Nextdoor Holdings, Inc. (NXDR) Earnings Call Transcript & Summary

All right. Good afternoon. I'm Jamesmichael Sherman-Lewis on the Citi Internet team here, and I'm thrilled to have with us today Nextdoor's Co-Founder and CEO, Nirav Tolia. For those newer to the story, Nextdoor is a local social app with a user base spanning over 100 million verified neighbors and the company's platform transformation, NEXT, just launched in July. Welcome Nirav. Thank you for joining us today.

Thank you for having me.

So Nirav, you've been back as CEO for about 1.5 years now. We're moving towards this new product-led growth culture. Help us understand that mission and how you're balancing transformation as well as building on the existing Nextdoor business.

Sure. So I'll start with talking a little bit about what we think the opportunity is, and that is to build the indispensable essential local application. That doesn't exist today. We don't go to Google or any of the Meta properties or ChatGPT to find out what's going on around us. That one place where you can find out immediately the things you need to know about where you live, that doesn't exist. It's been the mission for Nextdoor to try to be that entity from almost the very beginning. We started the company in the summer of 2010. And we think that we are now approaching a time where we have a large enough audience and there's enough content out there and there are technology tools like AI and other things that will enable us to be that place where, through the power of technology, you can quickly understand all the things you need to know about your neighborhood, your local community, where you live. And so that's the mission of the company. Now you mentioned product-led growth. A lot of people say, why haven't you been able to achieve this mission, right? And in Nextdoor's case, I came back 18 months ago because the company was not in great shape. We've been growing steadily for 9, 10, 11 years. We went public, and then we experienced some headwinds. I think the main reason for that is while the technology industry always evolves, products need to evolve as well. Nextdoor's product did not evolve. And if you don't evolve your product, your users ultimately don't get enough value from that product and they don't actually inspire other people to use the product, and that's what product-led growth is. Product-led growth is having a good enough product that the whole thing grows and ultimately, the whole thing means revenues and profits, because of the quality of the product, not the quality of the marketing, not some strategic thing that you figured out that locks you in as the only choice for consumers, but you have such a good experience and such a good product that it grows as a function of itself. And you mentioned that as part of the culture, I've been working in technology my entire career, the best companies are all ELG companies. They're all product-led growth companies. And so the big part of Nextdoor that has been missing and that we're trying to bring to the company now is a culture of building great product that ultimately will grow the company as a function of that product. What you refer to as NEXT, that was kind of our internal code name, we've rebranded it now, it's the New Nextdoor. So the New Nextdoor, we launched in the middle of July, and it was fundamentally different. It is fundamentally different than the old Nextdoor in a couple of ways, which I'm sure we'll get into. But that's really the very beginning for us of trying to put a better product on the field. And you use the word transformation, in this transformation that we are trying to ignite with Nextdoor, there are kind of 3 phases. One phase is the reset phase. You have to take an existing company, and you have to say, "Look, whatever we've been doing, it hasn't been working well enough, so let's reset the way we think about things. Let's reset the culture, let's reset the priorities, let's reset leadership," right? Those are very hard decisions that you need to make when you know that the trajectory is not the right trajectory. The second phase is rebuild. You have to take the things that you believe will ultimately deliver value and you need to either create them, restore them, repair them, whatever it takes. And then finally, reaccelerate, where you have hit a seam, you've understood an insight, you've been able to deliver value in a way that you weren't able to before, and that will drive growth. I would say that for Nextdoor, we are in the early stages of rebuild. So just to level set, right, the reset takes a little bit of time. You don't do these things sequentially. So while you're resetting, you can also be building and you can even be accelerating, right? But it's really about what is the highest percentage of your resources in any given phase going to. Is it resetting, rebuilding or reaccelerating? For us, right now, the highest priority is rebuild. And we're in the early stages of that. And if we build a better product and we have to learn from our users to do that, we're very confident that the reaccelerate will take care of itself.

