Good morning, everyone, and welcome to Autotrader's results presentation for the half year ended 30th of September, 2026. I'm joined today by our CFO, Jamie Warner, who will present our financial performance and an update on the market before I take you through how Autotrader is transitioning from a marketplace into an active specialized AI-powered platform for car buyers and retailers through Autotrader Intelligence. At the end of the presentation, Jamie and I will take questions in the room. The first half of financial year 2027 has been in line with our expectations, demonstrating the resilience of our marketplace, the importance of what we do for retailers, the strength of our brand, and the discipline of our teams. I'd like to thank everyone at Autotrader, our customers, shareholders, and wider stakeholders for their continuing trust and commitment. Starting with slide 2, which sets out our long-term performance across our key financial metrics over the last decade. As you can see from the charts, despite navigating various macro cycles, COVID disruptions, and recent supply shortages, Autotrader continues to deliver a consistent track record with very limited incremental capital requirements. This decade-long track record of compounding growth and strong margins provides the ideal foundation as we move into our next phase of growth, combining the strength of our core model with Autotrader Intelligence. Now for the headlines for the first half of the year. Basic earnings per share grew 5% year on year to 18.1 pence, supported by our accelerated share buyback program. As anticipated, revenue excluding vehicle and accessory sales was essentially flat at £301.8 million. This reflected the lower run rate exiting financial year 2026 due to a loss of some retailers and some spend moderation following feedback around our Deal Builder product rollout in late 2025. There are also tougher comparables following the conversion of a stock offer in June 2025. As we progress through the half, retailer forecourt numbers have recovered, growing from 13,600 at the end of March to 13,900 by the end of September. Group operating profit reached £198.5 million, with costs being well controlled. We maintained operating profit margins at 63%, or 66% when excluding pass-through vehicle and accessory sales. Our engagement metrics with UK car buyers remain unrivaled. Consumers spent 12 times more time on Autotrader than on our nearest competitor, which is a combination of 4 separate automotive marketplaces added together. Over 75% of all minutes spent on UK automotive marketplaces was spent on Autotrader. Almost 80% of buyers choose to come directly to us, underscoring the consumer trust, confidence in our brand, and the depth of experience needed to buy a car with confidence. In the emerging landscape of AI search, we're comfortably the most visible automotive marketplace as measured by Profound, who independently monitor generative engine optimization, or GEO. Referral traffic from AI assistants currently remains low, below 1%, although we continue to evolve our site search experience and plan to adapt as consumer habits and expectations shift. We've also continued to grow the percentage of users who are logged in on Autotrader. At the end of September, around 70% of our product page views were from logged-in users. During the 6-month period, cash from operations was £215.3 million, and we returned £330 million to shareholders through dividends and share buybacks. At the end of September, we had debt of £335 million with leverage at 0.8 times EBITDA. We're proposing an interim dividend of 4.2 pence per share, which is an increase of 11% on last year. Moving now to slide 4, which outlines our strategy. This does represent a fundamental shift in how we view and deploy artificial intelligence at Autotrader. In previous results presentations, much of our discussion around AI was centered on disintermediation risks or how we're already embedding AI tools at scale. We've been working with AI now for over a decade and LLMs since they rose to prominence. Based on that, we see the opportunities outweighing the risks. This technology is a once-in-a-generation step change, and just as print to digital, PC to mobile, data center to cloud, analytics to data science, we believe it represents an opportunity to make Autotrader more valuable to car buyers, more valuable to retailers, more valuable to employees and shareholders as well. It gives us the opportunity to expand our addressable market, to deepen our competitive moats, and unlock opportunities that were previously unobtainable. It expands the problems we can solve and the speed at which we can solve them. All with the goal of reaccelerating top-line growth. This is in addition to and not a replacement of our existing business model and activities. We still see stock prominence, Deal Builder, and annual packaging events as key drivers moving forward. Indeed, these will be, if anything, further strengthened by the strategy we'll outline through the presentation. We've consistently compounded revenue and earnings for many years and believe the AI technology shift will allow us to continue to do so for many years to come. The transition of Autotrader into an active intelligent AI platform powered by Autotrader Intelligence is already underway. It's powered by our cloud infrastructure, our AI coding agent, 57 specialized AI models, 200 million tracked consumer events a day, and 15 billion proprietary vehicle data points. We can do what generic LLMs simply cannot, and we will take out the heavy lifting and decision fatigue from buying, selling, and managing vehicles. As you can see on the slide, this strategy focuses on 3 core customer pillars, which I'll cover in more detail later in the presentation. But in short, for car buyers and their agents, we'll fully leverage our existing tools and AI to provide the deepest and most specialized AI-enabled car buying experience in the UK. We'll do this on Autotrader, we'll do it within AI Assistants, and we'll do it for AI Agents also, ensuring that wherever consumers' search starts, Autotrader remains the ultimate source of truth wherever and however they want. For retailers, we are transforming our retailer portal into a complete AI-native operating system rather than just generating leads and multiple data points. Autotrader Intelligence will cut through all that to provide the exact actions that a retailer should take alongside automated workflows for sourcing inventory, managing inventory, creating listings, for undertaking dynamic pricing and triaging leads. For vehicle manufacturers, we're moving beyond basic display advertising to build a whole-of-market new car deals experience, enabling them to sell cars more effectively and profitably. We've been seriously investing in our technology for the best part of 2 decades, and while things will no doubt evolve, for that reason we do not see this as a risk to our existing margin profile. But before we get further into the strategy, I'll hand over to Jamie, who'll talk through the financials and an update on the market. Thank you, Nathan, and good morning, everyone. I'll begin with a summary of our consolidated financial performance for the 6 months ended 30 September 2026. Total group revenue was £313.4 million, Within this, revenue excluding vehicle and accessory sales was £301.8 million compared to £301.2 million in the prior year. Vehicle and accessory sales, which carry an offsetting cost of goods sold, were £11.6 million, which was down 30% as we focus more on acting as a pure intermediary rather than taking principal risk. We now expect vehicle and accessory sales for the full year to be £25 million, which is lower than the originally guided £40 million, and this will mean cost of goods sold also being £25 million for the full year. Group operating costs decreased 2% to £117.2 million, reflecting our continued cost discipline. Our share of profit from joint ventures, representing our share in Dealer Auction, increased 10% to £2.3 million. Group operating profit was at £198.5 million, down 1%, and group operating profit margins were flat year on year at 63%. And flat at 66% when excluding vehicle and accessory sales. Basic earnings per share grew 5% to 18.1 pence, with EPS growth outpacing operating profit due to a 9% reduction in the weighted average number of shares in issue. Following our accelerated share buyback programme. Cash generated from operations remained broadly consistent with the prior year at £215.3 million. Turning now to revenue across our key customer segments. As explained at full-year results, with more than 50% of all lease deals being completed through Autotrader, we're now reporting our results under one operating segment. Total trade revenue was £267.2 million. Within trade, retailer revenue, our largest revenue stream, was £250.8 million. Down 1% year on year, with more detail on the following slide. Home Trader and other trade revenue grew 7% and 9% respectively. Consumer services revenue also grew 9% to £22.9 million. Within this, private revenue grew 8% to £14.2 million, underpinned by healthy private advert volumes, many of which were exclusively advertised. on Autotrader. Motoring services revenue increased 12% to £8.7 million, driven by stronger finance revenues. Leasing, manufacturer and agency revenue was £11.7 million. Within this line, the typical manufacturer and agency revenue, which we used to report separately, grew 18%. to £8.5 million, driven by advertising campaigns with new entrant brands seeking UK market share. Leasing commission and ancillary revenue was £3.2 million, down from £4.9 million in H1 of 2026, reflecting a lower number of deliveries. with lower van and pickup volumes and also lower average commissions as a number of funders moved to a fixed commission model. Moving to slide 8, which