Colin Sebastian, senior research analyst covering internet at Baird, and we're very happy to have with us today David Schwarzbach from Yelp, and Josh Willis as well, who's in the audience. For those of you here in person, there is an opportunity to submit questions via the email that's on the table there. So feel free to send those up. But before we move on, I wanna pass it to David for the safe harbor.
Thanks, Colin, 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.
Perfect. So I guess maybe to start and maybe level set, I mean, Yelp's a company that's been around for a while, one of the survivors among the internet landscape, which is a great thing. But at the same time, the business has evolved significantly over the years, and I think it'd be helpful to talk about, from your perspective, what's changed at Yelp and what the current priorities are.
Yeah. So I think there's really three big themes over the past number of years that we have focused on at Yelp. The first was really a transition from this sales headcount driven growth model to a product-led model. We'll talk more about that. The business has also really transformed from being very, very restaurant, retail, and other heavy, where in the first quarter, nearly 65% of our revenue actually came from services. So that's a significant shift in the way that we operate. I'd say more recently, we've been really focused on continuing to increase monetization of our existing organic traffic, but we see a large opportunity off Yelp for traffic, and we've done a variety of things there, but more recently, we've started to invest in paid search, something that we hadn't done in the past.
So those, I'd say, are the big three themes of how Yelp has changed over the past few years.
Then when you distill that in terms of the company's TAM, your competitive differentiation and where you see market share today, you know, how does it fall out in those respects?
Yeah. So I would say our competitive moat really comes from the fact that we have trusted content. We think that it's very high quality. We think it's differentiated from other sources. In general, we're very trusted for local data, but it is the ratings and reviews that we're best known for. That's one key part. We certainly have a lot of first-party data, which I think has emerged as, being extremely valuable, but also essential to competing in this era. And I think broadly, when you think about Yelp, the ability to apply new technologies to your existing products is also essential. We've been using machine learning for a very long time. We were actually using large language models well before OpenAI introduced ChatGPT. We have a trusted brand, which lowers the cost of consumer acquisition. I think that's also very important.
And finally, we are obviously an, a digital advertising platform, so you have to have a, a valuable audience to advertiser, advertisers. You have to have consumers that advertisers wanna reach, and more than 50% of our visitors come from households with income over $100,000, highly educated, very discerning. They like to do research, and so that audience is one that advertisers wanna reach.
Given where we are now, what do you think investors misunderstand or underappreciate about the company?
So-
-if anything?
There's probably a couple things. So first, I do think this transition to being focused on services, not everybody may have focused on that. As I mentioned, nearly 65% of our revenue in the first quarter came from services, and we really do believe that we are well positioned to lead in connecting consumers with great local businesses, but particularly service pros, and we've done a lot there to continue to enhance the experience. We did recently release our Yelp Assistant. It's a chatbot, and we think that it's quite differentiated. I'd encourage you to try it. You can see it in the app in the Projects tab. You can click on it, see for yourself, test it out. But that ability to integrate technology with experience, deliver value both for consumers and advertisers.
When the chatbot interacts with people, we get more information, and that really enables us to increase the value of the lead. I'd just say this is a fundamental element of long-term enduring value: is can you deliver value to the people who buy your product? I think we have demonstrated quite consistently our ability to do that, particularly in services, and that's represented by revenue per paying advertising location, particularly in services categories. So I think that's a part of it, that this product-led strategy has enabled us to do quite a bit to deliver value. I'd say that's one, significant, piece. We're also a diversified set of categories, and so that provides ballast across the business that we've been able to leverage. I think people underestimate how easy it is to do what we do.
It's very hard, as we know, to do content moderation. We do that really well. It's very hard to generate high-quality, user-generated content. I think we do that really well. So there's a lot of operational know-how. And then finally, on the capital efficiency side, we'll spend more time on it, but we have been very consistently returning capital to shareholders, and here in 2024, we expect to issue 65% fewer shares to employees. We have over $500 million authorized in share repurchases. We repurchased $62.5 million in the first quarter and obviously, prior grants will run off as time goes by. So you have this combination, I think, of the potential to lead in a very large TAM, services broadly and home services in particular.
I think we have a great technology platform, very high product velocity, which I think is essential to compete, and I think that we are really striving to be very efficient with our capital.
What are the top couple of KPIs that you look at when you evaluate the company? Either what you report quarterly or even internal metrics that you don't report externally.
