactical decision that there were -- like I said, we took extra 300 vans that passed through the balance sheet. They didn't sit there for very long. And just because the van volumes have been slightly lower, we felt that, that was a sensible thing to do.

The long-term strategy is still 100% seeing more volume delivered from the Auto Trader platform for users that are already there. And we're seeing some positive signs of that, albeit off a low base, but the Auto Trader volumes are growing or have grown pretty strongly in this first half. It is, as you'd imagine, heavily skewed towards cars over vans. And so this is -- some of what you're seeing in the first half is -- we're wearing a bit of a yield hit from that changing mix, which I think will probably play out a little bit in the second half. But if we continue to grow those volumes, we're still very optimistic in terms of hitting profitability and then hopefully seeing good growth and getting to the 20% to 30% margins that we set out when we acquired the business. It's very much part of -- there are a number of products that fit into the new car suite. It's very much part of that.

Thanks, everyone, for joining us.