details average revenue per retailer and retailer forecourt dynamics. Average monthly revenue per retailer, or ARPR, grew 1% or £43 to £3,037 per month. Looking at the 3 ARPR levers in turn, price contributed £117 to ARPR growth, driven by the successful execution of our annual pricing and product event on the 1st of April 2026. which incorporated a like-for-like price increase. Product contributed £15 to ARPR growth. This was driven by the integration of buying signals into our core advertising packages, offset by a lower customer package mix. We've recently completed changes to our packages which should support a better second half performance. Stock decreased ARPR by £89. Live car listings were 2% lower year on year as predominantly smaller independent retailers adjusted their slot volumes largely due to tougher trading conditions and feedback on the speed and scale of our Deal Builder product rollout in November 2025. The average number of retailer forecourts advertising on Autotrader during the first half was 13,762, down 2% from 14,080 in the first half of last year. However, forecourt recovery through the period has been encouraging. As Nathan said, forecourt numbers rose from 13,600 at the end of March to 13,900 by the end of September. On slide 9, we provide further granularity on livestock and retailer forecourts. Average live car stock on the platform during H1 averaged 454,000 cars, down 1% year on year. However, by the end of September 2026, total live car listings reached 464,000, which was a 4% increase year on year. We conducted a detailed analysis of the retailers who churned during this challenging period between November 2025 and January 2026. The data shows that of those customers that left the platform, we've re-signed 2/3 of those who are still trading, and we remain focused on continuing to engage with that remaining third. In leasing, total deliveries was 3,447 units. Lower van volumes were partly due to committing to purchase less vehicles, which was shown in the revenue through vehicle and accessory sales, and pickup volumes were impacted by the removal of tax-free. benefits. Turning to costs on slide 10. Total operating costs decreased by 2% to $117.2 million, but excluding cost of goods sold, costs marginally increased to $105.4 million. Within this, people costs decreased 2% to $51.5 million. Average FTEs decreased to 1,203, with lower headcount and reduced share-based payment charges being partially offset by higher underlying salaries. Marketing spend increased 32% to $12.9 million, although this increase reflects the point that H1 last year was unusually light. With spend this year being broadly in line with what we spent the year before last. Other costs, depreciation, amortization, and digital services tax were all flat year on year. As mentioned, total group operating profit decreased 1% to $198.5 million, and both group operating profit margins and those excluding vehicle accessory sales remained consistent year on year. Cash generated from operations is also consistent at $215.3 million. In May and then July of this year, we amended and extended our debt facilities, taking our total committed borrowing facilities to £600 million through September 2030. At 30 September 2026, gross bank debt drawn was £335 million, and net bank debt stood at £320 million after cash and cash equivalents of £15 million. This corresponds to a leverage ratio of 0.8 times. During H1 2027, we returned £330.2 million to shareholders through share buybacks of £269.6 million, purchasing nearly 7% of issued share capital. And dividends of £60.6 million paid in respect of the FY26 final dividend. Our capital allocation policy remains consistent, investing in organic growth, maintaining a dividend payout ratio of approximately 1/3 of net income, and deploying surplus cash flow and debt capacity into share buybacks. We still expect to complete approximately £500 million of share buybacks in financial year 2027. And finally, the board is proposing an interim dividend of 4.2 pence per share, an 11% increase. Now on to our outlook statement. First half trading has been in line with our expectations. We therefore maintain our full-year guidance that group operating profit will be between £395 million and £415 million, with accelerating revenue growth. in the second half. Moving now on to some slides covering the UK automotive market. Both new and used car transactions continue their post-COVID recovery. New car registrations grew an impressive 12% over the 6 months to 1.12 million vehicles, supported in both retail and fleet channels by new entrant manufacturers. Used car transactions grew 1% in Q1 of our financial year 2027 to 1.99 million transactions, with supply gradually recovering in younger age cohorts. The long-term fundamentals of the UK automotive sector remain good. The UK car park has continued to grow, and with strong new car registrations, we expect this to continue. Retail pricing, a key component of retailer profitability, has steadily increased over the past 6 months, with speed of sale being broadly consistent. Autotrader's competitive moat has never been wider. According to Comscore, consumers spent 12 times more time on Autotrader than on our nearest marketplace competitor, a combination of the 4 brands shown here. This has increased from 9 times over the same period last year, measured against the same competitors. Autotrader also continued to account for over 75% of all minutes spent across all UK automotive marketplaces. During the first half, we did see some softening in both time spent and monthly cross-platform visits as measured internally, which was particularly acute in June and July. When looking at our traffic share, we've grown against our competitors and are consistent against the websites of our largest customers, meaning trends are more a market reflection than anything Autotrader-specific. We also track our sales influence using data supplied directly to us by customers, which slightly increased year on year, giving us strong confidence in the very important role we play in buying and selling vehicles in the UK. I'll now hand back to Nathan to cover our strategy in more detail. Thanks, Jamie. Over the past 2 years, uh, like almost every technology company, a lot of our dialogue around AI has been about reassuring investors against disintermediation risks, or showcasing incremental feature rollouts. That narrative has obscured the opportunity the technology shift represents. General AI models are only as powerful as the data and context that they feed on. A general LLM trained on public web text cannot accurately price a specific 4-year-old electric vehicle, verify its provenance, or execute a complex regulated finance transaction on a dealer's forecourt. But when you do pair that state-of-the-art AI technology with Autotrader's 15 billion proprietary vehicle data points, 200 million consumer events tracked every day, 57 domain-specific models, and our deep real-time integrations across the UK automotive ecosystem, you do create a virtually impossible-to-replicate new capability that can underpin the best specialized AI-powered assistant for car buyers and retailers in the UK. This is not simple slideware. All of the foundations to do this are well understood within our business. They're established, they're productionized, and working for us already. Throughout Autotrader's history, when faced with significant technology change, it has been by leaning into the change that has proven to be our best defense. Through each of these transitions, our competitive position and value to shareholders has both increased. This same approach is at the heart of the fundamental shift we're making. Turning to slide 19, which brings us to the core engine of our strategy, which is Autotrader Intelligence. To understand where we're going, it's important to understand exactly what that actually is. This is not a single feature, it's not a chatbot or a third-party AI wrapper that's been bolted on the side of our existing website. Autotrader Intelligence is our central scalable AI platform, which is an intelligent orchestration layer powering every interaction across the platform. At its core, it combines over 57 specialized models, including machine learning and LLM models, alongside a platform that can swap and change any number of foundation models. which avoids lock-in and enables us to choose the very best model for each specific job. What makes that unique are the 3 pillars that a generic model or startup can't match. As mentioned, we have decades of proprietary data covering UK vehicles from their birth on the factory floor right through to scrappage, including exact fitted specifications, provenance, valuation movements, and much more. We have deep market integration directly into manufacturer build data and taxonomy, banks and lenders for real-time finance approvals, 350 automotive technology providers, and directly into thousands of retailers. Many of these integrations are two-way, as much of the automotive ecosystem draw on our APIs for decision-making and operational processes. Finally, as mentioned earlier, we record 200 million consumer events every day, which provides live supply and demand, allowing our models to predict pricing trends, buyer purchase intent, and vehicle turn rates in real time. All of this is only available to Autotrader. And it's this deep architecture which is why Autotrader Intelligence delivers a level of accuracy, speed, and trust that public models simply cannot achieve. And it's precisely this advantage that positions us to move from an advertising marketplace into the best specialist AI assistant for buying and retailing cars. in the UK. To capture this opportunity, we've organized our strategic choices around 3 core consumer customer pillars. The first is car buyers and their agents. The second is