We did a couple of years ago, really try to line up what we report externally with what we talk about internally. It's not every metric, but they are very aligned 'cause, you know, if you sat in a meeting at Yelp, my goal is that you wouldn't be surprised by the topics or the way that we were talking about it and the metrics that we were using. Again, there's a lot more depth to the things that we do, and we look at them more broadly. But at the end of the day, as an ad-supported business, clicks times CPCs equals ad revenue. If clicks are growing, that's a positive, and if CPCs are growing, then we need to be delivering more value per lead, or we can enhance that by actually seeing CPCs come down. That's what we saw several years ago.
More recently, what we've seen is ability to grow clicks faster than we grow CPCs. That's a good place to be 'cause that indicates that you're delivering more value. So what do I look at daily? I look at clicks and CPCs. Obviously, I look at ad budget by category and region, so generating demand is something that's important. That's certainly gonna show up ultimately in revenue per paying advertising location. We care about advertising locations, but I would underscore, we focus most on quality in our paying advertising locations. Paying advertising locations have been flat over the past several years, but revenue per paying advertising location, particularly in Services, has gone up quite consistently. If we can have an existing advertiser spend another dollar compared to getting that first dollar from a new advertiser, it's much more efficient for us to get that second dollar.
The final thing I would say is we do look at, obviously, clicks as a reflection of engagement on the platform and our ability to monetize those clicks, and you put all of that together, and those are the types of things I look at. With one addition, we have increased our expected spend here in 2024 on paid search, something that we hadn't spent on in the past. And so more recently, the metric that I'm most focused on is projects in the categories in which we are buying additional leads, because that's gonna be an indicator of our ability to pull in traffic, high-quality traffic, into services categories, particularly home services categories. So that's an example of, a few of the things that that I look at daily.
No, that's great. I do wanna step back. We are at partly a consumer conference, and one of the themes, of course, is the health of the consumer. So as we're asking in all companies, you know, how are you gauging the health of the consumer right now, and, and how are your end markets performing? Obviously, they're very different parts of the business.
They are very different parts of the business. One of, we talked a little bit about this on the Q4 call, but in late December, we started to see something happening with the U.S. consumer in our restaurant, retail, and other categories. It was pretty apparent by the second half of December in our dailies that something had shifted. It was not yet clear whether it was seasonal or something that was more enduring. I think there was a little bit of both going on. By early January, it was very clear that there'd been a significant shift, and I think as folks have reported over the course of, the past months, we've seen that start to show up where consumers do seem to be stretched.
That's been something that has made it harder, for instance, for restaurants, in the sense of being able to pass on price increases that they're seeing, whether it's labor cost or input costs. So, consumers have become more reluctant to do that, or they have decided to, downshift in where they're going out. So that's, that's very clearly, taking place, and I think we've also started to see that, consumer debt is increasing and maybe even delinquencies. Interestingly, on the home services side, in particular, we've seen much steadier consumer demand, and I do think there's a couple reasons for that. One, in general, homeowners have projects come up all the time. If you own a home, you know that you're gonna have something with the plumbing. You might have something with electric- the electrical. I had a sinking back deck.
I had to have that fixed. You might decide to do landscaping. As people stay in their homes longer, they do seem to be doing projects, and home equity has increased dramatically over the past five years. So I think people are investing in their homes because they're not moving. And so there, I would say that the demand on the consumer side, we think of consumers as the supply, with advertisers as the demand in our particular marketplace. But that supply of consumers has actually been quite steady and, you know, steady enough that we've thought that it was a great opportunity to go ahead and try to enhance the amount of traffic that we're getting there through paid search.
I guess on the advertising side, on the services part of the business, as long as the consumers are there,
Yeah.
fair to say that the advertisers are spending, or how should we think about that?
Yeah, there has been very consistent demand through the first quarter on the Services side from advertisers. It's very different in Restaurant, Retail & Other, where margins have been compressed, and we've most definitely seen folks shifting out of ad spend in order to either hold up margin or provide discounts, I think is probably the wave that's coming over the course of the summer, particularly in retail, but also in restaurants. So, it's much harder situation, I'd say broadly for the Restaurant, Retail & Other players from that margin perspective, and they're reacting to the lighter demand from consumers, so those two interact.
Is there anything that Yelp can do as a platform to help mitigate the trend, that trend in RR now?
So we definitely like having this high frequency set of categories because it's a reason to come to Yelp, it's a reason to come back to Yelp. But we make much, much more, more money on a given session and from a given click on the services side.
Mm-hmm.
