Good afternoon, everyone. Welcome to day three of Morgan Stanley TMT. My name's Matt von Baesty . I work on the Internet team here at Morgan Stanley. We are pleased to be joined by David Schwarzbach, CFO of Yelp. David, thanks so much for being here.
Thanks, Matt.
Absolutely. Got to go through a couple of disclosures, so just bear with me. For all disclosures, including Morgan Stanley Personal Holdings disclosures, you can find them on the Morgan Stanley public website at www.morganstanley.com/researchdisclosures. Some of the statements Yelp makes today may be considered forward-looking. Forward-looking statements are subject to a number of risks and uncertainties that could cause actual results to differ materially. Forward-looking statements are based on assumptions as of today. Yelp undertakes no obligation to update them. Please refer to Yelp's Form 10-K for discussion of the risk factors that could affect actual results.
Thank you.
With that, exactly.
That was the best reading.
I haven't memorized it at this point. Hopefully so. So we got through it.
That's good.
I want to start high level.
Yeah.
You've been at the company for five years. A lot has happened in the last five years.
Yes.
When you think about what's changed at Yelp, when you think about Yelp's position in the broader ad ecosystem, how would you frame that for investors, especially with what you've seen over those last five years?
Yeah. Again, Matt, it's great to be here. I appreciate it. There has been a tremendous amount of change for us in most companies over the past five years. I would say, most importantly, Yelp has gone through a very significant transformation and re-acceleration in the business, coming out of or going through COVID and coming out of COVID, and the most important shift really was moving from this sales headcount-driven growth model to a product-led growth model, and we've really built up that capability, which has been very important from a competitive position perspective, and we'll talk AI, but it's extremely important in this era as AI emerges, so that's one very, very important transformation that we've gone through.
Then I would say the second thing, which is probably still somewhat underappreciated, is that Services now comprises about two-thirds of our revenue compared to where it was sort of prior to when I joined and prior to 2020. We've really leaned in on Services. We have a great product there. Again, we'll talk about that as well. And then I'd say the last part that's really changed over the past five years is just the broad opportunity for us, whether that's continuing to expand our footprint in Services, but equally really taking advantage of the opportunity to deliver value to enterprise advertisers and not just small and medium-sized businesses. So I'd say those are the three big things that have changed over the past five years.
That's really helpful. We're going to dig into a lot of those product-level things in a moment. I want to stick on macro for a second. Macro has obviously been a bit of a bigger focus for people as of late, some pressure on the R&O category given consumer weakness. When you think about how macro is affecting the overall business more broadly, could you call out a couple of areas where you're seeing that? And how do you think about potential signs of stabilization on the macro side of things?
In 2024, our services business grew 11%. Our Restaurant, Retail, & Other business actually shrank, and we saw real pressure on these businesses from input costs rising, and so inflation for, in the restaurant side, say, ingredients, but this has been broadly true across Restaurant, Retail, & Other categories. Those direct costs have increased. Equally, there's been a significant amount of wage inflation, and quite a few states increased their minimum wage in 2024. California went to a $20 minimum wage, and that certainly affects all categories. It didn't apply to all areas, but it definitely has put pressure broadly, so as input costs have gone up, these businesses have not been able to pass along those increases to their customers or to consumers. In 2022 and 2023, consumers were more willing to accept price increases. That was not true in 2024.
And so what you had was this compression of margin. And typically what happens, as you know, is when margin gets compressed, there's more pressure on advertising dollars. And that's what we definitely saw in Restaurant, Retail, and Other as we went through the year. On our earnings call, February 13, what we shared at that time was that the dynamics that we saw in 2024 had persisted as we came into 2025. And on the call, we said it's unclear when that is going to recover. Our focus, though, is to be positioned to participate in that recovery when it does occur. And we're obviously still well known for restaurants in particular, but other of those categories that fall under R&O. And so we do invest there, but our incremental investments have now been going to services for quite a few years.
Gotcha. So when we think about, let's say, the macro environment persists as it is now, and we think about maybe one or two areas of focus for you, you mentioned it briefly, that you can hone in on to accelerate growth in 2025. What are maybe one or two callouts that you would say are most important?
