Thank you everyone for being here. My name is Matt Koss, Morgan Stanley US Internet team. Thrilled to be joined by David Schwarzbach, CFO of Yelp. Thank you so much for being here. I think we have dueling disclosures, so maybe you wanna kick it off.
All right, I'll go with the first disclosure. Thanks, Matt, for having us at the conference. We'll be making some forward-looking statements during the conversation today that are subject to risks and uncertainties. Please refer to our SEC filings for more information on the risk factors that may affect our results.
All right. On my end, for important disclosures, please see the Morgan Stanley Research Disclosures website at morganstanley.com/researchdisclosures. If you have any questions, please reach out to your MS sales rep. All right. Now that we're through that.
All right.
Thank you so much for being here. Let's start high level. Talk through in the six years I think you've been at Yelp now, some of the biggest strategic decisions that you've taken you know through today, and what you see as you know the strategic playbook for 2026 and beyond.
I did join in twenty twenty-.
Mm-hmm.
Right before COVID. COVID arrived, and that obviously drove a lot of change. The company had really already been on a trajectory to change from what had been a sales headcount-driven growth model to a product-driven growth model. That was one. A lot more self-serve, going upmarket and monetizing services leads at a higher rate. We really accelerated that transformation that we were going through and continued to execute on that through 2021, 2022, 2023. I would say in this current era, we're going through a new transformation like, I think, most companies, which is AI. We are very committed to transforming the business, the way that we engage with consumers, the way that we provide value to advertisers using these pretty extraordinary capabilities that have become available to all of us. Maybe a few things.
One is just reconceive the entire experience on Yelp with AI, make it much more conversational. The second is to power tooling for small businesses to enable them to be more effective and successful. The third is really to drive our partnerships with folks like Apple, Amazon, OpenAI, to bring our trusted content to consumers wherever they are.
Great. Let's stick with AI for a minute then. You just mentioned the OpenAI partnership. I think that's to provide AI-driven local discovery, but correct me if I'm wrong about that. Just tell us a little bit about what that partnership does, what it looks like from a financial standpoint, and what your expectations are for referral traffic to Yelp from ChatGPT.
For Yelp revenue from licensing falls in other revenue.
Mm-hmm.
Just as background, other revenue consists of subscription. That's a SaaS side transaction. I think we'll talk in a moment about the relationship with DoorDash. Then there's licensing. So around licensing as I already mentioned we have relationships with a number of different companies and most recently OpenAI. One of the things that we did not do early on was license our data for training. We didn't think that that was a good use of our data from the competitive perspective. We thought it would erode our competitive position, but also we wanted to really monetize at a much higher rate. Other revenue in aggregate grew 17% in 2025. In the fourth quarter, it grew 33%. That's a reflection of these partnerships that we've been able to enter into.
With regard to OpenAI or Amazon, or you'll find us in Apple Maps, it really is this fundamental authority that Yelp has around all things local. We've seen that if you go to a site and you just see a rating, say 4.2, but there's no brand or company name associated with it, you don't really know what that means. As soon as you add that Yelp logo or the Yelp name we find that consumers engage with it and trust it.
Mm.
The power of these partnerships isn't just that we generate revenue, it's that we are ubiquitous. People see our brand everywhere it does drive traffic back to Yelp. I think everybody's still figuring out though what does the search experience look like with AI.
Mm.
I think that's true. Google is still sorting out, what do AI Overviews do, when do they appear, when do you go to AI mode? I think that's also true across pretty much everybody else, and we believe we're very well positioned to power local search for those who aren't working with Google.
Got it. I wanna go to Dash 'cause you mentioned that in a minute here. Before we go there's something interesting that you've kinda touched on there about the Yelp logo communicating trust.
Yeah.
I guess that you know people who haven't been following the story closely probably have an outdated sense of what Yelp is all about. You know, people go looking for restaurant you know reviews or pictures of the menu. I guess what is your role as you think about it as sort of communicating trust in the ad ecosystem? Like, what is your position and value proposition in 2026?
I think we've all had the experience using an LLM and what it returned seemed inaccurate, if not made up.
Mm-hmm.