In general, where we wanna spend our marginal dollar at the moment is definitely on the Services side, which isn't to say that we're walking away from Restaurant, Retail & Other, but the cost to recover against a macro environment on the Restaurant, Retail & Other side, I think would be very high. Whereas the marginal benefit that we can see from a revenue perspective over the next couple of years in Services, at the moment, to me, seems much higher. We've definitely decided in 2024 here to focus on the Services side and continue to invest in Restaurant, Retail & Other, but not to increase that investment.
And so if you look out a few years beyond the current macro situation, is that mix shift going to continue? Are there things you're doing proactively to continue shifting more towards services?
I mean, sitting here, we did do 1% growth in restaurant, retail, and other in the first quarter, and we did 11% in services, so you're gonna continue to see that mix shift. I do wanna underscore, though, this diversified set of categories, and as things do better, sometimes others are less good, and as they do better, it can reverse up. So I think that's likely to be as we go through the cycle, that we will see, and we certainly wanna participate in a recovery in advertiser demand, as well as consumer engagement on the restaurant, retail, and other side. But I think in the near term, it's clear that share of revenue is gonna continue to increase in services.
All right, so I have a couple of questions on this year, and then we'll talk about some of the other initiatives, particularly on the technology side. But the paid search, the in-house marketing that you talked about, can you elaborate a bit on the goals of that program, what the expected payback periods are, and how you will evaluate that, you know, through the year?
So we took 2023 to really build out this paid search capability, and in order to find a return on advertising spend, ROAS, for us in paid search, we have to be able to very efficiently convert a click from Google into a project. So the first thing that we're looking at, and I already mentioned, is this growth in projects. They grew about 20% in the first quarter, up from about 5% in the fourth quarter of last year. Now, that was a mix of both improvements to the site as well as paid search, but focusing on paid search, you gotta land folks to a good experience, one that can convert. We've done a lot there over the course of 2023. We've built out the infrastructure, but we still have more to do.
What we wanna see is those projects converting into clicks, and we wanna see CPCs come down. Those are the first metrics that we look for. Now, in the past, when we've been able to lower CPCs, elasticity of demand has been reflected in the marketplace in general, and we've seen demand increase, so lower price, more demand. That's what we are looking for in order to sustain this level of spend, but in order to get that kind of read, you have to spend at a certain level. It will take 2024 for us to ascertain whether it's working from the most important perspective, which is we do see increased advertiser demand for advertising on Yelp by increasing the number of projects and clicks in these specific categories. That's what we'll be looking for.
In general, folks, when they look at ROAS, are looking at 12 to 24 month payback. We're similar to that, and there's a couple sources of value there. First of all, is obviously the budget piece. The second, of course, is we acquire these consumers, and we hope to see them returning, and they would be in the most valuable category. So that's something that we would factor in, but the first step is to get to break even on a certain horizon with the money that we're spending, and it's still a little bit early for us to be able to determine if that's the case.
I do wanna underscore that if you just take a step back, the midpoint of our guide for this year in Adjusted EBITDA is $320 million, the midpoint, and if we hadn't spent this $40 million, it'd be $360 million. I'd also point out that we took money that we reduced stock-based comp, we increased cash comp, so there's even a little bit more than that $360 million in effect from a, you know, apples-to-apples basis. If you divide that by the midpoint of our range on revenue at $1,430 million, that's about 25% Adjusted EBITDA margin. Last year, we did 25% Adjusted EBITDA margin. That's up six points from 2019. So the business is as efficient, if not a touch more efficient, here in 2024. We're also guiding to flat headcount as we were last year.
So if we are not seeing the returns that we hope for in Paid Search, then we have the opportunity just to flow that back through to Adjusted EBITDA. So we're very financially disciplined. We look for the ROI, and if it's not there, then we redeploy the spend to other more productive areas or areas, or we flow it through.
But you're committed to the spend for this calendar year?
For 2024, I think you've got to spend this money to get the read.
In the second half of the year, I believe the implied guidance is an acceleration in revenues, so that's not due to the spend in Paid Search, that's some other factors.
So we haven't recognized or put into our guide revenue from paid search in 2024. We'd expect that to show up in 2025, but we hope to see, obviously, indications of it. So in terms of the acceleration from the first quarter, our guide does imply acceleration in the second half, and we did 1% in the first quarter. More or less, I expect the macro environment to remain about the same as we go through the year. There's puts and takes with that, and, you know, large advertisers make a variety of decisions. And that's something that we certainly saw in the first quarter, where people did pull back.
