Hello, and welcome to the Angi second quarter 2026 earnings conference call. Today, all participants will be in a listen-only mode. Should you need assistance during today's call, please signal for a conference specialist by pressing the star key, followed by zero. After today's introductory remarks, there will be an opportunity to ask questions. To ask a question, you may press star, then one on your telephone keypad. To withdraw your question, please press star, then two. Please note that today's event is being recorded. I would now like to turn the conference over to Julie Hoarau, Chief Financial Officer. Please go ahead. Good morning, everyone. I'm Julie Hoarau, the CFO of Angi Inc. Welcome to the Angi Inc second quarter earnings call. Joining me today is Jeff Kip, CEO of Angi. Angi has published a shareholder letter, which is currently available on Angi's website in the investor relations section. We will not be reading the shareholder letter on this call. We will go through a few introductory remarks and then opening up to Q&A. Before we get to that, I'd like to remind you that during this presentation, we may make certain statements that are considered forward-looking under the federal securities law. These forward-looking statements may include statements related to our outlook, strategy, and future performance, and are based on our current expectations and on information currently available to us. Actual outcomes and results may differ materially from the future results expressed or implied in these statements due to a number of risks and uncertainties, including those contained in our most recently quarterly reports on Form 10-Q, our most recent annual reports on Form 10-K, and in the subsequent reports that we have filed with the SEC. The information provided on this conference call should be considered in light of such risks. We'll also discuss certain non-GAAP measures, which, as a reminder, include adjusted EBITDA, which we'll refer to today as EBITDA for simplicity during the call. I'll also refer you to our earnings release, shareholder letter, our public filing to the SEC, and again, to the investor relations section of our website for all comparable GAAP measures and full reconciliation for all material non-GAAP measures. Now, I'll pass it off to Jeff. Morning, everyone. Thanks for coming to the call. We're very happy with what we've been able to accomplish on our strategy over the last three months, and we believe we're well on track. Julie's going to give some commentary on the numbers, and then I'm going to come back and discuss our strategy somewhat comprehensively. Julie? Thank you, Jeff. Starting with our revenue. Our revenue for the second quarter was down 11% year-over-year. Three things drove it. First, we continued shifting away from our network channels. Network revenue was down 34% year-over-year. Second, we stepped back from lower quality-based marketing channels. We had ramps in Q2 last year, which makes this a harder compare. Third, our revenue in Q2 was materially impacted by a shift in mix and traffic starting about 10 days into March as oil and gas prices rose sharply following global events. We noticed that homeowner behavior changed, and demand moved away from larger jobs category, like roofing and HVAC, and that's where we have the most available pro capacity, towards smaller jobs, where we have less capacity, and that less capacity in those categories are unmonetized. We surveyed our homeowners consistently over the last few months. In April, we saw more jobs being canceled or postponed. More recently, overall spend is back at expectations, but each job is more expensive, so homeowners are doing fewer jobs, which ties right back to the lower demand we are seeing. We can see all of this in our metrics for the quarter. U.S. service requests are down -6%, while leads are down -13%. Our revenue per lead, however, is up +1% year-over-year. That's due to the mix-out of the old heavily discounted legacy ad product we had. As a result, compared to historical seasonality, the second quarter revenue is in the range of double-digit percent lower than on pre-March 11 run rates. We have seen things trending positively, but we remain below our pre-March 10 run rates, and we do expect modest improvements as the future quarters progress. On the profitability side, sequentially from Q1 to Q2 2026, revenue grew and adjusted EBITDA grew faster with about 50% flow-through. We have improved our marketing ROI, and we have reallocated about $6 million of inefficient TV spend in the second quarter towards higher ROI channels. TV has been less effective this year than in prior years, so we will be back. We do expect to take out a comparable amount moving into Q3. We remain on target with our overall adjusted EBITDA minus CapEx range for the year. Jeff said on the last earnings call that we would be happy with about $50 million a year of EBITDA minus CapEx, and we are right on track for that. July EBITDA margin is already several hundred basis points above Q2, and further reduction in ineffective TV spend should support a stronger Q3. I also want to discuss the impairment that we registered for goodwill and trade names. We recorded a non-cash charge of $235 million this quarter on our U.S. reporting unit. The trigger was the sustained declines in our market capitalization since year-end. That requires an interim test, which the test compares the estimated fair value of each reporting unit to its carrying value. We tested both reporting units, and the charge is entirely on