Yeah. I didn't really touch on the dynamics in services. So the dynamics in services have continued to be solid and robust for us. Obviously, 11% growth, we were pleased with that. And so I would say, first and foremost, our focus in 2025 is to continue to drive services performance. 2024 was more about home services. When we broadened that out, we did an acquisition. And we continue to believe that's an opportunity. When we think about where that incremental investment goes, again, we'll get to it in more detail, but we built a chatbot for services called Yelp Assistant. We've also continued to really enhance the ad technology. So we have a deep ad tech stack. We rebuilt that on neural nets.
And just in terms of consumer experience, which does go to Restaurant, Retail, and Other, we really revamped our Home Feed, made that much more visual, made it more relevant based on signals that we have on consumer interests, and modernized it. And so those things are obviously very important because consumer expectations of their experience with apps is ever-rising. And if you're not keeping pace, then you're falling behind. So that has still been an area of important focus for us.
And if we were to shift to more of a channel view of the business, multi-location, self-serve, local, how do you think about your outlook for those respective channels? And how do you think about respective growth as we head throughout 2025 there?
So going back five years ago, self-serve and multi-loc comprised about 30% combined of our ad revenue. That grew to 50% last year. And so we really feel like we made huge strides there. And that continues to be very important to us. Obviously, we really like self-serve because of the low cost of advertiser acquisition. The other thing is when you go through the self-serve flow, you're more likely to stay with us. You sold yourself. You learned the tool. You're familiar with the capabilities that are there. You're used to checking in. So we like that a lot. On the multi-location side, we definitely made progress. Now, a lot of our customers were restaurant retail and other enterprises. And so they have definitely been under pressure.
We see a very large opportunity, though, to go after enterprise customers on the services side, which for our services business ad revenue, only 20% comes from enterprise customers, so that clearly is a continued opportunity for us, and there's a lot of things that we've built out on the technology side, like APIs, in order to facilitate that business, but we've also had to do a lot to work with them and figure out how to integrate with their processes so that they can take a lead and convert it into a customer.
I want to hone in on some of the product improvements that you mentioned. You talked about the Home Feed revamp. You have business summaries. You also have review insights. When you think about how engagement on the platform has changed with those new user experiences, anything you could talk about there? And what are some of the signs that you're seeing as you've rolled those out?
I think everybody obviously is experimenting with AI and figuring out ways to use it best. One of the things that large language models are particularly good at is extracting information from a large amount of unstructured text data. We've really leaned into that because we are text-based. People are writing reviews. We're obviously known for the quality of those reviews. That's what distinguishes us. Being able to really, in context, figure out what is the most relevant snippet to surface. It has the beauty of it being something that was written by a person from their real experience. We love that, but you don't want to just pull out any random snippet. You want to pull out a snippet that's relevant to that query. I would just say that's also very important that we're query-based.
So we have clear clarity on what it is that the consumer is looking for. And then we have a lot of other signals. So we really like that. That goes true for, hey, in the Home Feed, what am I going to present? That may not be a large language model, but it is neural nets, which is effectively the same technology underneath the hood. And then just this idea that you can expand how relevant the information is to the consumer using a chat experience, I think, is really, really important. And as I mentioned, we've built out for services our Yelp Assistant.
On the Yelp Assistant, you rolled it out in services. What has been the consumer feedback so far? And how do you think about potentially rolling that out to other parts of the business?
Overall, it's been well received. We definitely think it leads in terms of chat experience for this purpose around services. From the third quarter to the fourth quarter, we were able to increase the number of projects being created through Yelp Assistant by 50%. So very substantial increase there. It's really powerful in that because it's dynamic, it's conversational, we're able to extract more relevant information from the consumer, which helps us to match them with the right service pro. So all of those things are really good and we think makes for a more relevant and engaging experience for consumers. On our earnings call, Jeremy mentioned that we were going to be looking to expand Yelp Assistant to all categories on Yelp. And I just think that's an exciting opportunity for us. And it's not just all categories, but it's every entry point to Yelp.