That's a little bit of the swing factor, but assuming that we were in a similar ballpark as we move through the year, then Services has to accelerate in the second half in order for us to achieve the guide that we've provided.
Okay. Now, looking beyond this year, I do wanna talk about the margin trajectory of the business. Obviously, the marketing spend, we'll understand better this time next year on the efficiency from that. But how do you think about, overall, the margin profile of the business? And I know you may wanna explain a little bit more about the shift from stock-based comp to cash compensation.
Yeah. Okay. So overall, I do believe that you can drive margin expansion over long term through a Product-Led Strategy. I think a lot of folks have demonstrated that quite convincingly. I think we've done that over the past several years, going from about 19% Adjusted EBITDA margin to the 25% Adjusted EBITDA margin that we achieved last year. Now, we've been able to do that by holding headcount flat over the past couple of years, and that means that the projects that the team are doing stack. This is really an essential component of a Product-Led Strategy, is that as you improve this site and make more improvements, those things continue to stack over time, so the contribution sustains. That's really the premise of a Product-Led Strategy.
We also did pick up a considerable amount of margin as we went to remote first as a company, and we've actually sublet a significant portion of our leases, or we've exited leases over the past several years. So then we've been able to redeploy that, a portion into Adjusted EBITDA, a portion into product and engineering investment. So fundamentally, we are very focused on continuing to drive margin, margin expansion for the long term. So that's, that's the approach that we're taking. In terms of stock-based comp, we got a lot of feedback from investors that stock-based comp was too high, and we committed to lower that to about 8% or less by the end of next year. We took a huge step forward, as I mentioned, where we expect to issue 65% fewer shares to employees here in 2024.
I think that's pretty big step towards that. But we did have to pay more in cash compensation. What we said in the Q3 call last year was that in 2023, all things being equal, we had moved about $20 million of stock-based comp into about $20 million of cash comp. But we would expect that to stack over time as we have new employees joining the company. But also just inherent to the way that the grants work, you should see some margin leverage from that as well. So, we remain committed to that. We think we've taken a big step towards that, and, you know, we'll continue to make progress there, I think.
You've diversified where your engineering headcount is, right? That's also part of-
So we started that. Yes, that's also a very good point. Thanks for that. We fundamentally stopped hiring in the U.S. for product and engineering probably 5 years ago, 4 years ago. I mean, the competition for engineers, if you'll recall, in the 2020, 2021, 2022 timeframe, was intense. Very, very intense. That softened considerably over the course of 2023 and into 2024. But we have built up our team in Canada and in the U.K. in particular, though we have a small team in Germany. There's terrific engineers in both of those countries. They contribute at a comparable level, but they tend to stay a lot longer, and obviously, if you have engineers stay longer, they contribute more 'cause they're ramped up and familiar with the code base. So that has been something that's important.
There's also a preference for cash over equity in those countries, and so that's something that we certainly were taking advantage of, from a cash, or a compensation mix perspective. Of course, paying them market competitive wages.
I do wanna ask about this, this Yelp Assistant. This seems very interesting, and I wanna talk about other product enhancements and technology as well because I think that's an important part of the story. But can you talk about the genesis of the Yelp Assistant and maybe how you're seeing people use it?
Yeah. Yeah, so I think we're all very surprised when ChatGPT first was released. It had been promised and talked about, and I think the quality of ChatGPT 3.5 was a surprise to many folks for being better. And certainly, it was something that we'd already been investigating, as I mentioned, but if you're an engineer, you wanna work on the new stuff. Certainly, large language models fall in the category of new stuff. But we also saw an opportunity to make Yelp a much more conversational experience, and maybe this is a point really worth underscoring. So we have a product called Request a Quote. That's when you come to Yelp, we ask you a few questions, we figure out what service pro you need, and then we match you with that service pro.
We gain information through that process and we realized that if you could make that more conversational, it could be a better experience for consumers, but you could probably learn more about the project. And we have absolutely found that to be the case. So the inspiration was, hey, this seems like a technology we could really leverage to create a better experience for both consumers and pros. We went through the process of validating that. We have seen that we get much better information out of consumers. That does improve the matching algorithm, and it enables us to much more precisely figure out which pros to show. Now, they like it because you get a higher value lead out of it. I would just expand on that, with the large language model in place, it gives us like a chat assistant.