the U.S. reporting unit. International came through with substantial headroom. This is not a reflection of our business on a consolidated basis. There is no effect on cash liquidity of our commitments. Looking ahead, our annual test for goodwill impairment is in October, and earlier if there's another triggering event. Further charge is possible if the market conditions, the valuation assumptions or the operating performance deteriorates. That's a function of the accounting rules. We've disclosed the sensitivities in the 10-Q. In closing, we're not reinstating guidance at this time. Our focus remains on building out the strategy Jeff described in the shareholder letter. We have laid out the reasons to believe, and we expect that to show up as acceleration in 2027, as Jeff explained on the last call. I will now pass it back to Jeff, who will go into more insights on our strategy. Thanks, Julie. Let's get up in the helicopter and look at our market opportunity and our strategy to win it comprehensively. I think as many people know, we estimate the overall market for completed home services work in the U.S. at about $700 billion. That is the total revenue available to our Pro customers. We estimate that less than 1.5% of that flows through our platform. We have a material opportunity to continue to penetrate that significant market. Our core lead business targets the $70 billion - $80 billion of total customer acquisition spend that all Pros make across the United States. About 65% of that spend comes from Pros with more than 20 employees. We have less than 0.5% of that market, and probably in the range of 4%-ish of the 35% of the market that is small to mid-size businesses. We under-index significantly against the large pro segment. If we achieve comparable share in the large pro segment to what we have in the small to mid-size segment, we'll reach something like $2.5 billion in revenue, and we think that's a very reasonable target for our existing core business. That, of course, ignores any improvements in Pro lifetime value and engagement. Reducing churn, which is a focus of our strategy, for example, by 25%, all else equal, would add 10 points to our annual growth rates versus what we can do otherwise. Executing on both opportunities, i.e. penetrating the large pro market and reducing our churn, will put us within striking distance of the $5 billion in revenue we talked about in our last letter. Before we penetrate the Pro software and services market at all, that market for Pro software and services to run their businesses and close down their leads is, we estimate, about the same size as the market for total spend on marketing and lead acquisition. We have a material opportunity in front of us. Our strategy is to be the trusted revenue partner for Pros and go after both markets. Effectively, the entire marketing and lead acquisition market for Pros and also the services and software business. What gives us the right to win in this $150 billion revenue market? Well, first, our core leads business. This is our competitive wedge in the Pro revenue cycle. We play a key role today at the top of funnel for hundreds of thousands of Pros, and with the improvements in lead quality and win rates we've made over the last couple of years, we're consistently improving our competitive position. Secondly, our market-leading distribution and customer acquisition assets. We have over 100,000 active Pros in the United States, a network any software and services company would love to have for distribution. We'll also acquire more than 70,000 new marketplace Pros per year in the coming years. Also a great distribution opportunity. Thirdly, our 30 years of brand equity in the industry, which opens many doors. Fourthly, our ability to generate cash to fund our strategy and continue delivering on our commitments. Finally, our AI strategy and development capabilities, which are already producing results. We have a homeowner agent already touching 50% of our homeowner traffic, converting that traffic at three times the rate of traffic that doesn't touch it, and contributing to our rise in success metrics. We've deployed our first Pro agent, the AI Front Desk. In just a few months, we're now in the market and booking appointments already at a solid baseline rate when compared to human call center performance that we observe. As we said on our last call, we believe that we're in the middle of the greatest technological transformation in a generation. We believe that AI affords us the ability to build products which greatly improve both the experience for and the success of our customers, and build them much faster. The three core footings of our overall strategy are, one, return the core business to growth through large pro market segment penetration. Two, finish building and migrate to our new AI-first single platform. Three, drive Pro win rate, success, and revenue through our AI strategy, consisting of A, the Angi Pro Chief Revenue Officer agent suite, and B, our Homeowner Agent. Let's walk through them one by one, starting with the large pro segment. We have a 10x opportunity in the large p ro segment by simply matching our small to mid-size segment penetration. We're already acting with velocity here and are watching the segment grow more than 20% year-over-year with less than a third of our fully staffed headcount in place year to date. How are we doing this? First, we're progressively building out a fully enriched target database and leveraging it. Secondly, we're putting the right team in place. We've