So let's talk about the rebuild. It's a great segue into the New Nextdoor. We have the 3 pillars: news, alerts, recommendations or Faves. Before we go deeper into the New Nextdoor, can you level set on what it is and how the core use cases of the app are changing?

Sure. So let's think maybe a step back about the world of social networking in general. Social networks have become the place you go to get information to read high-quality content about a particular slice in your life. You think about LinkedIn, it's not only the place where you connect with people, it is the place you go to find out what those people that you're connected to are doing professionally. You think about X, it's the place you go to find out what's going on in any number of subject areas. You think about Instagram, it's the place you go to find out, in many cases, things you don't know that are being published by people that you may know or not know, right? The same is true of TikTok. It's a place you go to consume content. In all of these cases, content is the thing that gets you to visit the service, right? So at a very high level, it's very, very, very simple with social networks. You need to have great content and you need to have enough of it that people are engaged on a daily basis or multiple times a day. And then you need to figure out the right distribution mechanisms for that content. And what we've really seen is personalization as the primary and most effective distribution mechanism, right? We are no longer in a place where even on LinkedIn, you only see content from people you're connected to. LinkedIn is now recommending content for you based on your profile. Facebook, there was an article about Facebook and Fast Company recently that said 70% of the content on Facebook is not from your friends. And so as we think about the New Nextdoor, the old Nextdoor was almost 100% content created by your neighbors. Your neighbors is defined by some circumference around where you live, and it was all what we think of as user-generated content, right? It turns out, though, that all the information you need locally is much broader than just the information that's in your neighbor's heads. What if it's an event that your neighbors don't know about yet? What if it's some natural disaster that's coming that meteorologists are talking about? What if it's local news? What if it's things about politics that politicians are talking about, right? What if it's a new restaurant opening? What if it's a small business that has a new offering in some way, shape or fashion? And so with the New Nextdoor, the first fundamental change we made was we said, look, we're no longer going to be just about neighbor-generated content. We are going to start to integrate what we think of as third-party content, so content outside of neighbor content onto the platform. The first part of that was local news. So we did partnerships with 5,000 local news publishers. We're bringing in 10,000 news articles per day. And we think that's a really good way to up the ante on the quality and quantity of content. It's just the beginning, though. We need to have every event that's going on this weekend in your neighborhood on the platform. We need to have every new restaurant that has come into your neighborhood, right, and even outside your neighborhood. We need to have every bit of information about the school calendar for the school that your kids go to. If you're religious, we need to have information from your church or your temple or whatever religious organization you're affiliated with onto Nextdoor as well because these are all planks of your local life. So with the New Nextdoor, we said, first of all, we're going to bring in local news. Now that seems kind of obvious that if you're thinking about what's going on locally, you want to read local news, right? But it was a little bit risky for Nextdoor because we had never let publishers into the platform before. So we needed to test it. Now we've got a very positive response and it gives us a lot of conviction and courage that third-party content can be valued by neighbors, that neighbors want more than just the word of mouth that their neighbors are sharing, they just want information. When they see the information, they start discussing those things. So there is a neighbor-generated content piece, even that gets attached to the third-party content, but that was one major change. The second major change is we know that Nextdoor is indispensable when there's a natural disaster or something that feels really important and high priority happening in the neighborhood. Typically, we think of natural disaster because it's easy to describe. It's a hurricane, a tornado, a fire. It could be a crime. But we've extended this now to include anything related to construction or traffic, anything that may be related to a power outage, right? You think about the utilities in your neighborhood when they go out, how do you get information about those services? And so not only did we start to integrate with those services so we could bring the content in, we took the content, we made sure that it was relevant to you because we know where you live and we know that the content is actually relevant for some service area, right? And then we created an entire kind of a surface, a map where you could go and look at all of those alerts in a way that you typically wouldn't before. Before you would just see a feed. And so you guys know how feeds work, right? If the feed is there when you visit, you see it. If the story and the feed is gone by the time you go there, you may not see it. But when it comes to alerts, you need something that's a little bit more permanent or at least permanent until that crisis passes, right? So that was the second biggest change. The third and final change is we know that we're in a world where AI is going to disrupt everything. So a big question that we need to ask ourselves, an existential question is, what is Nextdoor in a world of agents? What is Nextdoor in a world where people completely -- they will expect us to be using AI to improve user experiences? Now in Nextdoor's case, we're very lucky. We have our own proprietary content, so we don't need to go buy content from someone. That content has never been published publicly. So we don't need to be worried that people can get to that content in some other place through some other LLM that's already used the content. And finally, we have our own distribution mechanism. So we don't have to be worried that the only way people can find our content is through a search engine or through some marketing channel or some other way that we don't control. So I always felt that we had really good architectural foundation for playing in an AI world. What I wasn't sure of is when you take 14 years of neighbor conversations and you build an LLM essentially for every single neighborhood, what does that end up looking like? And what we found is it takes the content that is valuable and makes it incredibly valuable. It reads better. It's not just one point of view, it's multiple points of view. If you're looking for a service provider recommendation, you no longer just get a list of service providers, you can have context around those service providers. And then we can apply all the things that all the agentic companies are doing to make it better and better and better. And so that's the Faves part. And Faves is really favorites. So it's the neighborhood favorites that neighbors recommend to other neighbors when neighbors ask questions for service providers or things to do or attractions or really anything they want to know about the neighborhood. Now this architecture, news, alerts and Faves. And I would think of news more broadly than just news publishers, you can think of news as inclusive of alerts, inclusive of the City Hall calendar, inclusive of the things that are on the school calendar. You can think of it more as the local information that's not created by neighbors. News, alerts and recommendations, that's actually a durable architecture that we will be working on for the next 10 years. It's not just something that we worked on for a year, we released and now we're moving on to the next thing. We think that's ultimately what people want when they open a local app.