retailer customers, and the final is vehicle manufacturers. I'll touch on each of these in turn in a bit more detail now. Focusing first on car buyers and their agents. And where we're looking to embed Autotrader intelligence directly into the consumer journey to transform how they buy cars. Consumer search behavior is evolving. We're moving away from rigid keyword-based dropdown menus towards natural language conversational queries and AI-assisted search. Discovery. Filtering down vehicles will always be important to making a good decision on a car. However, how you do that is no doubt going to evolve. Our ability to do natural language better than other models is down to our detailed taxonomy covering all UK road vehicles. This granular vehicle data structure, which now allows nearly 200 different search attributes, everything from exact boot capacity to specific insurance tiers, running costs, and lifestyle attributes is critical. On a traditional website or mobile interface, such as those that we've operated historically, you can realistically only ever expose about 30 of those filters via dropdown boxes. Before the screen and user experience becomes completely cluttered and unusable for the average buyer. AI search eliminates that interface constraint. A consumer will be able to type or speak a complex multivariable request in plain English, such as find me a reliable electric family SUV under £20,000 with low insurance and a boot big enough for a double buggy. You'll get back a set of results that doesn't guess or approximate an answer, but delivers an accurate set of fast and comprehensive results that a generic engine or ungrounded LLM simply cannot replicate. However, consumer discovery won't just happen on Autotrader. We believe firmly it will happen inside third-party AI assistants and personal consumer agents. Which brings us to our agentic readiness and model context protocol integration. Rather than letting external LLMs scrape static, unverified, and unstructured web text, or trying to make that job difficult for them technically, we have deployed dedicated MCP servers and API endpoints. Those protocols allow external AI agents such as ChatGPT to directly query our live database in real time, accessing structured web standard data. Crucially, this means that wherever a consumer's car buying journey originates, it's Autotrader intelligence that remains the underlying source of truth. We provide the real-time inventory, the provenance of the vehicle, verified pricing, finance quotes, part exchange quotes, and then direct that high-intent, ready-to-transact consumer directly back onto Autotrader to complete the deal. Once a car buyer has identified a vehicle, the biggest remaining hurdle to a sale is trust, anxiety, and decision fatigue. Buying a vehicle is typically the second largest purchase a household ever makes, and navigating from hundreds of thousands of listings down to a single vehicle can feel daunting and error-prone. This is where Autotrader intelligence will transform the journey by converting those 15 billion vehicle data points and all our consumer data into instant objective confidence. For any buyer. We do this today with vehicle highlights, instantly pulling out rare features such as single owner status or practical attributes like a bigger boot alongside our market-leading price flags. Our ability to provide this useful assurance in a usable interface like search has also been unconstrained by the advent of AI. But building trust is only half the equation. We're also enabling buyers to take direct action and complete more of the buying journey on Autotrader, something that's also available— will be made available to AI assistance in a consistent way across all the inventory we have on our site. Deal Builder seamlessly connects digital discovery with forecourt fulfillment. It allows consumers to value their part exchange in real time, receive pre-approved finance quotes, and reserve the car instantly with a refundable £99 reservation. Since making Deal Builder a core part of our platform proposition, adoption has accelerated. Having addressed the concerns of some customers at the start of the calendar year, we now have around 9,300 retailers active on Deal Builder, more than double the 4,000 we had this time last year. We also remain on track to have all customers on by the end of this financial year. Deals submitted through Deal Builder continue to convert at sales at twice the rate of traditional leads, and over 50% of all deals are completed outside normal dealership operating hours. By removing friction and fear from the consumer experience, Autotrader Intelligence turns passive browsing into high-intent transactions, deepening our moat while making the lives of car buyers and retailers easier. Our second strategic pillar is transforming our retailer portal into an AI-native retailer operating system, becoming the AI assistant for car retailing. UK automotive retail remains highly fragmented, with many retailers managing disjointed software systems for sourcing, pricing, inventory, and lead management. With 18,500 daily active users on our retailer portal and 6.9 million daily calls to our API through Autotrader Connect, we are already the operational background for UK— backbone for UK dealers. By bringing more Autotrader intelligence capabilities directly into customer workflows, we aim to reduce system cost and complexity, save staff hours, and help retailers generate better returns on their efforts. Crucially, this shifts Autotrader on a retailer's P&L from being a discretionary advertising line item into a daily operational utility that dramatically improves their business. Providing a great foundation to monetize new capabilities and drive long-term revenue growth. Through Autotrader Intelligence, we've demonstrated this already with products rolled out in the past 18 months. We've previously referred to these as falling under the Co-Driver umbrella, which takes on much greater ambition as part of this strategy. We have computer vision models that reorder advert images based on buyer click-through data, while generative AI automatically drafts comprehensive vehicle descriptions and pulls out key selling points. This has reduced advert creation time from 28 minutes down to just 5 minutes. Machine learning propensity models trained on millions of historical sales score incoming leads in real time. Buying signals flag whether a buyer is local, highly qualified, and ready to purchase. Almost 800,000 inquiries have been flagged as high intent and convert into vehicle sales at twice the traditional inquiry average. Generating significant insight and efficiency for our customers. Our third pillar expands our role with vehicle manufacturers. The UK new car market is undergoing structural change driven by electrification, the ZEV mandate, and new brand entrants. Manufacturers spend over £750 million annually on UK marketing and over £5 billion on tactical pricing and incentives. which represent a very material opportunity. Many of you will know that we've been pursuing a strategy around new car for some time now, and every time we have made progress. Progress in terms of stock, retailers, and most importantly, audience. We have over 600,000 people engaging with new cars every week on Autotrader. We have excellent relationships with manufacturers, and through Autotrader Intelligence, we plan to leverage that to provide a cost-effective and high-converting advertising channel to a market that is growing fast and more competitive than ever. We're uniquely well-placed to provide consumers with new car— with a new car deals comparison. We have deep finance integrations with brands, with their captive finance houses, with retailers and leasing companies. With over 90% of new cars bought on finance, we'll enable consumers to compare cash discounts, PCP finance, and PCH leasing offers side by side with transparent deal scores. In conjunction with that, we can use real-time demand signals from our 80 million-plus monthly visits to allow manufacturers to run targeted conquest campaigns that reach buyers actively by comparing brands, delivering measurable sales volume. This will enable us to offer OEMs with a transparent, high-attribution conquest sales channel and grow their advertising spend with Autotrader. And as our data models build, we will be well placed to help them optimize their more than £5 billion of pricing and incentive spend in much the same way that we do with retailers. Today. Much of our shift in our strategy is enabled because we've dramatically increased our capacity to deliver software through deeply embedding AI in our engineering organization. We've long spoken about the strength of our platform. It is robust, scalable, and enables high-velocity software delivery. Over the past 12 months, we have built an Autotrader-specific AI capability that works on top of that platform. It is a model-agnostic coding agent that is not beholden to any one model provider, enabling us to choose the most efficient and effective model for the given job. We've then built into the AI the full context skills and tools required to operate effectively within Autotrader's environment, and if required, autonomously within guardrails that ensure speed. Speed, safety, security, accuracy, and scalability. It's built on our cloud-based delivery platform, which equips our 400-person product and technology team with model-agnostic coding agents operating in 2 complementary models. Engineering-directed, which is integrated into developers' IDEs and CLIs for real-time coding assistance and safety checks, or autonomous execution, which enables event-triggered background agents integrated with Slack, Jira, and GitHub to autonomously identify and execute tasks. This is uniquely tailored to our environment where it adds shared Autotrader engineering