achieved our growth rate to date with less than 10 sellers and will reach 30 by year-end. Thirdly, we've changed our go-to-market from here's a bunch of leads and here's a volume discount to an operating partnership. We make sure each Pro is set up to win with the right software and operational approaches, and we work through their lead to close funnels with them on a regular basis. Finally, the truth is that our core lead product is a better product market fit for large pros because, one, large pros already work against a high volume of leads, many different lead types, and they're focused on their overall cost of marketing versus won revenue rather than winning or losing each individual lead. Two, it helps that we've invested so much in our lead quality and win rates. We see our win rates up roughly 20% from a year ago and even more than that versus two years ago. We think we've gone from pros winning roughly one in nine leads two summers ago to roughly one in six now, and our pros experience this and lean into our product. Why couldn't we do this before? I'll take the blame. My first year in the job, we were looking at the wrong data with the wrong team, which meant the wrong execution. We started taking the segment and our operations apart a little over a year ago, and we've put everything back together to get to the trajectory we're now on. We're seeing real results, and we expect to accelerate from here. Let's talk about our progress on moving to a new AI-first platform. We've talked plenty, and we've covered all the ground regarding the limitations of our legacy technology. We froze the old stack, and we're now in full flight with the build of and migration to our new platform. We're building all new software and technology AI-first, meaning set up to deploy AI and our data assets across all product and platform surfaces. We're already hitting milestones in our re-platforming execution path. Our homeowner account experience is now live on the new technology. It's not visible to the eye because we've maintained the design and the UX. We've also implemented new messaging technology. It's the same technology which drove greater engagement and success when we deployed it internationally. We'll deploy AI-driven UX in the future on this surface, suggesting, curating, and automatically sending messages to get from contact to closed job, again, improving the experience and success rates for our core business. We expect to finish both building and migrating our homeowner experience to the new technology by year-end, and then we'll start iteratively improving that experience AI-first. At the same time, we've started working on our new pro experience platform, and we're targeting migrating our first test cohort by the end of the first quarter of 2027. Across all of this work, we're simplifying and removing friction from the product and customer experience, and we're merging the international and U.S. systems, creating a best-in-breed hybrid. It is worth noting that starting around six months or so of tenure, pro churn on the international platform is about half that of U.S. rate. We believe we can capture a chunk of this benefit through both platform migration and our Angi Pro CRO. Again, if we get half that delta, we'll have a 10% tailwind for future growth. We just need to execute. This is a core opportunity for us. It's not yet guidance, though. Let's talk about our AI strategy. There's two core topics to talk about. One, how AI is changing the traffic acquisition landscape. Two, how we plan to leverage AI strategically in that changing landscape. First, in terms of the landscape, we would say that AI today is compressing the value of surfacing information and discovery and impacting where homeowners look for help. LLM engines are taking a growing share of the informational searches that historically brought homeowners to marketplaces like us. The most visible near-term pressure is on unbranded organic search. However, Google's been putting its own pressure on unbranded organic search for years now. Effectively, they've reduced our reliance on their free search traffic. Our unbranded SEO channel is down close to 5% of our total service request volume, and our plan does not assume recovery there. What's our plan for LLM traffic? We intend to be present wherever demand forms, and we intend to match those homeowners to our pros on Angi through traditional search, social, and increasingly through LLMs and personal agents. We believe that at the same time AI is commoditizing informational inquiries, it's also increasing the relative value of matching homeowners to the right pro and getting the job won and done well and creating the data to reinforce that loop. As a side note, we're actually doing reasonably well with LLM share of voice. Our most recent data says we're at the top of the industry and double the share of our closest competitor. Share of voice on LLMs is not where we believe the action is, because it does not deliver conversion the way it does in SEO. Instead, our objective is to provide the pro supplier fulfillment layer for the industry on LLMs, Google, social, and everywhere else, which we always have, this objective now requires new tools and a new strategy. We've already built systems that can have a natural conversation about the work someone wants done in their home, in the homeowner's language rather than ours, in any channel. We can pick up the conversation at any point on any surface and either ask more questions based on context or directly surface pros, again, based on