That's a common experience. We've done our own market research, people are really looking for trust signals. Why should I believe that this LLM is returning a result that I can rely on to make a small decision, maybe going out to dinner or a big decision remodel my home? Our role around the quality of our ratings and reviews comes from human-generated content. We used to talk about user-generated content, but we've really now focused on this differentiation that we have, which is it's humans generated and human reliable. I hope all of you use Yelp, but my own experience, it's quite accurate. Like when it's 4.2, I have a 4.2 experience, and when it's lower, I have not as good an experience. If you think about it, where's Yelp at, in a sense, what's its core asset?
Mm-hmm.
It is this authority. It's how do I know that this is accurate?
Mm-hmm.
That's why these obviously large players are choosing to partner with us because we are very very focused on review quality. We have about 330 million reviews as of the end of last year. We don't show a quarter of them.
Mm-hmm.
That's a significant amount. To generate that amount of content that you don't show obviously is expensive, but we do that in order to ensure that quality. You can think of our defensible market position is around this corpus of reviews that we've managed to aggregate over time. It's actually also the moderation that we can apply to ensure that quality and the brand value that we have established over a long period of time and that enables us to partner across the web. Of course for many people, Yelp's a very beloved brand. They love Yelp. They love engaging with it. They love leaving reviews, and obviously we like that. The last point I would make is we've really shifted from a revenue perspective to services over the last number of years.
About 70% of our revenue comes from services, although we're best known for restaurant reviews.
Got it. Maybe let's stick with that last point, kind of about the relationship between RR&O and services. As you pointed out, services become the main growth driver of the business. Maybe talk about the relative strengths and weaknesses of RR&O versus services, and sort of the trajectory of these two businesses from here.
There's no doubt that just broadly, the local physical economy has been quite pressured.
Mm-hmm.
over the past couple of years. This is a combination of factors. Input costs have risen, whether that's ingredients or parts. Labor costs have risen. At the same time, while businesses were able to pass those price increases on to consumers in the 21, 22, 23 timeframe, by 24, consumers had had it. They were not willing to accept additional price increases, and that created margin compression. The other thing that definitely emerged over the course of 25 was just this very significant pressure on these businesses to continue to deliver value at the same time that consumers were trading down a bit. Instead of taking you know a marble countertop, they were doing quartzite. I'm gonna repair this instead of replace it, or I'm gonna try to fix it myself.
Mm-hmm.
We've definitely seen that ripple across broadly the local physical economy. In particular, it's really hit restaurants as well. In 2025, what we saw in the spring was there was just a lot of uncertainty that was introduced, particularly in the spring, around tariffs, and that impacted seemingly business owners' confidence in the economy. We really saw a dramatic change. We had seen double-digit growth in services for 13, 14, 15 quarters. We saw it again in the first quarter of 2025, and then that just slowed significantly, and that persisted through 2025. It has become more broad-based than it had previously been. I would say that you know or at least on the earnings call, what we shared was those dynamics had persisted into 2026.
Mm-hmm. Has that played differently across different sizes of advertisers? Like have you seen a divergence in behavior between enterprise and SME?
That's also a great question. One thing that historically we saw was large enterprises made very quick decisions. They have FP&A teams, they report earnings, and they're making rapid business decisions around marketing spend. Small business have been much slower to respond. That was not the case actually. In the spring of 2025, small businesses responded very quickly to the headlines, much faster than had historically been the case. There's clearly something different about that dynamic at that time compared to the prior years. Since then, i would say it's now gone back to where enterprises are making more rapid decisions and small businesses are making decisions less frequently, but they haven't returned to the level of spend that they had previously been at, or at least the growth in spend.
The dynamics did change, but at this point, o would say in essence, they're both operating about at the same speed because there's so much news all the time.
Yeah. Closely related to that is multi-location. I think it's something you've called out as an opportunity. I guess are there initiatives you're pursuing to push further into that, and how is that going?
We are very unrealized in services in both mid-market and enterprise, and we put a lot of effort in 2025 into tooling to enable us to serve those businesses, and we made a lot of progress. We're continuing to focus on that here in 2026. One of the things that we are doing is increasing spend on paid search because obviously if we're gonna buy a lead, we wanna land it to someone who's willing to pay for that lead. Connecting the purchase lead with being able to monetize is something that we've been very focused on. Those enterprise businesses, they want more leads. If we can deliver them they wanna buy them.