guidance, task-specific workflow safety controls, and connects AI clients to internal engineering tools, enabling engineers to use the workflows and tooling that are already in place today. Since launching in March, our coding agent has completed over 20,000 autonomous engineering jobs. Developer adoption has grown over 40% quarter on quarter. And we're processing around 10 billion tokens monthly. Crucially, under our centralized infrastructure and model tuning protocols, net monthly AI engineering spend has remained flat despite dramatically increasing workloads, decoupling software delivery velocity from platform operating cost. What we've talked about today isn't a distant roadmap or series of experimental AI prototypes. Autotrader Intelligence is live, productionized, and already delivering tangible value at scale right now. Leaning in and working with the technology early has always been at the heart of Autotrader's success. Because of this approach, we have always been well placed to benefit from technology change as opposed to managing it as a risk. And we see this time as being no different. We have a modern cloud-native infrastructure that processes billions of real-time market signals. We have over 55 domain-specific AI models running seamlessly in production. Many of them building on top of the best-performing open LLMs. More than 15 billion proprietary vehicle data points that no public LLM or new market entrant can replicate. Logged-in users accounting for 70% of all activity on Autotrader. 200 million recorded consumer events every single day, integrations throughout the industry that are called on 6.9 million times a day, embedded distribution with nearly 14,000 customers. And a 400-strong team of highly capable product engineers and data science specialists, 100% dedicated to making buying and selling cars in the UK better. And now their capacity has been multiplied by our own coding agents. Delivering software faster, smarter, and more efficiently than at any point in our history, such that the scale advantage that Autotrader once had has now become much, much greater. For nearly 5 decades, Autotrader has led the UK automotive industry through every major technological evolution, from print to digital, desktop to mobile, physical to cloud, and data warehouse to data science. We're now taking our biggest leap forward yet by transforming Autotrader from a passive listings marketplace into an active AI-native operating system. We believe this can expand our addressable market, deepen our competitive moat, and underpin compounding revenue and earnings growth for many years to come. We have a strong core business and the data network effects, technology and team, not just to adapt to AI, but to embrace it to revolutionize buying and selling of cars. In the UK. That's why we feel good about the opportunity ahead. History has shown us time and time again that we are able to lean into technology change and benefit car buyers, customers, and shareholders too. Thank you very much for joining us today. Jamie and I will now take questions from people in the room. Usual, start at the front and work our way back. Start down, down there. Thanks. It's Andrew here from Barclays. 2, please. First one, I guess, is to press you a bit harder on the revenue outlook in H2, where you haven't been very specific. I'm not going to say accelerating. So could you give us a bit more help in the moving parts on revenue in H2, particularly in Meet Martha, and how you're thinking about stock from here? where the product can be a bit better as some of the changes in the packages run through, and then also price. And then the second question is the inevitable one on FY28. Appreciating that you communicate the price increase to dealers in January and you won't have made all of the decisions yet, but can you give us insight as it stands in terms of what products are going to be part of the toolset to help that conversation, and as it stands today, how you kind of feel about your ability to drive more of a mid-single-digit ARPU growth, uh, in FY28? Yep. So yeah, so I think the first thing, first thing to say on the sort of outlook is like there isn't, there isn't really a change at the sort of bottom line revenue number, so the revenue excluding vehicle accessory sales. Obviously, I laid out a change to vehicle accessory sales that has no impact on profit. So In the round, where what we said at the full year, where consensus is, we're happy. But you're absolutely right, Andrew, there are, in the minutiae of the detail, there are some things that are marginally more positive or negative. Retailer forecasts probably are a little bit further ahead than what people thought, which is obviously extremely positive. I suspect the exit rates probably similar to where we would have expected to be, but obviously it's going to be a little bit more first half weighted. That better retailer number is marginally dilutive to ARPA. And as you kind of aggregate in those higher forecourts, forecourts typically smaller that are joining, you get the kind of dilution of ARPA. So the product lever, whilst it looks very big in the first half, will moderate down. And I think the guidance is still absolutely fine. Similarly, stock. You know, the stock lever, whilst it looks very negative in the first half, if you think about the second half of last year was negative 85, all of that negative was in the fourth quarter. Which actually means that fourth quarter was more like minus 170. So actually to report negative 89 shows good progress and momentum. And I think there's great confidence that it's going to improve from first half into the second half, and again, the 30 to 40, you know, reasonably well, well underpinned, I think. Product lever, I think, is maybe where it's a little bit softer than we thought, and that's really down to the package upgrades, albeit we knew that we would only have the changes that we've made ready by the end of September. So those changes are now in place. So we've got good levels of optimism that we'll see better customer penetration in the second half, more incentive to upgrade, and the product lever to improve from that £15. So, you know, if you take marginally better 4 quarters, maybe a marginally softer product lever, you still get to the same retailer revenue. Then outside of that, consumer services is certainly better than people were expecting. It's better than I was expecting. Manufacturing agency is probably also a little bit better. Leasing is probably a little bit softer. So again, in the round, it will, you know, if people didn't move their numbers, I think it's no problem because it does all aggregate down to the from the same point. In terms of FY28 outlook, as you know, we don't do guidance on the year ahead, but in terms of the event, I think The big message is that the whole Deal Builder thing, all of that is well and truly behind us. We feel very comfortable with that. So in some sense, you know, we've taken our lessons from that, we've listened to retailers, and that's actually really informed the products that we're talking about for next year. Some of these which are already in test and going live, there's no reason why we'll hold those back for a magical moment in April. We'll get them live as soon as we can. At the moment, that includes WhatsApp as a messaging platform and intelligent call tracking, which you can think of as rather than getting a missed call, you get a detailed lead like you would get online. What's the vehicle that they're looking at, the buying signals, all those pieces of information that go along with an online lead, which means you can actually govern and track and manage the lead in the same way you would others. Today, in many dealers, they simply don't know what to do with those. You get on average 3 calls, that's not 3 leads. Who do you call back in what order? So it's actually, while sounding not exactly the most glamorous of products, practically very, very, very useful to a retailer converting sales more effectively. Then we've got a couple there we haven't decided of these ones, which we would also go with in buying signals and appointments. So appointments is linked to Deal Builder. Obviously, some customers said, I don't want a £99 reservation fee. We kind of listened to that, adapted to enable people to request reservations. Some of them would like the ability for them to request an appointment and provide some times. So you've got that in there, and then a further enhancement to buying signals. So it's a pretty strong lineup. Deal Builder, the Deal Builder episode is well and truly in the rear vision mirror. So I think we revert back to what we've always said, that we expect these events to be in total combination of price and product around 5 to 7%, and there's no reason for us to think that that would change moving forward. I knew you were going to ask that. I mean, I think, look, getting there. But look, I think if you look at the trajectory, we've delivered pretty much flat in the first half. The steer for this year is Consensus is low single digit. Consensus for next year, I think retail revenue is just under 6%, total revenue is at 5.5%. That's a good trajectory. to be on, and we're pretty comfortable with where all of those numbers are. I wouldn't suggest that we're here saying all those are moving up, but hopefully we feel like there's good confidence in them being underpinned. We just work our way along the front and then go back to the next row. Yeah, hi, it's Adam Berlin from Goldman Sachs. 