context. We can drive a better match with this better context than our proprietary data and knowledge of our pros and their preferences, skills, and availability. We've already built this with our ChatGPT app. We'll do this for Amazon Alexa, and we're working on other significant integrations, multiple of which we believe we'll announce soon. We're also doing this by buying ChatGPT ads. We're now spending profitably in that channel, and we're at a roughly $3 million revenue run rate on that platform. That sounds small, but a year and a half ago, our Meta business was half that, and we're now approaching a profitable $100 million revenue run rate on Meta. We're optimistic about the marketing opportunities on LLM surfaces. We believe ChatGPT will grow and that Google will continue to offer significant ad inventory. Along the same lines, we've already developed and deployed using the same approach, our Homeowner Agent 1.0, which we've called our AI Helper to date. As we've said, 50% of our customers use this agent, have improved success rates, and we're building data off this usage to deploy across all our other surfaces. We expect to further develop the homeowner agent going forward and deploy it deeper in the funnel to clarify project details, provide cost ranges, identify appropriate and available professionals, and move towards contacting and booking appointments through voice or text through this agent. Critically, this will drive better matches and jobs won well for our pros and deliver outcomes and data to win more homeowner traffic. I've put us already into the middle of the second topic, which is our AI strategy. Fundamentally, the underlying job won well and job done well both still require a match and a still skilled Pro, and this is our role as the supply and fulfillment layer in the industry. AI may be shifting the metaphorical front door for the homeowner to walk through to gather information, but it is not going to eliminate what has to happen after that threshold's been passed. The right pro match still needs to be found in terms of preference, skills, and availability, and the right pro still needs to understand, assess, price, schedule, win, and complete the job. This is where both our assets and our strategy position us to win. We have the pro capacity to complete more work than any other marketplace. We have years of proprietary reviews, matching, and job completion and cost data, and we can make sure that a homeowner searching on an LLM finds not just the standard pros that their model surfaces, but the right pro for that task at the right time when the homeowner needs it to be done. We've got the best engine in the industry to acquire, recruit, and onboard pros, screen where applicable, understand their skills, preferences, and service areas, and know whether they're available. Now we're expanding that toolkit and that engine and our strategy, and we're leveraging AI to be the full trusted revenue partner of the pro and provide the tools that allow pros to win work wherever homeowners are searching, in turn, creating that supply and fulfillment layer for LLMs and all other surfaces. As we've said, our core lead business is our competitive wedge. $35 billion of annualized job volume enters our platform. Our challenge is that only about $10 billion is completed by Angi pros. Thus, we need to build agents to help pros already receiving this demand convert more of it. The homeowner agent will play a role here, but more importantly, we're building the Angi Pro Chief Revenue Officer, which is an AI-driven revenue system that performs the high-effort, cumbersome work between receiving the lead and winning the job. The Pro Chief Revenue Officer will respond immediately to the homeowner, answer calls, schedule and optimize appointments and routing, provide sales coaching, prepare estimates, and follow up consistently, leading to more winning. More winning equals more retention and greater lifetime value and more pro capacity in our supply layer to serve homeowners on any surface. Again, this is real opportunity. Our best evidence looking across our businesses over time points to doubling win rate, cutting churn in half. Again, not guidance, but data we see on our platforms. When we look at the results larger customers have had with some of the AI call center businesses that have gotten out there first, we see that their win rates have as much as doubled, so we're very optimistic we can replicate that. The Pro Chief Revenue Officer will also ensure that we know the pro skills, preferences, success data, and availability, and leverage that data and information to match each pro to the right customer jobs across all surfaces. Again, we're already doing a version of this, but we will able to do so at even higher fidelity once we implement our strategy. Again, better matches equal more jobs than done well, more data, and a flywheel that wins. As we said in the letter, our first agent, the AI Front Desk, is live. We have dozens of pros on board, and we've made dozens of appointment booked already at rates within the range of what we see from human call centers. Around 80% of the pros we've onboarded still are using the service, which is a good rate for an MVP pilot. We were not expecting perfection. We're very happy with our progress. We now want to iterate from good to great and start to scale up. Our next agent will be a receptionist receiving calls on behalf of pros with the ability to ask questions and do more than just book an appointment. Alongside that, we'll be thinking about