So a lot to jump off there. I want to dig a little deeper here on the content because I think it's so key. News is now 5% of the feed. We know over half of users are engaging with publisher posts. When thinking about the AI content relevancy and ranking, how do you think about the optimal balance of publisher content with UGC? And is there a role for AI-generated content in the future?

It's a great question. So there are actually two different questions. So the first one, how do we think about the optimal balance? Increasingly, the expectation consumers have is for full personalization. So for every user, there is a different profile that needs to be built on what they expect. Some users actually want all news. All they want is news. They don't want to see any neighbor-generated content. Some users actually just want to see neighbor-generated content. Other users want to see a mix of the two, right? There are going to be some users who want to see new restaurant openings, they're going to be others that don't. So it's incumbent on us with ML to actually start to build candidate sets and profiles so that when you come to Nextdoor, you see what matters to you. Now we have a huge advantage there, which is we know exactly where you live, right? You're not just a user that's cookied, and we followed your cookie across the Internet, right? We know your exact street address. And then if you think about matching that exact street address with all of the kinds of information that exists out there to help us understand the kind of person that lives in a type of house, a type of neighborhood, a type of area, right, we can start to build a pretty sophisticated profile before you even start using the service. When you use the service, when you browse the feed, when you respond to notifications, when you post, then it starts to build a really rich profile for what you want and what you expect, right? The only thing that's really missing is the profile gets richest, the more rich content you have. And so this brings me back to my first point, which is we need much more content on Nextdoor. If we just rely on neighbor-generated content, the only way that we're really going to see a nonlinear growth in content is by seeing a nonlinear growth in users. And we already have 100 million users. So while we can be a lot bigger in terms of users, we really need to be thinking about how do we 10x content with our existing user base. Now that gets to the question you said, right, which is we think about using AI to do that. So one way that we have used AI to get more content into the feed is there's a news article, AI can be used to analyze the news article and AI can pick out a leading question to ask based on the article. So let's take some examples. There can be an article about something that's just happened in a city. There's a new building that's been sold. The AI can look at the article and say, "What do you think is the impact of this sale?" And that starts a conversation. We call them conversation starters, right? We saw a huge bump in the number of comments that were generated when we actually use these AI conversation starters. So that's kind of the very beginning. I'll tell you something that we find a lot more interesting. We have this Faves section that you talked about. And if you go on the Faves section, which is not live across the country, it was launched in 6 DMAs, and we're systematically rolling out new DMAs. But if you go in there, you can ask questions like what's a great place to get ice cream with my kids? It's natural language conversational search versus text-based search. That's the way to think about it, right? Now that's natural for Nextdoor because Nextdoor has always been a conversational platform, unlike search engines where you go there and you type in plumber, right? The way Nextdoor has worked is the user has come on and said, does anyone know of a good plumber, right? And that's a different query than just typing in plumber, right? So what we see is every single week, tens of thousands of those posts are not answered immediately on Nextdoor. It may be 24 hours after they're answered, it may be 48 hours. In some cases, they don't get answered at all. That doesn't mean that we don't have the content. So what we should be doing and what we will be doing is we will take the AI results that we've already generated through the LLMs for each neighborhood and when someone asks a question like, "I'm looking for a great dentist for my kids. I'm looking for something to do this weekend, are there any suggestions to replace to do a great hike? I'm looking for a place to board my dog when I'm out of town." Things that typically are asked over and over and over again, right, if someone happens to see it when you post it and they respond, great. But the clock starts ticking as soon as you post. At some point, the clock hits some number, whether it's 30 minutes, 3 hours or a day where it's going to make sense for us to use AI to generate an answer based on the corpus of data that we already have, that will be a game-changing experience on Nextdoor because now you'll go to Nextdoor and you know that you'll either get a completely fresh answer from your neighbors or if you don't get that, you'll get an AI-generated answer that's based on previous conversations. And in that answer, we can of course put at the very end and if there are any neighbors who also recommend this, chime in here, right? So there are ways that we can use AI to lubricate these conversations and create more content in a way that we never could before.