A reason why that's valuable to us, of course, is, first of all, this being able to get the right information, the relevant information to get better at matching. We want to extend that everywhere. There's another dimension to it is that once you're in a conversation with the consumer, you have an opportunity to ask them some questions like, do you have any other projects? Is there anything else I could help you with? We know that you came and searched for this in the past. Is that still something that you're interested in? It really changes this text box query where I'm going to type a couple of words to a dialogue with the consumer. I think that presents an enormous number of opportunities for us.
Because we have this very trustworthy content underlying it, it also means that the consumer can ask about all kinds of things that aren't immediately obvious from clicking on a link. I think we're really well positioned to leverage this large language model technology to apply it broadly to the experience. Not everybody, however, wants to use a large language model. We're going to preserve those other entry points for folks who are like, hey, I like a simple query. I'm happy to click through. And I want to reach out to that business myself. We're not getting rid of that. Consumer preferences will differ. We're going to be open to that.
Are there any complexities that exist in the other parts of the business that you're rolling this out to that maybe didn't exist in services? Or should it be a fairly seamless rollout?
One thing that we have really learned is that it takes a lot of iteration and product work to create a better experience than the status quo. It is not simply add LLM magic happens, it's better. So the short answer is we've tackled one of the harder problems, which is creating a chatbot for Request a Quote. And if I can just detour for one second, one thing that's very important to appreciate about services on Yelp is this is a workflow. It's not just an ad. When I go to Yelp and I say, I need a plumber, it's not just any plumber. It's a plumber that has the experience or is focused in the area that I actually need them. That's true of almost all services jobs.
So it's, hey, gather the information, figure out exactly what the need is, figure out who would be most relevant to return as a result, and connect them through our Message Center and help facilitate the transaction. That's a whole series of steps. It's not simply click on an ad, make a call. So when you build a chatbot to manage that experience, it's not just out of the box. It's going to be unique to your environment and the way consumers use your product historically while also helping them to understand how they might be able to use it differently in the future. So as we think about rolling that out, not just to services, but just to the rest of restaurant retail and other, there will definitely be this product iteration that's required. But we have the know-how. We've done this.
We have the technology, the tech stack, the first-party data. We're very good with large language models. We've actually been using them for quite a few years, even before the big announcement from OpenAI. And so when you put all of that together, we have the capability, but there is no "add LLM and magic occurs." You've still got to do the product work.
It's clearly been a lot of thought that's gone into your AI work over the last couple of years. How has that shaped how you think about investing in AI going forward? Where are your priorities? What initiatives are you most focused on as we head throughout 2025, 2026, and so on?
So we are very, very disciplined from a capital allocation perspective. We look at the ROI on all of our incremental spend, and whether that's incremental ad dollars, our own advertising, whether it's incremental headcount, and for incremental headcount, whether that's in sales or engineering or other parts of the business in order to earn a return. So we're very rigorous about that. We have a very, I think, rigorous planning process that sets the priorities. And then through experimentation, we are agile in redirecting those investments to the places where they're working and taking them away from areas that may not be working. So I think that that's a good product cycle with good feedback loop there overall. Now, as we think about where should these incremental spend go with regard to large language models, as I mentioned, it's not just the large language model.
It's really the application of neural nets pretty much across all of the experiences, the ad tech stack, the Home Feed, as well as the chatbot itself. So we have directed resources against AI and large language models, but there too we're very disciplined. We're looking for a return on that incremental investment. So that continues certainly to be an important source of investment for us. The other thing I would just say on AI is that there is also improving the productivity of the enterprise at large. And so we are definitely looking at ways to leverage this technology to make our people more productive, whether that's in product and engineering, which I think has certainly been at the forefront, but we're also looking at customer success and the post-sales experience, the sales experience, and even in G&A, where can we successfully apply this technology?
And I think there's been certainly a lot of statements out there about how much productivity can be driven. I'm probably a little more skeptical or conservative on how you apply this technology into driving productivity, but I certainly think there's real potential to year in and year out increase productivity for the organization pretty much in every area.