3 questions if I could. So you gave us quite a detailed presentation there about the new strategy and how AI feeds into that. Can you put that to any numbers? So in the past, you were delivering high single digit 8% organic growth in your core Autotrader business. Does this get you back to that? Does that get you beyond that? How should we think about that? Or is it just more about what you need to do to fight off the competition? Second question, um, you talked a little bit about minutes being down this half, maybe because of the World Cup. But can you talk a little bit about leads? Um, are your— the number of leads you're sending to dealers, are they still holding, or is there a problem that people are skipping Autotrader and going directly? And any confidence you can give us that's not happening. Um, and then thirdly, on just Deal Builder, You did talk about how many dealers are now using it, but I think it would be more helpful to understand how many transactions or what percent of leads is Deal Builder actually driving today, and is that material yet? understand how important the product's becoming. First one, second. Yep. Okay. So I'll take the first one. No, we're not putting long-term guidance around what the strategy means for our numbers. I'd say we're pragmatists, as you know, and at the moment our focus is, well, let's get to the sorts of numbers that Jamie spoke about and then go from there. The reality is that the revenue that we're— what's implied in everything that we're talking about is a deeper play into dealer solutions and dealer software, which isn't a market that we've really got any share of, any meaningful share of. We've built dealer systems, obviously we know how to do it, but it's all just their activity on Autotrader. What we're talking about is going much, much broader than that. Implicit within that is the fact that we will likely provide choice. We're not talking about an expansion of a hard bundling strategy. We're still going to do that, but that's kind of part of the strategy over here. And what we're talking about is on top of that. So in that sense, yes, we're expecting it to be incremental, but it'll be something that by its nature you're selling to a retailer potentially one by one by one at good yields, but these things tend to build over time. So I'd be surprised if— I'd like to see customer penetration next year. I'd be surprised if it was big enough to be able to make its way to meaningful numbers, but we are doing this because we see this as an opportunity to— a pretty unique opportunity. We can build stuff much, much quicker, and we've got big advantages in building B2B software because we run the marketplace. as well. So it is an over-and-above strategy. Apologies, I can't give you the exact number though. And then— oh, sorry, I should just say quickly, no, it's not about fighting off competition. We just— the competitive environment in some ways, we opened our platform to enable anyone to tackle the dealer software space. And we're still going to do that, but we're going to play a much bigger role ourselves. So it's very much a proactive move than anything we feel like we need to do defensively. Sorry. No, I mean, the most important, then on just the audience, the most important thing that we track at that sort of bottom of the funnel is sales influence. So we get actually a lot of sales data from customers. They tell us the postcode of the person that's bought the car, and every search on Autotrader involves a postcode, so we can very accurately match whether the person viewed that car. Or a car that was also at that forecourt, and they've walked in, and we can attach lead volumes to it. And that marginally increased half on half. So we feel very confident around the fact that that traffic, as you say, there's a bit of seasonality, as much as it's maybe a bit of market, but it doesn't feel like it's having any impact on consumers that are genuinely there to buy cars, going through the funnel, using Autotrader, and resulting in sales. Then from the Deal Builder transactions, obviously we've changed the product where not everyone has the full reservation. Some people have requested a reservation, as you can imagine. They are put through in higher volume and don't convert at quite as high a rate. And the way that we've— this is actually how we've always built Deal Builder and a deal specifically. But it is actually more of a packet of data. So you create a deal and it has a buyer, a seller with the dealer, and an individual vehicle, and then you attach buying signals to it. There may be a lead or a message, but there may not. And this is what a deal really is, and it's become a lot richer. Obviously initially with a reservation, but then with buying signals. And so Not that I want to dodge the question, like the full-fat reservation is probably now 10% of a dealer's sales influence. We used to talk 5 to 10, so we have seen a kind of improvement in it. But actually the volume of deals as it's defined where someone's requesting a reservation or just engaging with the deal widget, they might be doing a part exchange, it might be finance intent, you can just put a message in there. is actually quite a significant scale of the number of leads. But I think where we're going is that every lead with the call tracking changes that Nathan talked to, everything will be a deal and every consumer interaction where we think you're near the bottom of the funnel and we're observing you as high intent, we will have that packet of data that then we work out. You can't pass it across, wouldn't pass it across to the customer without consumer consent, but they're things that we're capturing as this is someone that's really interested in this car. How do we personalize the journey, present relevant journeys, and ultimately try and get them better connected to the retailer? That's the direction that we're going in. Will Pack from BNP Paribas. A couple for me, please. Firstly, if we look back, I think probably an underestimated headwind for Autotrader in the help of your dealer customer base in terms of gross profit growth, their cost headwinds. You painted a pretty benign picture in the slides today on the underlying market conditions. I suppose you're more talking to the total market rather than specific dealers. We have fewer listed dealers. When I speak to dealers, it sounds like, you know, they're still finding things quite tough. Could you talk to how your customers are feeling at the moment in regards to their gross profit and cost headwinds? And then Additionally, um, Oh, actually, you answered my second question, so So on You always find retailers that are doing well and retailers that aren't doing well. I would say on average, um, I speak to a lot of retailers, they're actually not doing too badly at the moment. Sales good, gross margin good, trade prices are falling at the moment, retail prices are holding. So that's like a very, very good market to be in. Independence, if you're totally focused on used cars, that's definitely, definitely the case. I think you are right around the headwinds, actually. And that has all been below the gross margin line. Independent retailers, by and large, seem like they're doing just fine, actually. I mean, I had a couple of weeks ago, our customer advisory group, best part of 20, 25 retailers, we spoke about the market and, um, they were reasonably unanimated, which generally means things are really, really good. Being that that's kind of in their nature. But I think that the cost pressures are still there. I mean, last year it was quite acute because of the living wage, minimum wage, NI increases and interest rates. Obviously, I haven't seen what's been decided today, but a lot of retailers are highly exposed to stocking loans. It's one of their biggest costs in their business, and if interest rates go up that one's hard to— that's a hard cost, an impossible cost, near an impossible cost to mitigate. So retailers are still very, very focused on cost. Um, but the imperative around it feels— it doesn't feel nearly as acute as, as last year. It's worth saying, any— for franchise retailers, the difference for them is they're caught up in this new car market, and depending on what portfolio of brands you have will depend on how you're feeling. If you've got lots of the new entrants, um, you know, if you signed up early with them or during the course of the last few years, those cars are great value, selling really fast and good margin. And the, the manufacturers are really good to deal with. They're quick, they're like low maintenance in a sense compared to some of the others. Um, if you're playing in a lot more of a legacy, um, some of the legacy manufacturers are having a tough time of it. They're up against in terms of competition they're up against in terms of the ZEV mandate. And whenever there's some level of strain in the system, it does get shared in reality between manufacturers and, and dealers. So Um, some, some franchise retailers, um, you know, there is a little bit of strain in some of those groups that haven't necessarily set their portfolios up, but, you know, there's not many listed ones now, and the ones that are doing kind of okay, and I don't think that's not representative. Thanks, sounds encouraging. Just a quick follow-up in terms of the AI offerings on the B2B side to your customers. It sounds like you're kind of leaning into that a bit more than you had previously. Is there scope in a medium-term view for your customers to transition to Autotrader being their DMS and allow them to drive material savings in that way, or is that a bridge too far? Um, so I think that the word DMS, and as we've decoupled this ourselves from a strategy perspective, is actually kind of a reasonably clumsy term for a whole bunch of different functionality. We all do the same, by the way, so it's not a comment on you using it, and that's everything from lead management through to invoicing through to booking after-sales. I think we feel like we're not going to solve problems that aren't already well solved today. We're really leaning into the areas where dealers aren't being served today, but some of that functionality