That makes a lot of sense. So we're still early here in Faves in search. But is there a revenue and a monetization opportunity attached as you think about search and your data corpus?

Yes, there are a couple of things that we should be able to do. So the first and simplest is we are an ad-based revenue model today, right? So the more people use our platform, the more ad inventory we should create, the more revenue we should generate, the higher the top line is. And we've already said that starting Q4 of this year and in 2026, we're going to be breakeven. So in that case, any incremental revenue will actually flow right to the bottom line, right? So that's a very simple, more engagement, more revenue, more revenue, more profit, right? It should actually be that simple. In addition, though, what we should be thinking about is in those conversations that neighbors have or that AI generates about small businesses, is there a way for small businesses to be part of the conversation? Not by putting an ad there, but by literally integrating into that dialogue, right? That's something they want, right? They even actually want to be involved in dialogues where neighbors are talking about their competitors, right? And so the thing that we're thinking about right now is if you could alert a small business, every time someone was talking about them or one of their competitors or really the space overall, and you could then send a notification to that small business to be part of that conversation, that would be commercially very valuable to the business, right? It's almost like a reverse search. Why do small businesses buy their name on Google? Because when someone goes to Google and they type in the name of the small business or the name of a competitor, the business wants to have a presence there, right? The same way when neighbors are talking about your business on Nextdoor, we should give them a way to read the conversation or be part of it. That's a different business model than just an ad that goes right in there, right? So that's one of the ways that we should be able to start to extend so that it's not just advertising based on inventory, it's advertising based on intent.

Yes, very interesting. I think we should take a step back here because the next -- the New Nextdoor, pardon me, is really interesting. But how do new and existing users find out about the New Nextdoor? And then given platform WAUs are about 1/5 of overall verified neighbors, I'd imagine reattracting dormant users is maybe the first key focus here.