You mentioned the application of neural nets to the ad tech stack. When we think about the tools you're rolling out for advertisers, these AI neural net-based tools, what have you seen from an adoption perspective? What have you seen from a ROAS uplift perspective? What has been the response to some of these capabilities?
So for folks who are unfamiliar, one thing that's important about the way that we work and work with advertisers is advertisers give us a budget, and we optimize that budget on their behalf. The cost per click that's imputed is based on an auction, but it's really a way for us to determine who should rank and how should we absorb or deploy the budget. So we're almost already, we're almost entirely doing everything for the advertiser in the first place. And so we're applying lots of AI. Now, at the margin, there are areas where advertisers can make choices. For instance, hey, what photos do I want to display on my business page? We have a tool that lets them recommend which photos to show. They could also pick the photos they prefer. That's like a very small part of it.
But obviously, in the photos that we're recommending, we think those are the ones that will get them the most clicks from valuable leads. And so I would say we are really at the center of the optimization function as opposed to providing lots of knobs for the advertisers themselves to tune. The other thing I would simply say is that we're obviously very large with small and medium-sized businesses. They don't have the time to spend trying to optimize their advertising. That's why we're doing it on their behalf. Overall, we definitely believe that we continue to deliver additional value to advertisers. One way to measure that is what is the revenue per paid advertising location? Restaurant, retail, and others have obviously been challenged. So that has gone up, but that's more because the base has shrunk a bit. But on the services side, it's gone up dramatically.
We think that's a really strong indication that we're delivering ROAS to them.
How do you talk about the small and medium-sized advertisers who may not even be familiar that Yelp offers these tools? How do you go about that process of familiarizing advertisers with the suite of capabilities that Yelp offers?
When you start a business in the United States, you're going to register. We're a directory. If you want visibility for your business, one of the things that you're going to do is you're going to register on Google. You're going to register on Yelp, and we really think there are these two directories. It's Google and us, so if you don't want to work with Google, we're the directory that you're going to use. Now, when someone goes and signs up on Yelp, that's exactly our opportunity to, of course, make it easy for them to put their information on the site, but it's also the opportunity for us to familiarize them with the ads that they could choose to use, and that's why I mentioned earlier we really like self-serve because I go through that flow. Everyone has to go through that flow.
Now, if they do not end up buying advertising, it turns out that we have a great opportunity for our sales team to reach out to them. So in a way, it makes the whole sales funnel more efficient. But personally, I think one of the real advantages from self-serve is the fact that they're familiar with the capabilities. We're able to introduce those capabilities to them. And so they're comfortable and familiar, and then they can judge whether the return is there and they increase their budget.
And when we think about your new advertiser acquisition strategy, are there specific advertisers you're targeting? And longer term, how do you go about maintaining that relationship now that they're familiar with the capabilities? They've been on the platform for some time.
Sure, so new business creation, very important part. Now, we also, because we're a directory, we know all the businesses. So we can also call them back. So we do that as well. In terms of after they've signed up, that really does flip over to the post-sales side, and so we're doing outbounding. We're doing email. We're engaging them. I should just clarify. We're really responding to inbound calls. We don't do a lot of outbounding on the post-sales side because that's very expensive. But we are in product reaching out to them, and then we're using email to prompt them to log into what we call the Business Owner Account page, and then there's in-product marketing that presents an opportunity.
If we just extrapolate from what I said on the consumer side, where once you're in a conversation using an LLM, there's an opportunity to prompt someone to perhaps do more, I do think in time we'll have that same opportunity on the business owner side.
When you think about all these initiatives and you think about allocating capital to the various different focuses of the business, what are the key KPIs you look at? How do you evaluate the return of these respective initiatives? And some may be a little bit more tangible. Others may be a little bit more high level. So how do you think about that process? And even between AI and non-AI initiatives, how are you making that evaluation as you go?
So we are very experimentation-driven. I would say our method of development is experimentation-driven agile development. And so we start with the hypothesis. We have some data informed that hypothesis. And then we're looking for ways to test it. Now, to your point, and then once we run the A/B test, we're figuring out what the return is. Now, you can't A/B test everything, and sometimes you get uncertain results. And so really then the question is, is this product effort, for instance, in line with the strategy that we're setting? Or, hey, we know we need to modernize, and so we're not going to A/B test modernizing every element of the Home Feed. You need a Home Feed. It needs to be visual. That's pretty clear. That's the way consumers prefer to interact with information today.