Got it. Maybe let's go on to Dash. You know, I think you've talked in the past about seeing some competitive pressure from food delivery, and at the same time, you have this partnership with DoorDash, I think also one with Grubhub. How do those partnerships work and what's the progress on growing them to date?
Those, we had historically worked with Grubhub. That relationship shifted primarily to DoorDash last year.
Mm.
It's gone really well.
Mm.
We're pleased to be able to partner with them and to surface Yelp content also on DoorDash and obviously be able to refer folks from Yelp to DoorDash, and that's a pay-per-lead type of model.
Mm.
with them. Obviously DoorDash and just food delivery has really emerged almost a secular shift in the way that people eat. They don't eat out, they eat at home with food delivered-
Mm-hmm.
has been a big shift. Just as a quick calibration only about 10% of our revenue now comes from restaurants so it's a small portion although 85% of our traffic comes broadly from restaurant, retail, and other. Just in terms of economically that shift to delivery has had really only a marginal impact on our business. We're certainly pleased with and happy to have the partnership with DoorDash because people do come to Yelp, they look at a restaurant, if they can order through DoorDash from Yelp, then that's certainly a convenience.
Got it. I think that you saw last quarter paying ad locations go down by about 5%, and mostly driven by RR&O. You know, how has that churn trended across segments? Is it really just an RR&O phenomenon? Is it coming from customer churn or ad spend per customer?
It's a good question. It is definitely predominantly on the restaurant, retail, and other side for all the reasons that I already discussed, though, we did see a modest amount of pressure on the services side as well. I think at the end of the day, what's our job? To deliver valuable leads to businesses, whether it's a restaurant getting someone to come in, or it is a plumber who wants to do that next job.
Mm.
What we remain focused on is delivering incremental value to the advertisers through becoming more efficient at matching, by ensuring that the conversation between that advertiser and that consumer is as good as it can be. One of the things that we did, and we'll talk more about it I know in a moment, but we did make this acquisition of Hatch. Hatch is an AI lead management solution, we think, best in the market. If you wanna interact with consumer, you wanna do it in a way that leads to a lead.
Mm.
That's our job, is to really make those things more valuable to the advertiser. The other thing I would say is we did dramatically increase our monetization of services traffic over the 2020, 2021, 2022, 2023 timeframe. We realized it wasn't good enough to just add leads, you had to add quality leads. We did focus for a period of time around quality.
Mm.
Yet there is still headroom to go on the services side to increase monetization, and we actually have focused less on increasing monetization in restaurant, retail and others. Regardless, we do see that there's an opportunity to increase that given the traffic we have.
Got it. Let's stick with Hatch. I mean you know you introduced it as an acquisition you made recently. Talk about how it enhances your, I think, previous Yelp Receptionist product.
Yes. We had built a voice AI product for service pros that would pick up the phone when someone called. They're busy with the job. And we had actually partnered with Hatch. Hatch works across Yelp, Thumbtack, Angi, Google, it's across all platforms, and it is very effective at increasing the probability that someone is gonna pick that particular business for whatever job they have, whether it's plumbing, electrical, it could be landscaping. The acquisition, we're very pleased, it just closed a few weeks ago, and we have focused on that as opposed to the Yelp Receptionist product that we had been building. There's no doubt, and we're gonna get to AI, that things are changing very rapidly.
This idea that an AI agent can successfully engage with the person to understand better what their need is and then to ensure that they are getting the service that they expect, that is a clear opportunity, and we do see improved conversion when you add these AI agents to the conversation.
Got it. Let's stick with AI. I mean i think in 2025, you talked about adding over 50 new products and features on the AI's front, or many of them are powered by AI. Can you talk about which of them have seen the highest adoption, and what the results that you're driving from them are?