is, in some cases, provided by those DMSs. So, are we going to go into, you know, the Pinewoods and the Keyloops of the world, those really hardcore systems that sit right at the heart? Thank you. Giles from Jefferies. My first question is on Autotrader Intelligence. When can we expect it to be powering the monetization for an agentic service or some outcome-based service? Second question, the 1/3 of dealers that churned that are still holding out that you mentioned in the prepared during the presentation. Are the package of changes you put through coming out of the dealer advisory councils enough to get them back? No, because we don't see that by doing those any differently, you can create a lot more profitability for retailers, and that's really where we're focused. But some of those systems will do various levels of pricing, inventory management, online advertising, dealing with leads, not just Autotrader leads, finance systems, which we've got a unified finance platform. So there is a lot for us to go at. And I think it is fair to say we've been building the foundations, but we haven't necessarily lent in as well as we're able to now because we can build a lot more software a lot quicker, and our platform means all the services are there. And building services, AI is not as good at, but building stuff on top of services and interfaces and those sorts of things, It is excellent at. Okay, that sounds Strange sentence, but hopefully that makes sense. And the stock available for reservation is now reaching a point that it feels like you could start to change the marketing message to consumers. around being able to make a reservation. Are you thinking the same way too? Anyone? I'll get that first. Yes. So, I'm slightly in danger of repeating a previous answer. Like, I think we definitely see more opportunity in what we've laid out. Being standalone, being monetized, that's the place that we're going to, going to start appreciating each time that we develop one of these products or features. We will still go through that process of, is this one of these things that benefits everyone, or is it something that could be monetized on a standalone basis? So I think Nathan said it in the presentation, we feel like it can, um, can be, can be incrementally pretty meaningful. I think we're, we're just here today setting out that opportunity where it sits. We're not here, you know, talking about FY28, '29, '30. Um, so that'll come, that'll come down the line in, in future presentations. Um, on the second question, which I was writing down at the time you asked it, just to make sure I've got it right, are we doing enough to win back the rest of the retailers? That essentially, yeah. You constituted the customer advisory councils, you published— WhatsApp messaging was one of the work streams. Is it enough for those churners to come back? Yeah, we think so. There is like a bit of art and science to working out who's churned and who's come back. Being a car dealer is a regulated activity, so understanding who's left, who's churned, who's left the industry entirely. I mean, that was the majority of retailers when we looked at the numbers that actually gone out of business. But, um, I don't think we can do anything more, actually, and it feels like sentiment is definitely shifting now. direction. The, um, whilst customer advisory groups around 20 to 25, there are people there that representing 1,000 independent retailers, and the response that they've said to us is you've listened, you're changing stuff. And then it comes down to offering the best sales channel for retailers. And I think if anything, we've seen as many retailers opt in and come back of their own accord as opposed to us acquiring them back because they've experienced other platforms and and found that they get more sales quicker on Autotrader. And in that respect, from a competitive position, which Jamie pointed out, you know, it's better than it's it's ever been. So, um, whether the retailer number reverts back to falling down a little bit as things get harder and the industry reverts back to its more long-term average, who knows. But I think we should be able to get back to you know, that same number of retailers or the same level of penetration that we did have back then. I don't think there is anything more we can do. I think we genuinely listen, we've genuinely changed stuff, we're launching products. We've engaged more broadly. We've always engaged with retailers a lot. It was the smaller retailers, they're quite hard to reach, that we're now doing a better job of, I think. I should say, actually, there are there are a few areas. So if you're a small retailer selling very cheap cars, it is worth, um, recognizing that the Autotrader rate card doesn't really work on value of value of cars. So for some of those retailers, actually, the proposition might not be as economic for them because the effective cost per sale as a percentage of gross margin is high. So we are looking at, in parallel with many of the changes we've talked about, we have an economy stock package out there to reach some of those retailers that still lots of retailers will advertise those cars, but if you're not going to use Autotrader, that might be one of the reasons. And there's a bit of an economic reason behind it. So we're looking at— there's a lot more stuff that we're doing in parallel that I think will help us get there. Do you— over the marketing message? So yeah, I think there's more, there's more scope to lean into promoting more from a consumer perspective the journeys that we have available on Autotrader, but it's almost all tied up in what we're trying to do with Autotrader Intelligence, with CarBuy as their agent, more personalization, better, more relevant listings, smarter search journeys, and all of it ends with that deal packet that I talked to, and there'll be the option to reserve, request a reservation, understand financing options, part exchange. potentially for some customers' appointments, as Nathan mentioned. So we'll definitely start to evolve the consumer experience where it talks more about those bottom-of-funnel experiences. And ultimately, what we're trying to do with it is deliver more highly qualified buyers to our retailer customers. You know, it all ends with the dealer. It all ends with their ancillary products. And hopefully it's complementary to what they're trying to do in terms of their gross profits. Get along. Thank you. Morning, it's Lara Simpson from JPMorgan. Sorry, another question just on the dealer and retailer numbers. Obviously, you've spoken about the ones that have returned, but I think of the ones you lost, you said also about 54% are no longer trading. So can you just talk a little bit more around that closure? story or consolidation, and if you feel retailer consolidation is becoming a bit more of a risk going forward and how that shapes your outlook for retailers, not into the second half but more into 2022. 8 and going forward, both in terms of absolute retailer numbers, but then how that consolidation could feed into weaker product price mix. And then my second question was just on the margin. You've obviously held the core margin. I think people cost were down slightly to cushion some of the marketing spend. You obviously are talking about the intelligence platform funding itself. How should we think about the building blocks to margin from a midterm view? Are you looking to hold it and maybe optimize headcount and people costs? a little bit more to fund the step up in marketing and AI, just to understand that. And then my last question, sorry, is just Amazon Auto, I suppose, is a bit of an uncertainty as we look into the new year. Would be interested just to hear your thoughts or expectations around a potential timeline launch there, and if the risk of them coming to the market in the midterm is affecting how you're thinking about price discussions or spreads. strategy around marketing next year? Just your considerations there. I'll start with the first one. I'll do the second. Yeah, okay. Um, So on the dealership, a short answer to your question is I don't think the closures are anything abnormal, and I don't think dealer consolidation is an increasing risk at all. I've never thought it's never really. kind of surface that way. What we actually see, and the reason for that actually, is the barriers to entry are just very, very low. And where there was some potential consolidation in franchise retailers, which was more down to OEMs reducing their footprints. That's now been totally filled out, and we've run out of space for franchise dealers because the Many of the Chinese new entrants are looking to distribute things in a very— in a traditional way. And I don't mean traditional in a bad way. I think it's exactly the right way to go. Even over the past few years, if you look at the very biggest dealer groups. I think there'll be some, some will get bigger, but for everyone we've seen get bigger, we've seen some split up and spin out big chunks. And that's partly down actually to the stress I was speaking to Giles about. got— actually, it might have been Will— if you've got those legacy manufacturers, actually, you might want to think about your portfolio different. differently, and you have seen a number of groups publicly jettison certain brands from their portfolio. So if anything, the swings and roundabouts, I think, you know, we used to see deal numbers kind of reduced by around half a percent a year. This was back pre-COVID. We used to talk about that. There was never, never anything in that that we thought was like a structural thing or the fragmentation increasing. At all. These closures, we don't talk about them a lot, but whenever you look at— to be honest, whenever you look at Autotrader's retailer number, it looks remarkably consistent on the top. But when it comes