It's a great question. So we have 100 million verified neighbors, a little over 100 million verified neighbors, and we have just under 25 million that are visiting every week, right? So those are the statistics that you just said. At this point, we think it's a lot more important to take the 25 million that we have out of the 100 million, right, and turn that into 30, 35, 40, right? Then think about let's take the 100 to 105, 110, et cetera, right? Because ultimately, what you'd like is you'd want as large a percentage of your verified neighbor registered base to be actively using the product as possible, right? And that's really the opportunity for Nextdoor. The weakness of Nextdoor is, as you said, only 1/5 of that installed base is used as a product every week, right? What if it were 1/3, what if it were 1/2, what if it were 80%, right? That would radically change the economics of the business, even on the existing user base. So that then begs the question, okay, how do you get the rest of those people to use Nextdoor? They clearly thought there was some value proposition because they joined at some point. And then for whatever reason, they're not using it frequently or they've churned or whatever the case is, right, that's where we think deeply about what we call reactivation. You and I have talked about this, right? We have not actually started the reactivation yet. The reason for that is because I don't think you get multiple opportunities to tell someone who's decided, this isn't as valuable as I thought it was. You don't get multiple opportunities to go back to them and say, "Actually, you're wrong, come back." When you say, actually, you're wrong, please give us another chance, you want to show them something that you know is going to be really compelling. In fact, the way we should be thinking about reactivation is not just broadly, okay, we have people who haven't used the product in 6 months, let's send them all an e-mail saying, here's the New Nextdoor. No, we shouldn't do that for sure. What we should do is we should say, where are the neighborhoods where there has been a super compelling conversation. Now let's take that conversation and let's send it to that dormant user and say, here's what you've missed recently on Nextdoor, please come back. Now that's a wholly different kind of approach because that's not a generic marketing message, which is there's a New Nextdoor, please come back, right? That's -- there's something very specific that you've missed. And in many cases, it will typify what we're trying to build with the New Nextdoor, which is a fear of missing out if you're not on the platform all the time, but that's something that takes a little bit more time to build, right, because it has to be personalization, you have to actually look at the quality of content that's there and then you pick your moments and you initiate reactivation. That's the work that we're doing now. And I would expect that, that's going to happen in Q3 and Q4. Now again, we could send a can and blast out to all of those dormant users. And for sure, we would get some performance back, right? And you'd see it in the metrics and that would be great. We'd go out on a quarterly call, and we'd say, "Look, look what happened to our users, right?" But we want to do this in a way that whatever that bump is, it sets a new baseline. We don't want it to bump up and then it goes back down because people say, "Oh, there's a New Nextdoor, I'm going to go visit it." Well, what if they visit and the first thing they see is not a very compelling piece of content. Then they're never going to come back, and we cannot reactivate them again, right? So I think the reactivation thesis is a very powerful one for us, but we need to take our time and do it the right way.

Yes, that makes sense. As we build off that reactivation of dormant users and we think about attracting new users, are there more demos that we could expect to be added to the platform? I'm thinking about specifically with Faves and generative AI in search. Could younger demos potentially be added as well?