In the spring of 2025, we introduced Yelp Assistant that was focused on services. We have something called Request a Quote. You come to Yelp, you can either just search for a plumber and pick a plumber, or we ask you, "Tell us a little bit more about your job so we can better match you with the right service pro." Request a Quote. That Request a Quote was definitely a question tree. If the person asks this and gives that answer, yes, no, what you have, what you may have, it would go to that next fixed question. The beauty of a conversation in the LLM is it's dynamic. What is the most optimal next question I should ask that person? You get a lot better information, and it improves the matching.
That has seen significant traction over the course of the past year. In 2025, Request a Quote requests increased by 400% compared to 2024. The other thing that's very important about Request Quote is 100% monetized, almost always 100% monetized. We show four results instead of person clicking on one. That inherently increases the monetization without a perception of increased ad load. It's actually a better consumer experience, and it monetizes at a higher rate. The Hatch product fits right there in Message Center where it's managing that conversation between the service pro and the individual and doing that across a variety of platforms. Broadly speaking, voice AI is something that is rapidly emerging as a capability.
We believe that we have put together a equally best-in-class product to what's offered on the market, either by startups or other enterprises, more established enterprises. Voice quality is a complex topic. It's not simply, can the AI understand what's happening, though that's obviously essential, and we think that we're very good at that, but there's a lot of dimensions. People make a decision in that first second or two whether to engage with the AI or not.
Mm-hmm.
The voice quality, the emotional character of it, the intonation the cadence all of those matter tremendously in actually having a person choose to talk with the AI.
Mm-hmm.
That's something that we think that we're doing really well. It's something that we think we can bring to Hatch's product, and it's certainly powering our Yelp Host product, which is answering phones at restaurants.
Yep. You talked or touched on Yelp Assistant there for a second. I guess you've highlighted the past couple of quarters a lot of momentum with that product. I think you also announced plans to roll out cross-category this quarter.
Yeah.
What, what actions are you planning to integrate that product beyond service requests in 2026, and how do you see that impacting revenue?
Thematically for us, it's really this shift that's going on in search from links to answers and actions.
Mm-hmm.
People really want an answer, and they want an action. They don't just want that for services that they want that across the Yelp experience. If you think of Yelp as a specialized LLM or chatbot versus general chatbot, what differentiates Yelp? Why would someone come to Yelp to engage as opposed to going to another platform? It is going to be, i can do a search for anything local-
Mm-hmm.
across all the categories we serve. Clearly, it was very important for us to bring Yelp Assistant not just to services, to all categories. We're doing that. We're very excited about the product. We've made tremendous product progress around it. We're building the integrations to enable that action layer after you ask the question to still have that, the Yelp voice in the experience is very important. Brand expressed as voice is something that has emerged as quite an important dimension to all of these products. We think we do that well. We're very excited about the opportunity there.
Most importantly, it's now a conversation with the consumer, and you can really learn about what their preferences are you can understand better nuance of whatever it is that they're trying to do, and you're able to serve that more effectively by surfacing the right businesses or the right information to them. It really is this idea that we are reconceiving Yelp around a conversation powered by LLMs with that authority, that ability to show the evidence. Why do we have confidence that what we are telling you is accurate? 'Cause we have the photo, we have the review, and we have the rating. That's what differentiates us. That's why you would choose to go to a specialized chatbot for local, Yelp, as opposed to a general experience.
That concept of brand as voice is a really interesting one. I think it's especially interesting for Yelp because and this is supposition on my part, but I would guess that many consumers historically kind of viewed Yelp as a neutral intermediary, where it's a place where you go to find user human-generated reviews and they're aggregated on this site. What is the brand promise that you're putting into a voice with Yelp?
Yeah.
how are consumers reacting to that?
This is a great question. For folks who are passionate Yelp users, what they love is that Yelp can, of course, has that authority and is able to offer the evidence, but it's more playful, it's more fun. It's not going to be exactly the experience that you might expect that is a little more rigid or corporate.
Mm-hmm.
That's not why people come to Yelp. We really want to ensure that that tone comes through in the way that the Yelp Assistant engages people.
Mm-hmm.
Now, of course, it's gotta be respectful, and we have a supervisory LLM sitting on top of it, and we want it to be appropriate, and we've taken all those precautions from a trust and safety perspective. Once you get beyond that, you can be authoritative and playful. They're not mutually inconsistent as long as you're offering the evidence for why that information is being presented.