to independent retailers, and sometimes we don't even know which ones are leaving and coming back as the same business, there is an opening and closing that is happening. all the time. And I don't think over that period of time it was way out of kilter with history, although trading has got— COVID was a great time, really great time, everyone was making money. It's reverted back to normal, so it's got a little bit tougher and you got hit with a bunch of costs in terms of NI and minimum wage and stocking loans. Maybe at that period of time was a little more acute, but where we are now, I think the churn levels are back to— everything's returned to normal. Churn was almost exactly the same number in the first half of this year as the first half of last year. Almost to the dealer. Sorry, you done? Yes, I'm done. On the margins, so confidence around the long-term margins. For a start, we talked already in this call about revenue growth improving. So that helps. So there is more capacity to take some cost growth. Heads were down in the first half, although it's worth really pointing out to people that most of that is down to the integration of leasing. So it's nothing to do with engineers, developers, data scientists, anything like that. They're relatively consistent. And I think I would model over the next 2 to 3, 4 years— it's always quite hard to know exactly, but I think heads relatively flat and some salary increases. As average salaries tick up, marketing should be—it's already taken a step back. That was just really because last year was very low. It'll be consistent with revenue growth, and then we've always. for almost the best part of 10 years. Those other costs, we do have a real process around how we manage them, and we really make sure that we're super disciplined on what that's going on and how we manage it. And I think the propensity for us to do that. You know, we feel it's one of the areas that we feel confident around, and hopefully we demonstrate in this first half, you know, there is practically no revenue growth and the margins haven't moved. So yeah, I think we feel okay about it. Those are the moving parts. And Amazon Autos, we're not complacent about anything. So we pay attention to all of that. You know, I think we've got a great business and that partly underlines why as we have over the years, but I think some of it's also hopefully good judgment and paying attention to these things. What we've What we know about Amazon is they're fantastic at new goods, at consumer marketing, at cross-selling, at things that fit in envelopes and things that can leverage their logistics warehouse. What a car transaction is, is a really big metal thing where people self-finance. It's used, so every single one's unique, and it typically involves a part exchange as well. So Whenever we look at these competitors, we think almost solely about the consumer side and what is the thing there that can solve a problem that we're not solving today. And then we kind of crack on with making sure that we solve that problem if we weren't doing it already. We don't see what that is in this situation. There is a thing around trust. They're working with— I mean, they will be here and they will launch early next year. We're pretty confident that that's highly likely. But what they can do for consumers is a little more tricky to see. It's very easy to launch one of these propositions with retailers. You just tell them it's free. And they'll give you some cars, especially if you're a brand like Amazon. But the only thing that matters to a dealer, ultimately, is car buys. I have been in the US now for, I think, 2 to 3 years, and we're pretty well connected with people over there. It's not something that people really talk about as doing a lot. It might be actually that they're going for their advertising opportunity, trying to collect more data on people's vehicle interests. And actually, I could see that potentially being a good lucrative space for them, but it's— that's more a Google, Meta, Amazon space as opposed to anything that we play in. Good job. Thanks. Jessica Pok from Peel Hunt. Just a few for me, please. Can I just understand right now, you know, you've got ChatGPT app, but have you actually opened your stock to also AI agents right now? In that I mean, um, the UK one, I guess Instinct, neither in the UK yet, but would you open it to me if it landed in the UK? The second one is just on— I think I might have missed it. You know, Nathan, you talked about what you can offer identically now and kind of in the future. So what is available now versus what you're planning for in the future? I think listings, yes. Vehicle check, finance, deal builder, which parts of the chain are available? Through, um, LM agents now. And the final one is just You know, obviously you're not commenting exactly on when the revenue can accelerate into the double digits, but in terms of the product launches, Um, can we expect the pace of the product launches to accelerate in the near term? So I can take the first one. Have we opened up the inventory? Yes, through MCP connectors and app integrations within some of those AI assistants. Um, have we opened our— at the moment, our inventory is not, um, is not, uh, crawlable by bots like Google, and when you're using, um, when you're using real-time inventory within an AI system, what they're typically doing is hitting a search engine. That's why it hasn't historically— it doesn't give great results for real-time inventory. It's not quick enough, and very often those search engines don't have access, um, to listings. They certainly don't have access to Autotrader listings. I think what we're, what we're flagging is our approach to search engines and generally Uh, will we open up our— if you hit our search API on an open non-password basis, will we just dump the inventory? No, we'll be far more selective than that because we think the people that are doing things well you know, there's 3 of them and they're the ones that we should work with because they're the ones that will be used. Um, we don't, we don't see any reason to take the risk of just opening up, um, Autotrader to be used kind of in a headless sense, because with those custom integrations, um, which goes to your second question, we can choose exactly, um, what we expose and how we expose it, mostly actually to offer the best experience to a car buyer. We answered it before, Jamie mentioned it before, when talking about the audience. You can expect us to absolutely focus on protecting the bottom of the funnel, and we think we probably— we will do a lot to make sure that's the case. case, but I suspect we might not have to do as much as people might think because when you use those interfaces, if you have used Instinct— well, at the moment Instinct is taking a scrape of Autotrader. Very often you get a cookie notice, which isn't necessarily certainly isn't the ideal user experience, but even within something like ChatGPT where you've got a custom integration through MCP, the results you get back It is much, much better if you've got the app installed. It's now inline just when you use the search, and the results are actually really, really good with the obvious next step unless you're willing to buy a car. one photo is to then pick and dive in. And I think that's— when it comes to the questions around audience, we are not going to pretend for a minute that more people aren't going to do that. We're all using AI assistants. It would seem ridiculous to suggest that That's not going to happen. The top of funnel has never been something that— an area where we're focused. There's a lot of other people that might get impacted by that. We really focus on now you're ready to make a decision, you want to see what's really available for sale. That's where we step in. I think that's linked to your second question, which is exactly the same thing we went through when we looked at Google. App Store and previous. The answer is we don't know exactly what we'll expose and we'll work our way through it. We'll typically— you can always step back from these decisions. Actually, you can kind of open and close things, but I think we'll gradually look at opening things up, understanding what happens from a user behavior perspective, whether it's valuable to them. Actually, this is a bit around the opportunity versus risk. One thing people haven't necessarily focused on is what happened when Google came around, is we ended up with more, much, much more audience, much, much higher market share than we ever had before. And I think AI assistance helping people that are a bit less confident to dive into Autotrader at the right place is actually a big opportunity for us to reach people that might not use Autotrader today. And believe it or not, there are some people in that bucket that go directly to franchise retailers. They generally don't use other sites like Autotrader, but there's a chunk of people there where there is an opportunity to make Autotrader stronger from an audience perspective by leveraging what they're doing. Yeah, and just on the third one, As we sort of touched on, there's no sort of grand promises or announcements around like the revenues, like pretty comfortable with where fiscal '27, fiscal '28, fiscal '29 consensus numbers are. But for the second part of your question, 100%, the pace is you will see more products from us. And it's really down to what Nathan was referring to around the AI coding agent, our ability to ship more software has pretty dramatically increased. Just add. Lost my seat there. Ed Young from Morgan Stanley. I've just got one, please. It's actually a follow-up on the minutes point that was raised for the presentation. Previous question. Your visits minutes are down, but your gap to number 2 is extended. So overall, clearly engagement in car marketplaces is down. At least