But all of the subtleties of that interaction are, by and large, getting A/B tested on the site. So we love data. We love to run the experiment. And then we definitely act on the results that we see. So that's the overall capital allocation process. We think we do that quite rigorously. And then I would just say more broadly, as we think about capital allocation, we committed to flat headcount here in 2025. We were flat headcount in 2024. We were flat headcount in 2023. So this thesis that a product-led strategy can lead to margin expansion is very central to our approach. So put differently, we have set this constraint, which is without more people, how are you going to continue to grow the business and drive margin expansion?
That in itself, I would say overarching, drives a capital allocation discipline that has worked really, really well for us. So even aside from the A/B testing, I've got to make a choice. Where is the best place for me to put my effort? Every person, team needs to make those decisions. And I just think that's an overarching framework that has led us to be able to expand margin by a number of points over the past several years. And that's flowed through to EPS was $0.50 in 2022, and it was $1.88 in 2024, 40% growth in EPS last year as a reflection of, I think, both the growth in the top line, but also the margin that we've been able to drive.
I want to dig into margins a little bit more. Maybe when we think about the Q1 margin guide implies 25% full-year guide implies around 23%-24%, how should we think about margin trajectory throughout the year, and how should we think about maybe those areas where you do want to put a greater emphasis on the investment side of things versus areas where you may see leverage under the surface when we think about the building blocks to that overall margin?
There's probably three important considerations around the margin for 2025. The first I would just say is RepairPal has a lower margin than Yelp. And so there's just an impact from that margin. What we said when we made the acquisition was that they were about break-even. So obviously you have more revenue, but we're not focused today on margin. We're focused on the revenue opportunity associated with RepairPal. So that's one component. The second component is obviously it's still very early in the year. We pointed out there's quite a few risks and uncertainties in the macro environment right now. I think that's an important consideration for us. So we need to have a little bit of flexibility there. And then we do want a little bit of flexibility as we go through the year in order to double down on things that are working well.
And I would just point out, in 2024, we entered the year. We said that we were going to spend on paid search, consumer acquisition for projects. And we shared with investors our expected spend on that. We also said that if we did not see the return that we expected, that we would not spend. And that's exactly what happened. We started out the year. We saw a lot of good signal. As we continue to iterate and refine, we realized that the ROI wasn't quite as strong as we anticipated. And we ended up taking that spend down. And we were able to flow that back through to EBITDA. So I'd just say philosophically, we're going to be disciplined. If we can't spend it productively, we're going to flow it through to margin.
And I would also say that it is really important to us to ensure that we are driving margin for the long term. So long-term profitable growth is the approach that we take overall. And what you saw in the fourth quarter is that as we exceeded guidance on revenue, we flowed that basically 100% through to EBITDA. And that philosophically is how we approach it. And if we're flat headcount, you're basically a fixed cost business. And so you get a lot of leverage. If, say, R&O were to recover, that gives us the opportunity either to reinvest it or to flow it through. And I think that we've shown discipline on that. If we're going to reinvest it, we want a return. If we're not going to invest it, we are going to drive margin.
That makes a lot of sense. You mentioned the focus with RepairPal on the revenue opportunity near term. The acquisition strengthens your autos vertical. How do you think about the acquisition strategy in 2025? Are there other verticals, subverticals that you think you can execute a similar strategy with? And maybe longer term, how do you think about potential synergies that could exist more on the margin side of things following the acquisition?
Sure. So we do hold cash on the balance sheet for acquisitions. And we used balance sheet cash to make that acquisition. The acquisition itself was a little more than a quarter of free cash flow. So obviously it was something that we could absorb. And then the way that we think about these acquisitions, does it fit in our strategy? Is there something they bring that makes us better? Is there something that we could apply that would make them better? And are there other areas that we could go into to drive that top line performance? Now, in the case of RepairPal, they do repair. They don't do maintenance. They don't do collision. They don't do warranty work and/or recalls. So we think that there's ways to continue to drive that. But I think we all agree, if you make an acquisition, don't break it.