Got it. Really interesting. I guess kind of sticking with the AI theme, let's move on to risk. It's obviously been a very hot topic, probably you know something that comes up in every session at the conference for the past couple of years. You know, on the other side of agentic commerce, you have this concern that investors have that consumers will increasingly find local businesses, for example, through an LLM platform instead of going perhaps to Yelp. Is there a risk there? Is there an opportunity there? What's your response to those concerns?
far for us, this has been a tremendous opportunity.
Mm-hmm.
Not per se a, an immediate risk of any sort. Things are changing so rapidly, so I'm confident that Yelp is very durable and things are still emerging. With that caveat, why do I have confidence in the durability of Yelp? It just comes back to this theme that I've already been sharing, human-generated content.
Mm-hmm.
Trustworthiness. I still wanna make the decision, show me the restaurants, don't just pick the restaurant for me.
Mm-hmm.
There's that nuance. A service pro's coming into your home. How can I trust them? What do I know about them? Am I gonna trust the LLM to make that decision for me? Or do I want the evidence that, no, this person is very trustworthy, they were on time, they did the job well, and I was able to resolve anything that came up during the job. That isn't easy to replicate by just turning to an LLM and say, "I need a plumber.
Mm.
It's insufficient, in my opinion.
Mm.
We are very much architecting the experience to continue to convey that the detail and level of information to the consumer. Again, this theme of differentiation, why would I pick Yelp? It's because I know that that information is trustworthy, can inform a good decision, and I will get what I expect.
Let's talk about how people are finding Yelp. I think you know the latest figures, you have 74 million direct visitors to the site and the app. Do you rely on search or you know paid marketing to get people to Yelp, and do you expect that to change at all going forward? It's obviously a really fast-evolving ecosystem.
Absolutely. Yelp, because of its strong brand, has a very, very low cost of traffic acquisition.
Mm.
We have our app, 28 million app uniques in 2025, then we have our desktop experience and our mobile web experience. The Comscore number is usually between 70 million-80 million, the 74 million that you referenced, but it's actually 38 million and 60 million for desktop web and mobile web. We have this huge audience already.
Mm.
We do we spend very, very little on traffic acquisition, TAC. We see an opportunity to do that, as I mentioned before, for enterprise customer, really bring that lead directly to them and be able to monetize at a high rate around that. Certainly we benefit from SEO. We're, we show up in Google results. We're highly trusted. We think that we are gonna remain very relevant in this era of chatbots, where they're looking for trusted content to surface. That's why people are partnering with us and licensing our information, as I already talked about. Overall, very confident that this core asset, highly trustworthy, human-generated content is something that people are gonna continue to seek out.
Got it. Thinking about strategy for a second, if I think back about what we've talked about up to this point, there's a couple different categories of things that you're doing. There's Yelp Assistant kind of AI on the consumer-facing side. There's Yelp Host, which is kind of playing intermediary, but it's an AI product for advertisers and businesses. Then you have you know you're plugging into some partnerships, like with OpenAI. How are you thinking about allocating investments and energy and the strategic priority of those kind of very different functions of the Yelp platform?
We are very rigorous in our planning process. We're very, very ROI-driven, and we're also taking the long-term view and building the product not just for 2026, but for the years to come. I think we've all heard this, that there was a huge shift around December in the capability of coding tools, particularly Claude Code.
Mm.
It is absolutely starting to show up.
Mm-hmm
in the way that we think about how to build, how fast you can build, how fast you could iterate, try a feature. One of the things that has really emerged in our thinking over the past several weeks is an understanding, historically, when you choose to build something, it's expensive.
Mm.
You have to have a lot of confidence that what you are building is the right thing.
Mm.
That leads to a lot of decision-making overhead.
Mm.
In a world where you can generate code much more quickly, spin up features rapidly, put them in front of your audience, and get their feedback, A/B tested or just looking at the metrics, it starts to change how you approach product management and product engineering.
Mm.
I think this is going to be true across the board, where you're gonna have this compression of the risk management that has historically gone into writing code.
Mm.
It's very exciting. It will demand a different way of working. It will demand different processes. With this, sort of we passed this event horizon of some sort in December, where it is now not just possible, it's not just a capability, but it's an expectation.