there's maybe some noise. Just more broadly, are you expecting Autotrader Intelligence to reduce the amount of time people spend on site, maybe because they come in at the right point? I'm just trying to think about how we should track or think about minutes and visits as a measure of moat or a measure of overall strength of system. I know you spoke about sales influence, but obviously we can't see that, so just some thought process about how we should interpret both the current market situation and those trends over time. Thanks. Yeah, second part— well, both questions are good, but the second one's particularly good. I think yes. I mean, no one wants to talk about the weather and sporting matches during results presentation, but it was just a bit softer for whatever reason. during the course of summer, and that was a market thing. We speak to retailers all the time. We saw it in leads, and you saw it in subsequent sales as well. It's been a very lumpy period for sales, not uncommon for a retailer. to ping me in 2 weeks in and say, this month's terrible, what's going on? And then 2 weeks later to say, actually, it ended up totally, totally fine. So it has been, um, it has been a bit strange. You're right, we do focus on this, the structure. There's not been any real shift, and if anything, we've kind of gained share. So it does look like it is just a market thing which could be linked to just demand softening. We are seeing some demand measures starting to soften up a little bit. That typically just adjusts used car prices, which we are seeing in trade. But I do want to be really clear, like we know people are using the internet differently, like we all are, and that will change some of the patterns. We just don't think— we think it could actually strengthen Autotrader's position overall as opposed to weaken it. How do you work out what using AI on Autotrader means for a specific measure like time on site is a really, really hard question to answer. And I think, you know, we'll have nothing to hide on this. We're going to build it out and we think it's the right thing to do. What we'll focus on is human beings at the top and connections at the bottom. That is— they're the numbers that we really pay a lot of attention to. It's a bit like 5 to 10 years ago, Many of us spoke about page impressions, but the best way to increase page impressions is to obscure photos, for example. It's an absolutely cracking way to probably get 10% out. And time on site's the same, just slow up the servers a little bit or add a few more steps into a process. So we never optimize for it. You might then say, well, why do you talk about it? The reason is it's engagement that matters against our competitors, and it's the best measure of engagement when you look at something like Comscore. So it's admittedly limited, but it's the best one that we've got. And I think that's why we'll focus a bit more on the ratio and talk to how it looks at the bottom of the funnel. Sure. Joe from UBS. Just 3 more. I'm joking, just one. So on your $500 million buyback for '27, does that take you to your level of leverage that you see as ultimately optimal, or is that simply the level of buyback you thought was rational for this year? I'm obviously trying to figure out whether you're willing to take more leverage next year to buy over and above free cash flow. Thank you. Yes, I think we said it in the results in May. I think based on what we knew at the time, what we know now, $500 million seemed like the right number. One turn of leverage, it gets us to— that's not a maximum or a target. If we still saw it as the right decision, we would go higher. We would go higher than that point. Obviously, it's all a bit dependent on lots of factors that you take into account. But no, look, we feel good with the $300 million we've bought to date. And as I said, we expect to do the $500 million. Not an ideological cap. Time for the last one. We'll go to Sean right at the back. Sorry. Sorry, my fault for coming in a little bit late. I'm sitting towards the back there. A couple from me if I can. I guess firstly, just on the OEM opportunity that you set out, Nathan. What are the near-term steps you want to take, levers you want to pull to push that forwards? And then just secondly, as You know, maybe, maybe the way that people engage with Autotrader changes very slightly. Are you seeing any weakening in your ability to attribute sales or any changes in the way you need to collect data, interact with consumers, retailers to maintain strong attribution? Yeah, good questions to finish on. So in terms of the manufacturer opportunity, Retailer— so there are 3 parts to it. There's leasing, there's what we do with retailers, and what we do with manufacturers. And we have got 3 different focuses in those areas. Retailer and new, real new car stock on Autotrader is doing well, and it has been doing well for some time now. It's just carrying on. Penetrations over more than half franchise retailers, and I've got new franchise retailers, the new stock. So that one feels relatively easy. I know someone in the room would say that's not really fair, Nathan. But you know, I think we've got a pretty. a pretty good path laid out there. What those new cars don't have on them, and this I will admit is a little more work, is finance offers. They're cash-priced new cars. No one buys new cars on cash price. The interesting thing is lots of people don't buy used cars on cash price, but everyone uses a cash price to gauge what vehicle they're going to get. So it's still super helpful to the 600,000 people that come to Autotrader every week to look at new cars, but we think we can go a step further by telling you what the actual deal is, because very often the deal on a new car is buried into the finance offer through a deposit contribution or APR subrogation. There is— that's really— that finance focus actually does spread across all of the new cars. So when we talked about a deals hub, we've got the leasing deals, which is essentially a finance offer. We need to work on getting the finance offers on the physically available new cars, and then we can compare, compare the two across HP, PCP, and leasing. And then we'll be able to say to users, this is this place, this is the only place where you can get the best deal across those 3 measures. That is about finance integration. And it is about probably getting more retailers on. As you might imagine, finance companies are not the quickest to integrate with, and to be fair, they've got to deal with regulation and compliance. But continuing to roll that out is something we've been making progress on, and we're going to double down and then build into that. To your question, what do we want to do in the nearer term? is to allow those to be compared across one another and in a single place, ideally with some measure or flag. I'd be like what we do with price flags as to which ones might be the best deal. for a specific consumer. Then with the OEMs specifically, there's lots of new OEMs. They've all got brands that no one knows in the UK. I would say they're all very pragmatic, and so they're looking— they see Autotrader as being a place where they can quickly get known for cars. So making sure that we've got the relationships and all the solutions to allow them to do as much as they possibly can or want to do on Autotrader partly underlies the stronger M&A numbers that Jamie spoke about earlier, and there's an opportunity for us to just do more than that. More of that, sorry. Then it's about leads and lead generation, and that's the bit where it's still a little bit nascent, actually. A lot of OEMs measure all their effectiveness on inquiries. So how we turn that activity into hard leads for an OEM specifically, not a retailer, is kind of that third area of focus. On the final question, we are getting close to time. The changes with AI and consumer patterns, does it impact our ability to attribute sales. I would actually say, and happily show any analysts that want to see this, our attribution has just taken a very big step forward this year, literally in the last few months. Jamie talked about match sales, which is where we're provided the data for a retailer's sales and we can match it up very specifically, not just with a user and say, oh, this postcode looked at this car, and go to a much more granular level and give them confidence as to, yes, it was Autotrader that influenced this sale. We'll give them time date stamps on what they did, who else they looked at. And then it gets a propensity score of the likelihood that we've influenced it, influenced that sale. We don't really get pushback from retailers anymore on that. Now, the interesting thing is that that is on match data. What we've now done, again, to labor the point perhaps more than I'd like, using an AI model, what we now do is model that behavior. And apply that to every single dealer on Autotrader, whether or not we get their sales data. So we provide a predicted sales number whether or not we have your sales data. And that also provides you with exactly the same granular level of detail, and we can start having that conversation with dealers. So we're no longer dependent on sales data coming in to be able to predict sales, and it opens the right conversation with the retailer if they say, Well, you've got my sales number wrong. We just say, we'll provide you match data and we can give you all the insight. And then they can get insight into geography, where they're selling cars, what are the people doing, what's the pattern of behaviors, what they might want to source. So, um, It's not something that we've talked about, but I'm super proud actually of what the team have done there. I think it is some of the best that I've seen from marketplaces around the world, actually. Great. Thanks everyone for joining this morning. Thanks a lot.