schedule optimization and routing, and then likely start looking at the visit itself with quoting and sales coaching functionality. As noted in our letter, we plan to demo the full Pro CRO 1.0 suite, both live agents and prototypes at our investor day on November 17th. We expect that the Angi Pro Chief Revenue Officer will change the experience and the economics for everyone. The homeowner will be more likely to get the job done, delivering outcomes. The pro will win more and enjoy a better return from Angi, driving retention and pro capacity for our supply and fulfillment layer. Angi will build a deeper network and more data to improve the next match, appointment, and job. We'll also have flexibility in how we monetize the relationship. In our core product, that could be per lead, per appointment, or per job won, and we'll also have the opportunity to earn more revenue through both pro LTV. Higher winning means higher retention, means more revenue, and we may also elect to charge a usage or subscription fee for our agents, building a whole new revenue stream for the business. Again, we're very happy with our progress to date, and we're very optimistic about our opportunity to win the significant market in front of us. The landscape may be changing, but we believe we're well positioned. One, to win the large pro segment with our core business and our new go-to-market. Two, to improve the customer experience and our ability to innovate effectively by getting to a new AI-first single platform. Three, driving greater customer success, stickiness, retention, and repeat with better outcomes, more jobs won well and done well with our AI strategy by both building the Angi Pro CRO suite and further developing our homeowner agent. With that, we'll take questions. Thank you. We will now begin the question and answer session. To ask a question, you may press star then one on your telephone keypad. If you're using a speakerphone, please pick up your handset before pressing the keys. If your question has been addressed and you would like to withdraw it, please press star then two. At this time, we will pause momentarily to assemble our roster. Today's first question comes from Dan Kurnos with StoneX. Please proceed. Great, thanks. Good morning. One for Julie, one for Jeff here. Julie, you said that the trends improved exiting the quarter. We can all, I think, it's been a long earning season, but we can all kind of do math, and it implies sort of some sequential improvement throughout the balance of the year. How should investors think about that sequential cadence on revenue and margin from here? For Jeff, you've talked a lot about this kind of year of steady improvements, before we see your strategy, the AI strategy, begin to accelerate that growth into 2027. A, how much of that depends on large pro? You gave a lot of details in your prepared remarks around how that's initially tracking, but just some incremental color there would be helpful. B, obviously, I'm assuming we're targeting growth in 2027, but I don't know if you wanna commit to that. Thanks. Thanks, Dan. You're right. As I mentioned earlier, there will be modest improvements as the quarters roll forward. That means we expect maybe a little bit of improvement each quarter, but we're not guiding on it. Regarding margin, as I said earlier, July was strong. Q3 is looking good, but Q4 usually comes down a little bit due to seasonality. As we look ahead, we think that the large pro opportunity is really core to growing again. We think the SMB business will level out at some point, particularly as we get online enroll live on the new platform and ramp. We do think that the large pro opportunity is the key to growth again. As you sort of pointed out, we're not committing to timing, but if we execute our strategy, we should be growing again sometime in 2027. Again, not point guidance, not anything, but we feel good about our momentum. We feel good about the opportunity, and we think we can really double down and drive significant growth through the large pro segment. Great. Thanks, guys. Thank you. The next question is from Brad Erickson with RBC. Please proceed. Good morning. Thanks for taking the questions. I guess two kind of related questions. First, you gave several metrics in the letter around just kind of the core blocking and tackling of the business that are all generally showing improvement. If you had to focus it a little bit, what do you kind of view as the most instructive metric or two as we think about this return to growth. The second, what is kind of like the specific bottleneck on that return to growth? Do you need a certain level of agent adoption at a certain level, or is it just more as simple as completing the traffic cleanse you've been kind of going through? What are the most important gating factors if you were to hit that 2027 kind of rough, not guidance, but target that you just mentioned? Thanks. Let me take that, Julie can correct me or add if needed. I think the core thing we focus on in our customer experience is win rate. The inverse of that is homeowner job completion. Our pros are our paying customers, the more they win, the more they stay. That is a really critical overall success metric. In terms of the business, we're focused on overall capacity growth. We cited a few metrics surrounding that. At the end of the day, we want to acquire more capacity than we churn, and as we build capacity, we point ourselves back to having the capacity to grow the number of leads and grow the revenue in the business. That's why