Yes. So historically, the demos where we've performed the best are suburban neighborhoods with folks that own their home or have made a significant investment in living in the neighborhood. Typically, they have kids or they are of the age where they typically have kids, right? And so that's kind of our sweet spot. Areas where we've struggled a little more urban environments with younger folks, I wouldn't necessarily say that college students are a vibrant potential opportunity for us. But certainly, once you graduate from college and you move some place for the first time, you're trying to figure out what's going on. I would point to 2 types of content. I don't think it's really in the generative AI space per se, but 2 types of content. You can think of generative AI as a foundational layer for us that should make everything we do better versus it's a vertical initiative that may attract a certain type of demographic. Content on the other hand is the thing that ultimately gets people to visit the service. And so there are 2 types of content, for example, that will be much more valuable or equally valuable to suburban folks as well as urban folks, right? Alerts. If there's a power outage, if there's inclement weather, if there's crime, it doesn't matter if you're 8 years old or if you're 80 years old, you want to know about it. That's why we've really emphasize that kind of content because it is really kind of universally relevant, right? It's not for a particular demo. It's for everyone who lives in the neighborhood. Now an example of one that's a little bit more unique or tuned for a particular demo is we are in the midst of testing events content. So one of the next big things you're going to see from Nextdoor is a big push on, on Thursday, I want to know what's going on this weekend in my neighborhood, right? We've not done that in a systematic way. It could have been random and neighbor would have said, "Hey, this weekend, I'm going to do X, Y, Z or tomorrow, I'm going to do this thing or tomorrow, there's this book signing," right? But we haven't done it in a very systematic, predictable way, which is what people expect from us, right? So as we roll out events, I can tell you as someone who has a full-time job or multiple full-time jobs and 3 kids under the age of 13, I don't have a lot of time to think about events outside of the events that I already have in my family, which are typically my kids sports, right? But when I was your age, right, when I was newly married or I was single, right, I wanted to know what was going on in my neighborhood. It was much more tuned for my age, where I was in my life and my demographic even where I lived, right? So as we roll out events, I fully expect that the younger demos will find that much more interesting than maybe they find the discussion of where can I find a trusted babysitter, right? You don't have kids, the babysitter thing doesn't matter at all. This then goes hand in hand with the ML that I was talking about before, which is: should we even show the babysitter post to someone who doesn't have kids? No, we shouldn't, right? And should we show the nightlife-centric posts to the person who has 3 kids or who has a newborn? No, probably not, right? So these things kind of work in concert with each other. But again, the common denominator is a lot more valuable content. And so the thing that we have to ask ourselves all the time is, what are all the possible pieces of local content that you could imagine finding, all of that needs to be on Nextdoor, all of it. And unfortunately, or fortunately, if you just have one type of content, it's not sufficient to turn Nextdoor into a daily utility. But if you start to layer on multiple pieces of content, you should start to see the reason and the frequency for visitation to go up.

That's great. As we build this New Nextdoor vision, can we talk briefly on the monetization. Self-serve revenues have accelerated now 58% of total. And we also have this emerging nascent programmatic opportunity. What do you see as kind of the core revenue growth levers going forward and ad tech improvements that we can look forward...

So I'll start with what I think is table stakes, and then I'll get to maybe something that's a little bit more aspirational. So the table stakes, if you're a social network today, is to have a suite of technology tools that make it super easy for anyone to come and advertise on your platform, that's the self-service thing you're talking about. And then that ultimately create better performance over time through learning who is the right person to show this add to, what is the right time and what do we ultimately think will increase click-through rate, right? And so that's something that we've actually had some success building. You can see it in all of our metrics, the CPM is going up, the percentage of self-service revenue going up. And frankly, the fact that we've been able to sustain our revenue even amidst the platform that has historically not been growing that well, right? And so all of the work that we've done on the ad stack, while it is table stakes, it's working. We can actually point very quantitatively to how it's working and we feel good about that. That also gives me faith that we have the right personnel internally in product development that if we can do that on the ad stack side, we should be able to do that on a consumer offering side as well, right? But I do think that's table stakes. Where it gets a little bit more aspirational is, first of all, if we can really drive engagement up, then obviously, we are driving much more revenue because our ad stack platform is now mature enough to take advantage of all of that opportunity, right? So that's a near-term big driver of revenue, right? But it's not easy to actually move engagement. I mean that's the reality. You look at all of these companies of a particular scale, right, it's not easy to double your engagement. It's not even easy to make engagement go up 10%, right? So that's kind of the slow and steady path. The thing that's really aspirational, you talk about programmatic, what are all the additional ways that if we have a healthy, vibrant ecosystem, we could generate revenue? Programmatic is certainly one of them, right? Thinking about percentage of transaction and this idea that we talked about, which is almost a reverse search engine that we internally called opportunity alerts, where we tell a small business or we tell a business that there is an opportunity for business based on a conversation on Nextdoor, those are the things that are pretty exciting because they're not reliant on the supply of advertising. And the reason that's so important is we know how to generate more revenue today, show more ads. Now there are 2 ways you can show more ads, right? One way is you can create more inventory. The other is you can increase ad load. One of the things that we've been told over and over again by our users is that our product feels too commercial. That's basically, I'm seeing too many ads, right? So we've made decisions in many cases to reduce ad load, right? And lots of analysts out there and lots of investors look at us and say, "Hey, you've been testing this reduced ad load thing, when is the ad load going to come back?" Well, I don't know if it should come back. I actually think a better way to solve that problem and generate more revenue is deliver more value to users so that they don't feel like we're just adding more ads. Now you can do it 1 of 2 ways. You can increase the amount of noncommercial content or you can make the ads feel like content. We know that Instagram, I mean, if these days, if you look at Instagram, of the first 10 posts you see on Instagram, I'm sure over 50% are going to be ads. It's kind of mind blowing, right? But people still use Instagram because the ads are very good, right? So as we learn more about our users, I think we do have opportunities to make the ads feel like content, but we're very far from that today. And so I would set expectations that if you're a founder of a company and founders think about the long term, founders think about foundational value, right? I didn't come back to Nextdoor because I wanted to snap my fingers and do all the short-term things that would do short-term revenue or do short-term engagement. I came to Nextdoor because I believe there was an opportunity to build the definitive app in local. And I don't think the way to get there is to increase ad load. I think the way to get there is to increase value. Now that can be frustrating because ad load is an easy way to generate more revenue. But in the long term, it's not something that users appreciate.

