Yelp Inc. (YELP) Earnings Call Transcript & Summary

Good morning, everyone. Cory Carpenter, Internet analyst at JPMorgan. Pleased to have David, CFO of Yelp, with us. Thanks for coming.

Thanks, Cory, 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.

All right. [Operator Instructions] So David, just to kick off, Yelp has changed a lot over the years. Just for those newer to the story, could you kick things off with how Yelp has evolved and just the state of the business today?

Sure. So for those of you who haven't checked in on Yelp in a few years, we've really transformed the company over the past 6 years, and it's changed in 3 important ways: the first, Yelp very much grew up as a sales-driven company with growing local sales headcount leading to growth in revenue. And a number of years ago, we shifted to a product-led strategy, and we'll talk a lot more about that; the second, of course, is that Yelp is very well known for restaurants. And in the first quarter, nearly 65% of our revenue came from services. So that's the second transformation; and then the third theme that's more recent, and again, we'll get into more detail around this but we're really focused not only on continuing to generate a significant amount of organic traffic, but we're very focused on off-Yelp traffic and tapping into that, and we've done quite a few things to enable that. So overall, in terms of the state of the business, we've really invested in creating a great experience, really updating the experience and have a significant amount of momentum.

And from a reporting perspective, you split the business between services and then our RR&O or Restaurant, Retail and Other. Could you just walk through at a high level, the trends impacting each of those segments? And then we'll go deeper into each.

Yes. So on the Restaurant, Retail and Other side, we talked about this on the Q1 call, we actually talked about on the Q4 call. We started to see this in late December actually. It definitely carried through the first quarter on the Restaurant, Retail and Other side, which is inflation continues to be a real pressure, both for consumers and for businesses. On the consumer side, things have continued to get more expensive. And I think consumers have become more resistant to the price increases that, particularly in Restaurant and Retail, those companies were able to pass through. On the company side, the input costs have continued to rise. You've all seen, I'm sure, the increase in minimum wage, particularly in California, but it's not unique to California. So labor costs have gone up significantly. Input costs continue to rise and margins have been compressed. So when you combine weakening consumer sentiment with pressure on margins, that's definitely impacted advertising and Restaurant Retail and Other from where we sit. On the services side, in contrast, through the first quarter, momentum continued. It does seem like the significant increase in home values has really enabled homeowners to continue to spend on projects. They may not be doing the biggest projects but they are continuing to do projects. And in the first quarter, we continue to see real strength in our home services categories and across services broadly. So it's quite different between those 2 sides of the business right now.

So let's start with services. I think you mentioned about 65% of your revenue. What are the main subcategories within this segment? What's resonating well right now that's driving the double-digit growth you've been seeing for a couple of years now?

Yes. So services itself encompasses quite a broad set of subcategories. The one that we are really focused on is home services, but it's not only home services, it's local services, it's auto, it's professional. It's actually pets like grooming, things like that. So it's a broad set of those categories. And particularly in home services, we've really been focused on creating the best experience for consumers meeting pros in order to do the projects they need to get done.

So last quarter, and you mentioned this earlier, you announced plans to expand paid search to acquire services projects. What did you see in testing that gave you the confidence to ramp this up? And kind of why did you get to that $40 million or over $40 million number in terms of level of SEM spend this year?

So just to step back for those of you who aren't familiar with the product, we have a feature called Request-a-Quote. Request-a-Quote is meant to enable a service pro to respond to an inquiry from the consumer and we take the consumer through traditionally what was a question tree to really refine what it was exactly that they were looking for. Now we've introduced our own AI chatbot, which we call Yelp Assistant, which we can talk more about as well, which we're really excited about but that's gathering that information, and then we take it, we go -- we use a matching algorithm and we match a set of pros with that consumer. So one thing that we said that we'd start doing at the beginning of last year was build out the capability to acquire traffic off-Yelp through paid search. And the way that we are doing that is we are advertising on Google, and then we are bringing people onto Yelp into the Request-a-Quote flow, to create projects. Now projects in the first quarter were up approximately 20%, that's overall compared to the fourth quarter of last year when they were up 5%. Now that's a combination about the improvements that we're making to get people to use the Request-a-Quote product, but also bringing people into that funnel. And we increased spend in the fourth quarter over the first quarter from -- to $7 million from $2 million in the first -- fourth quarter of last year. And overall, for the year, we said that we'd spend $40 million or more on acquiring projects through paid search. So $40 million was [ $7 million ], [ $33 million ], about $11 million a quarter, roughly through the rest of the year. And the reason that we've taken the level of spend up is several fold. The first, the metrics that we look for, first and foremost, of course, is are we actually growing the total project volume, that's the first indicator. Now Yelp is a marketplace. We are matching supply and demand. So the supply is consumers coming to the site, demand is advertising dollars, we run an algorithm to find the market clearing price. And so when we bring more projects in what we'd expect to see is increasing clicks in the categories where we're buying and declining CPCs. And we'll get to how that interacts with revenue in a moment. But those are the first metrics that we look for, increased projects, increased clicks and declining CPCs in the categories where we're buying projects.

And could you just talk a bit about what categories you're initially focused on for paid search in a bit -- maybe a bit more on how you're measuring ROI?