moving from down 13% in the first quarter on pro capacity to down 2% as of June, and we're about flat in July year-over-year, is really important to our future trajectory. Having that pro capacity is the most important gating thing in terms of growing in the future. If we don't have pros with capacity to pay for leads, we can't market into it, we can't grow the revenue. The biggest gating items there are, A, Dan sort of hit it earlier, our ability to penetrate the large pro market at the rate it looks like we can penetrate, given our extremely low penetration now. Then, of course, our ability to keep driving that win rate and pro experience, which then increases retention and thus increases the number of pros and the capacity available. Our strategy is built around those two objectives from the large pro go-to-market to the new platform and the Angi Pro CRO. In terms of timing, Jeff said on the last earnings call in May that we'd focus on our strategy, as I mentioned earlier, probably seeing by that sequential improvements. He also mentioned that in about a year, which puts us around May next year, we'd start to accelerate. We're not guiding, we're not putting a date on it, assuming we execute our strategy, we'll be growing again at some point in 2027. Understood. Thanks. Thank you. The next question comes from Tarini Padmanabhan with UBS. Please go ahead. Hi, this is Tarini dialing on for Stephen Ju. I have two questions. First, can we talk about pro capacity? You've talked about it, but what's driving the growth there for you? Should we be thinking of the growth here as Angi through putting higher quality leads to the SPs versus other solutions they can be using? Second, as you roll out agents for pros, you said that you may monetize either through improved core business retention or LTV or through a feed model. What do you think the revenue model could be or should be? Sorry, I didn't get the first part of your second question. I got the part about the revenue model. What was the first part of your second question? Yeah. As you roll out the agents for the pros, you said that you may monetize either through improved core business retention and LTV or through a fee model. What do you think the revenue model could be or should be? Okay, great. I can take the first one. Yep. We define pro capacity as the budget available from each pro or the actual spend if they don't have a budget, but most of our U.S. pro base has set budgets. When we're growing our pro capacity, that means we're adding to our ability to buy service requests and monetize that capacity. We don't think about how much pro capacity we have in relation to the market. We're currently utilizing about two-thirds to three-fourths of the total pro capacity right now. As Jeff said, our revenue represents less than 1.5% of the total marketplace. We're likely still below 2% in terms of total pro capacity. We're currently growing our total pro capacity, this is despite the fact that our nominal pro count is down, that's because we're growing our average per pro capacity, which grew about 13% year-over-year this quarter. We're doing that by retaining larger pros and targeting and acquiring larger pros, that ties directly to our large pro strategy. Let me talk about the agents and economic models and maybe even some sort of constructs on models you might use. First, our first priority is that we are trying to improve pro win rate, drive the pro revenue cycle, and thus become the trusted partner of the Pro. We do that, the revenue's going to follow. I think one way it will follow is what I outlined in my remarks, which is driving win rate consistently drives retention. Again, we have evidence that you double the win rate, you may cut churn in half. If I can improve churn by 25%, that's a 10% annual growth tailwind. If I can change it even 10%, that's a 4% annual growth tailwind. Obviously, both of those are key monetization. You would say pros are pretty used to paying for software to help run their business. When we think about there's a couple of ways that this can manifest itself. One is, if we're overall driving the success of our core product, we have pricing ability. Two is, we can charge a usage or a flat fee, and it sort of doesn't matter, but to get to an average monthly amount. We could charge that for our leads, we could charge that to pros to use our software and agents for other leads. Your software works really well for Angi Leads. Can I use it on my Google LSAs? Can I use it on my inbound phone calls? Can I use it on one of your competitors? If you think about it, our average small pro is paying about $600 a month for 12 leads, and they're winning about two of those. If they start winning one more, i.e., a 50% improvement in their revenue, would they pay $50 a month in either usage or fee for that? Would they pay $100 to use it for other platforms? Maybe. I think it's actually pretty fair, and I think it's pretty reasonable if they're getting much more value that they pay for where they get the value from. If you want to think about a hypothetical model, I just lay this out because the real power in the business model here is our existing distribution. We are building agents, i.e., software, that goes right next to and with the core product we're already selling to wedge ourselves in the revenue system. It's a natural add-on. It's like getting some fries with your burger and a soda, like a meal deal. We tag that in, and we send it out, and it's almost costless CAC. With 100,000 pros today and then 6,000 pros a month, you