It gives us experience and capabilities to extend that into other experiences. For instance, could we help service pros to improve their experience in responding to inquiries from consumers, is an area, and I think everybody's just looking broadly, how can you use large language models in customer success and in sales? We're looking even in finance and in people operations, how we can apply these models. I do think that engineers can be more productive using them for sure. There's clear use cases in customer success, and when you have all of that integrated, that data set, you're, it's, it's mutually reinforcing because you're able to share that information across the different platforms and continue to improve. So Yelp Assistant in itself, I think, again, I think it's a terrific experience, but you'll judge for yourself.
But it doesn't stop there. It's really just a starting point for making this much more conversational across Yelp. I do think, you know, it's still really early on things like voice AI, but I think there's a lot of potential there. ChatGPT-4o showed, at least in the videos, I don't know if it's actually in production, some real-time interactions that were quite impressive.
Now, I noticed in some of these, alternative chatbots that, Yelp's data is available. How should we think about, you know, your ability to potentially monetize those relationships? And then how do you think that could potentially affect the competitive landscape?
So broadly, the way we think about value here is the lowest order value is using data for training. And if someone wants to pay us, we're more than open to that, but that's not where our focus is. The second order is using large language models to process information in some way. So we recently released a product called Yelp Fusion AI API. Let me unpack this for you. So we've been licensing our content for some time. You may have seen it in Apple Maps, used it on Alexa, but that's where we get a query for a request to return search results that's already been structured by whomever's using it. That's the API. The AI API is where we get unstructured text. We process that using a large language model and return results.
That is what I mean by next level of value or second level of value. Third level of value is actually a chatbot that's been trained on your data, can interact, can extract, more information during the conversation, and it can export that in a way that you can use algorithms to improve whatever is you're trying to do, in this case, matching. So that's the third order. Once we have this AI API in place, you know, in time, it creates a set of opportunities for us to actually look at pushing our Yelp Assistant further outside of Yelp, and so that's exciting 'cause the more data you have, obviously, the better. Now, we have licensed this, particularly to the Yelp Fusion AI API, is something that Expedia is using in the Romie product, but, we also have, licensed our content to Perplexity.
If you're familiar with them, which is an LLM-based search engine. So we obviously are happy to license the content. Where we see people using our content beyond what has been deemed fair use, we're very aggressive in protecting our IP.
Yeah, I've seen it in Perplexity recently.
Yeah. And so, Perplexity is a licensee of Yelp Fusion.
Yeah. So one question popped up from someone in the audience. Uses of cash.
Uses of cash.
Capital allocation.
I'm sorry?
Capital allocation.
Yeah, capital allocation. So just from a capital allocation perspective, we're committing to—committed to returning cash in excess of our target cash balance. The way that we come to our target cash balance is we have an operating amount of cash, we have a buffer on top of that, and then we do hold cash on the balance sheet for acquisitions. We do actively evaluate acquisitions, but we're also extremely disciplined about that. Types of acquisitions that could be of interest would be things that accelerate time to market for us. We've added a lot more video into the experience. Video editing tool might be an example. In our ad tech stack, there are things like pacers and bidders that might be of interest to us that would help us to do a better job in matching advertisers with consumers.
Those are all examples of domains where we might look to make acquisitions.
What's the last search that you've made on Yelp?
Best-
... David?
Well, actually, I did, I did go last night with my nephews to have Korean barbecue, so I used Yelp to find an awesome Korean barbecue place in Koreatown. Happy to share that with you after the talk. But my favorite search of recent times was actually best chocolate chip cookie, and I happen to love chocolate chip cookies, but I also think it's instructive. We've heard a lot about, will large language models disintermediate sites like Yelp? And I actually don't think that's gonna happen for a few reasons, not least of which, and this is the phrase that I've started to use: I think the search experience is fit for purpose. So we talked about Yelp Assistant.
In services where you need to gain some more information and use that to do matching, I think a large language model makes a lot of sense. But try searching best chocolate chip cookie and do it on a variety of sites. Do you really need, like, a lengthy write-up about whether it's gooey, or whether it's thin, or whether it's a big chunky one, or it's extra buttery, or does it have pecans, or does it have almonds? No, I just wanna know, what are the best bakeries near me? And, when you do that on Yelp, you get exactly the four or five best bakeries near you, and I think that's just an example. In general, if you want the answer to a specific thing, just give me the answer. Don't disintermediate or don't intermediate it with a large language model.
I think that's the sort of likely future for search, at least in the near term.
And Yelp will be sponsoring chocolate chip cookies.
Yeah, chocolate chip cookies.
... outside for anyone who's in the audience here.
Exactly.
Well, thanks, David, very much for attending.