Mm.
The speed with which you develop, the expectation is now that that's going to go even faster.
Mm.
We made the shift to product-led a number of years ago, and we had already changed a lot of our process to speed time to market. I believe that's gonna further accelerate that. That's an exciting place to be because if you have that depth in visitors, if you have that expertise, if you build this variety of capabilities and you can iterate more quickly, then you can bring things to market that much faster that have the potential to be very successful. If it's not, it's okay. You can move on to the next feature.
Got it. Maybe in the last few minutes, we can hit a few on the financial side. I think you guided to EBITDA margin compression of three or four points this year driven by investments in AI and in Hatch. How do you expect you know those AI transformation costs to continue over the course of this year and going forward, and how should we think about your margin trajectory over the medium term?
We did generate $369 million of Adjusted EBITDA in 2025, 25% Adjusted EBITDA margin. We guided lower this year at $310-$330. It's an investment year, and we are investing for the longer term. We are in this seminal moment, this secular shift to a new way, as we were just talking about of building and bringing product and features to market, and we wanna take advantage of that. Our expectation though is over time we are going to get the benefit of that. There's fundamental premise when you're product-led which is that the improvements that you create stack over time and drive margin. That's one element. Of course, our focus at Yelp is very much around doing more.
We wanna drive top line with all of these improvements and create this set of capabilities. My expectation is that over the longer term all of this leads to margin expansion, and I think that's gonna be generally true as people adopt these capabilities. For sure, 2026 is all about investing for the future.
Got it. Thinking about capital allocation, I think you bought back a little bit less than $300 million in stock last year, raised or authorized another $500 million in February. Given the investments that you're making in 2026, how are you thinking about balancing capital allocation across reinvestment, share buybacks, and you know M&A?
Yeah. Maybe just a couple of stats first. $324 million in free cash flow in 2025, 22% free cash flow margin. As I already mentioned, $369 million in Adjusted EBITDA. We drove stock-based compensation as a percentage of revenue below 8%. three years ago, we told people we would do that. We did that. We've set a target of less than 6% by 2027. We've been very focused on capital management, capital allocation efficiency over a long period of time. By the way, we generated $2.24 of EPS up from $0.50 just a couple of years ago. We think that that gave us or gives us the financial strength to make the investments that we're making this year.
With the investment in Hatch, we have definitely shifted our capital allocation policy a bit in that we're using our capital to make those acquisitions even as we continue to expect to buy shares in the market subject to economic and financial conditions. We continue to see that as a important part of the approach. Obviously, the Board approved $500 million in incremental share repurchases, I would just say the balance is shifting a little bit towards holding less cash on the balance sheet, doing acquisitions, doing share repurchases.
Got it. Maybe we can close on sort of a big picture question about AI. It's been a lot of the content that.
Yeah.
hit on today. I guess, what would you point out to people as the most underappreciated opportunity from AI and maybe you know the most underappreciated challenge?
We're still figuring out this interaction model. That's a challenge.
Mm.
it's gonna take a lot of iteration. I talked about Yelp Assistant in services a year ago. The thing that was a surprise is when we got better information, our matching got worse.
Mm.
We're like, "How can this be?" It was so counterintuitive.
Mm.
It turned out that the matching algorithm had been tuned to less precise information.
Mm.
We had to retune the algorithm to absorb this better information. I would just say you can't underestimate that it's. Product development is still a thing. Iterating is still a thing. Understanding how that you present that information and all the underlying systems take advantage of that is still emerging.
Mm.
That's just an unknown, but an amazing opportunity. I think the thing that is underappreciated probably about AI at the moment is just how comprehensive it is across the enterprise and across the consumer-facing and, in our case advertiser-facing surfaces. There is going to be this incremental intelligence and this conversational interaction that just changes all of the way that we all interact with information. We're all obviously using these tool but I don't think that we've yet realized the full potential of them in the way that we work learn and choose to go out to eat or have someone come to our home.
Great. David, thanks so much for being here.
Thank you.
All right. Excellent. Thank you so much. It was great.
Yeah. This is definitely in the crazy rate of change right now.