So drive usage and targeted relevance to get revenue?

Yes. I think there are 3 things: more usage, better targeting and then non-news feed ad opportunities.

Yes. Last couple of minutes, do we have any questions in the audience?

Regarding opportunity alerts, have you talked about a time line for when that would...

I would call that beta at best, which is to say very early. Now here's what's not early: some very large double-digit percentage of the conversations on Nextdoor are about commercial service providers. So in terms of the demand of that kind of content, we're not early. We're early in thinking about what is the best way to take that existing demand and turn it into a commercial opportunity, right? So we don't have to convince our users to ask for plumbers and dentists and doctors and babysitters, right? We just need to find a more intelligent way to monetize that versus just sandwich adds between those posts in the feed. And that's where we're really early, right? And it's exciting because a lot of times, I mean in these marketplaces you're building, you have to build supply and demand, right? Certainly, we've got the supply at that point of that kind of commercial content. We have to figure out the right way to connect it to the commercial entities. I wouldn't put that in any financial model, though, anytime soon. I would consider that to be upside.

All right. Last minute here. Let's hit on profitability. Revenue per employee is up 58%. We have a recent restructuring plan as well. Where should we think about areas of relative product investments relative to operating leverage?

So we have over $400 million in cash. And so we don't have the existential risk of we're going to run out of money that's why we need to get to profitability, right? But that said, as someone who started 3 companies and primarily done start-ups, where you are at existential risk every single day, I felt like it was very important culturally for us to say as a company, every dollar we spend is going to be sacred. And the way to do that is to say, we're going to be breakeven, we're going to be breakeven starting in Q4, we're going to be breakeven in 2026. And even though we have all the cash, it doesn't mean that we should use it. That said, if we saw something, let's take opportunity alerts as an example. And it required a $10 million investment in 10 new amazing AI or ML engineers, right? Or we felt like there was a marketing campaign we could do that would be really effective in reactivating dormant users, right? We would certainly do those things, right? So we're not ideological about we're not spending any of the money, but we're setting that as the foundation because that makes necessity the mother of invention versus a war chest being the mother of invention. So I would think about the breakeven, and this is both Q4 and 2026. I would think of it less as, oh my gosh, we've made so much progress from a business standpoint that we're ready to start generating cash, right? And I would think of it much more as we are embracing the cultural norm that we're going to treat every dollar like it's our last. And when we start to inflect all of that incremental revenue, we can make a decision on are we going to reinvest it, are we going to decide that it's going to flow to the bottom line as profit, right? But we're going to be very, very judicious with the capital and with the way that we think about building this company in general.

Great to hear. All right. With that, I think we're out of time. Thank you so much for joining us, Nirav.

Thank you for having me.