then just get into, if this is a freemium product, how many convert to paid, and then what's the retention? On a very simple level, you can build out a model that says, once we roll this out, which is probably not before next year and more likely in the second quarter or later, where am I in 15 months? If I put 100,000 pros on my platform and I keep 25%- 33% of them, and I keep 25% - 33%, that's the conversion to paid of all the new pros, and I have something like, I don't know, a 3% monthly churn or a 2% - 4% monthly churn, you can imagine getting to numbers that look like 30,000 or 40,000 paying customers after 15 months simply because of the installed distribution and the natural synergy of these agents with our lead product. If I have 30,000 - 40,000 customers at $50 a month, I start to have a decent monthly run rate in revenue, and I also have the opportunity to upsell these people into larger packages. Pros are already paying $1,000, $2,000, $3,000 a month for software packages to run their business. We think that at lower price points, given our CAC, we can actually build a pretty decent recurring revenue base. Again, I think you can build that model and do the math yourself, but I think it would be a very nice lift on our existing revenue and profit base, and give us something really to power the business going forward. I think those are a couple different ways to think about it. We're really excited about it because, again, we think we have the assets to really make this work, and we've already got proof of concept with our first agent. Great. Thanks. Thank you. Our next question is from Sergio Segura with KeyBanc. Please go ahead. Great. Thanks for taking the questions. I had a couple on consumer marketing and traffic. Consumer marketing did increase as a percentage of revenue, just in the quarter, just curious how you're thinking about the trade-off between growth and efficiency today in the current macro environment. Then, if you could provide any color on what portion of your marketing spend is variable, and can be dialed up and down based on kind of market demands versus investments that are more fixed in nature. That's question number one. Then the second one, related on to ChatGPT. You talked about the growing traction there. Can you just speak to the economics of that channel relative to traditional search and what you've learned so far about the quality and conversion of traffic through ChatGPT? Thank you. I'm going to go in reverse. The economics on ChatGPT are we're making a nice profit margin. It's not quite as good right now as Meta or Google, but it's also at very low volume and early stage, and we moved this thing from losing a little bit of money to making a little bit of money pretty quickly. ChatGPT/OpenAI, they're optimizing, and they're continuing to work through these tests. It's making money, not quite as much we actually anticipated, that it'll probably be similar, because that's what it'll take to compete in the market with the other platforms. In terms of marketing spend and efficiency, we have a pretty rigorous discipline around making our last dollar by break-even margin. That's probably easiest in Google, where we can work through their profit curves, and we can work with their interface. We exercise that discipline with consistent analysis and lean testing across other platforms. Our marketing spend is variable, but we're going to buy until we're not making any money, and we're not going to pull it back to make more money because if we pull it back, generally that means we're making less money. The really critical piece is having the pro capacity to buy SRs that match into that capacity and generate revenue by creating leads. That's really how we drive it. I think the exception is probably our brand marketing, where you have two pieces, which is TV and non-TV, which is largely social. We also look pretty consistently, it's just not as easy to read as quickly. We look pretty consistently at the ROI of TV. We use iSpot, and we look at the response to the ads, and we triangulate. Our TV was not working nearly as efficiently as it has in years past, according to our prior analysis. We've pulled that back, Julie mentioned significantly, $5 million or $6 million from Q1 to Q2, and we'll probably do about the same in Q3. We're going to go back to the drawing board and look at our channel and daypart mix, and we'll look again at our creative. We try and exercise the same discipline on TV as we do on performance. There's just a little bit of lead lag there. On social, we've been able to drive a lot of impressions and traffic with our social brand activity. This isn't big dollars. There's several million dollars there. We do have the ability to pull that back if we don't think it's working, but we also continue to believe that we need to keep our brand impressions and our market leading brand awareness out there. We're always balancing that. We're not aggressive there. We could pull it back a little bit, but we think it's important. I think maybe I got all your questions, Sergio. You can let me know if I missed something. I think the only thing you may have missed is just if you could talk about the quality and conversion of traffic from ChatGPT, that's any different from your other channels. We don't see anything worse. We don't see anything better. It's a little hard to read at the volume we have because our sample size isn't really big enough to distinguish it, but we see it tracking with other channels right now. Understood. Thanks, Jeff. Thank you. The next