So the first step is, they, get them on board, make sure it's operating, get them comfortable in a public company environment. So that's the stage that we're at. But we're absolutely looking for other opportunities to lean in to continue to drive our services broad categories. And then within that, other areas that are professional services or local services would be areas that we could be interested in. But we definitely continue to have cash on the balance sheet for acquisitions. And we're actively looking for those opportunities.
When we think about Yelp's competitive positioning, specifically in search, how do you view that competitive position today? How do you think about potentially innovating in that area going forward?
So I think overarching our view is that search is moving from results to answers and actions. I think it's really early on the actions side, but it's certainly emerging answers as opposed to links. And because we have this first, the really important thing is we have first-party data. We go to great lengths to ensure that it's quality data. And we do have a trusted brand so that when we surface information, and you get the review, get the rating, you can see for yourself. And that's actually by someone who's used the service or visited the restaurant. We think all of that positions us really well to provide people with answers. And recently, Amazon announced their revamped Alexa. They mentioned us as being able to enable the agent part of that for booking a restaurant reservation. So we think that we have the content first and foremost.
We have the delivery platform for that. We're investing heavily in these large language models. And we have a set of capabilities that enables us to take action on behalf of the consumer. So when you combine all of those things, we're actually excited by the opportunity that this technology presents us.
We've seen the emergence of several agents, but it is still early days. You mentioned the Alexa partnership. How do you evaluate those partnerships when they come up? What are the key factors you're thinking about where you say, okay, we do want to plug into the next agent that emerges? And what are the key considerations there?
One of the things that I would just say is really important to us is our hundreds of millions of reviews are our competitive differentiator. So what we don't want to do is give up our competitive differentiator by making this freely available for training. So that's the first thing. Also, it's really for us low value to sell our data as training data. So just to say that. But what is important to us, of course, is to surface Yelp data across the internet, not just within the four walls of Yelp. So when you have a partner like an Amazon or an Apple Maps, we're in the OEM, the infotainment systems for many of the car manufacturers. Perplexity uses our product. We want that reach.
And if you think about it, going back to if there's two directories for businesses in the United States, Google and us, then we think this is a valuable capability that folks want. And we obviously want to be fairly paid for the value. We want to deliver more value. We'd love to power a chatbot for services on other sites. We'd be completely open to that. We have an API to support that. But in general, where we see ourselves positioned is value added. We don't want to be in that commodity side of answers for queries that consumers might have.
That is very clear. We're in the last couple of minutes here. I'm going to try to squeeze in two if I can. You talked about giving yourself the opportunity to double down if things are working well. What are maybe one or two things that you're most excited about heading into 2025 that you think could be areas that could warrant some doubling down if you saw what you expect to see?
Yeah. I mean, for sure, this idea that we can extend the Yelp Assistant across all of Yelp, far and away at the moment, is probably my leading thing. I'm also intermediating the interaction between consumers and service pros is really interesting. We revamped our Message Center. We've done a lot of work there. But at the end of the day, if you accept this thesis that services is not an ad, it's a workflow. And one of the most important parts of that workflow is the communication between the service pro and the consumer.
If we can make this an experience that consumers feel comfortable with and they're willing to share their information, and if we can make it valuable to advertisers by intermediating that conversation, there's a lot of technology that gets applied there, including large language models, then I actually think that's a terrific way for us to not only be perceived as adding value, but actually adding value because it's a qualified lead for a service pro who's responded in the best way they can, and that consumer decides to use them. That's at the heart of the value that we can deliver to consumers and advertisers.
That's very clear. On the flip side of that, maybe one or two of the biggest challenges as you head into 2025?
I mean, again, I would just say the macro front is probably top of mind. Obviously, we'd love Restaurant, Retail, and Other to recover and participate in that recovery. I think that's, I mean, probably far and away for me the top of mind.
Very clear. Well, David, thank you very much. Really appreciate it. Thank you, everyone, for joining. Enjoy the rest of your time at the conference.