question comes from Youssef Squali with Truist. Please proceed. Great. Thanks for taking the question. This is Robert on for Youssef. What's the plan to drive more traffic and jobs to the platform as capacity builds? What are your capital allocation priorities over the next year, and how are you thinking about your bonds? I can take the capital allocation priorities question. The last big capital allocation decision was to buy in bonds at a discount. We're obviously looking ahead at those bonds coming due in two years and going current in a year. Thinking about our bonds and our refinancing is our top capital priority right now, and we'll take care of that in due time. In terms of driving more traffic and service requests to the platform, we believe we have ample opportunity in our existing channels with the expansion of Pro capacity. We've had a mix shift from last year and the beginning of the year, which has limited the volume we can get at the CPAs that break even with the capacity we have. As we expand our Pro capacity and we fill in task and location, we believe there's ample opportunity to scale in our existing channels. On top of that, we are always testing new channels, looking at new partnerships and so on. The most promising channel we have right now is our ChatGPT test, where we're scaling that up bit by bit. We're reaching profitability, and they appear fully engaged in growing that business. We think there's real potential there. A channel we haven't made work yet from an economic basis, but we continue to test, is TikTok. We look at the other major platforms. We also have multiple partnerships. For example, the partnership we have with Anywhere Compass Group. After their merger, we're still executing there to bring in jobs through their agents. We have multiple other partnerships, including our retail partnerships with Walmart, Wayfair, et cetera. We have an active business development opportunity. We have some real opportunities there. We're continuing to look at the LLMs and the new platforms. We believe we're in position to grow into our capacity as the year goes on within the approach we've outlined. The last thing I'd point out that's important is that our repeat rate has been running up 20% over the last couple of quarters, and that's pretty critical in terms of supporting our brand traffic going forward at lower TV spend. When that repeat traffic comes in through paid channels, it improves our conversion and cuts our CPA and allows us to spend more. I think that one of the sort of biggest hidden turns in our business over the last quarter has been the return to growth in homeowner repeat, which we're pretty excited about, and we think that that's a valuable asset in our whole mix as well. Thank you. Thank you. The next question is from Eric Sheridan with Goldman Sachs. Please proceed. Thanks so much for taking the questions. Maybe two, if I could. The first, building on the comments earlier about the improvements you saw in July, is there any way to sort of tease out how much of that might be an easing on the macroeconomic headwinds that the economy was broadly facing relative to some of the improvements you're trying to drive into the business organically? That'd be number one. When you talk about the migration and some of the investments you want to make to be an AI-first platform, can you just refresh whether some of the commentary we're getting today on those investments also fits inside the parameters of the annual cash flow framework that you gave last quarter? Thanks so much. The answer to the second question is yes, it's all inclusive. We're not talking about incremental headcount or investments. The answer to the first question is, of course, nuanced, which is, we do believe we've seen recovery in the homeowner, not fully. We think we still have some mix in traffic impact. We also think that a significant piece of this is driving our Pro capacity back towards growth. Our utilization remains down below where it was tracking previously. That continues to slow us down. We think that is mix and availability at the same number of leads for SR. We think there's a bit of a nuance in there, which is in response to the shifts in mix and traffic, we've sort of grinded through and retuned our marketing machine. We think we're making a little more money on a little less revenue. Apples to apples, we're actually doing a bit better. I think I'd sort of slice and dice it that way, if that helps, Eric. It does. Thanks, appreciate it. Thank you. This does conclude our question and answer session for today. I would now like to turn the conference back over to Jeff Kip for any closing remarks. Yeah. Look, on a very simple level, thank you to everybody for joining and following along. We're obviously very excited and optimistic about what we've been able to do in the last few months. As we look at the opportunity in front of us, we're accelerating our strong momentum in the large pro segment. We're on track with our new platform work, and we think there's real upside as we complete each piece. We're growing our LLM presence, and we're also making real progress with our homeowner agent. We have our first Pro agent live and performing really above our expectations. We expect to move forward with all deliberate speed there. We're on track with our strategy, and we're looking forward to accelerating in the quarters to come. Thanks everybody for your support. Have a good day. The conference is now concluded. Thank you for attending today's presentation, and you may now disconnect.