Yes. So the -- we're looking at Home Services categories in particular. And it's across a number of the fairly obvious ones that you'd expect in those categories. But we're testing. We're looking at the different categories, and we're also doing it by geography. One of the things that's very important, obviously, with paid search to get to ROI is to be able to purchase projects in the categories that you care about at a price that is going to yield a return in the geographies where you want them. And so 2023 was very much around building that scalable platform. I'd say we are still relatively early in the total capacity or capability of that platform, but obviously, advanced enough for us to spend $40 million or more. But it's that targeting by category and geography that ends up mattering a great deal. And then in time, what we do want to do is be able to also focus on particular audiences, homeowners obviously would be attractive for us, and we want those folks to come back. So you need the economics -- for the economics to work, you need to bring someone in, get the matching, have the advertising, have advertisers spend more with you and get consumers to come back. So that's still very early in terms of being able to determine what the ROI is. And we are very much experimentation driven. I believe we're very disciplined financially. And so as we see those results come in, we'll adjust the spend depending on the performance, but obviously, we want to generate margin out of this effort.

So you kind of alluded to this earlier, but on earnings, you talked about this dynamic of a lag between advertisers seeing the lower -- or seeing the clicks and the lower CPC and then improving their ad spend. Could you just expand on that dynamic and then also how it impacted your outlook this year?

Yes. So we're definitely, as I said, a finding the market clearing price, so supply and demand. What does that mean in effect? It means that we are not trying to control clicks or CPCs. It does mean that if we deliver more value through lower CPCs or more competitive CPCs compared to other sites, we believe that will generate more demand. It's a very simple premise, which is that there is elasticity and advertiser demand for advertising on publishers like Yelp. And that takes time to show up because those publishers actually need to see that improving return on their investment in advertising on Yelp. And so well, in the past, when we have done this, we've seen clearly that we've been able to generate more ad budget. That's not always true. But in general, that's what we have seen. And so that's what we're banking on in terms of the current effort. But of course, we have to see it all play out.

Let's talk more about Yelp Assistant, which you mentioned. So conversational AI feature. It helps match consumers to service pros. Could you just expand on how this works and the initial customer response you're seeing?

So obviously, we've all been very focused on the advent introduction, widespread use of LLMs, we're very, very excited about this technology. And just for background, Yelp has been using AI models for more than a decade. So this isn't something that's new to us. And I just underscore the very important element of being able to leverage these models as a publisher's first-party data. The technical capability to train those models and then, of course, to deploy them in a way that generates revenue. In terms of Yelp Assistant, we have built our own chat bot. That chat bot, you can find if you download the app or use the app in the project tab, it will be in the lower right corner there, so you can see it for yourself. And it dynamically will ask follow-up questions when you start to talk about a services project. Now there is a supervisory model resting on top of that, like other sites to prevent the person from trying to make queries that are really not relevant to Yelp or get the chat bot to say things that maybe it shouldn't. But once that has been cleared, personally, I found that it's very intuitive. And the questions that it asks aren't always obvious when you go in to starting the Request-a-Quote process and make you think about what it is that you need to accomplish, we gather a lot of information out of that and improves our matching significantly. So from our perspective, the more information that we can gain with the fewest number of questions, the better off we are being able to match and create a great consumer experience. So we've rolled that out in the project tab to 100% of iOS app users. And we're looking at how to introduce that across the rest of the site. So one thing that we've -- will also -- is also a theme is, well, when are you going to roll out a chat bot for every category. And one of the things that we've learned recently that's really, I think, very interesting and was a bit of a [indiscernible] for me was, not every query wants a chat bot in order to efficiently find the results. And let me give you my current favorite example. I happen to love chocolate chip cookies. So I looked for best chocolate chip cookies. And Yelp, and what comes u, that's chocolate chip cookies where I live. And it came up instantly. Now if you do that same search on Google, it varies because there are often changing the interaction. It also produces a list of bakeries with great chocolate chip cookies. Do you need an LLM for a chocolate chip cookie query, not really. So I think what we're going to see even on Google you'll see, even though they're rolling out these summaries, you'll see that for some queries, they're not going to summarize because there is nothing to summarize. That's not the most efficient way to get the consumer to the business that they're looking for. So that has been very interesting. I think it's going to evolve a lot. I think consumer preference and for how to interact is also going to evolve in time as we all figure out how to deploy these models.

My next question was how you plan to roll it out to other categories. I think you answered it. So I want to talk about, you announced a deal with Perplexity, you announced an AI API, you just talked about your approach to data licensing and how you think about this as a monetization opportunity or something you want to keep in-house to power things like Assistant.

So we have been monetizing data for some time. We have 2 products, Yelp Fusion. That's where we're providing content to other sites. I'll come back to Perplexity and AI API. And then we have something called Yelp Knowledge, which is much more for marketing to businesses. So it's a B2B insights product. We do sell both of those -- that revenue shows up in other revenue. And we have quite a few different customers existing for the Yelp Fusion product. Again, that's the content piece. So we're in Apple Maps, we're on Alexa. And most recently, we licensed that to Perplexity for their queries. For those of you who aren't familiar, Perplexity is a LLM based search site that aims to compete with Google to produce results. So we're definitely participating there. Now the AI API is -- so we have the API, which is I'm going to return results from Yelp. The AI API is where we get a free form tax input, process that with a large language model and then return results. And that has -- is an advancement because obviously, the site does not need to worry about filtering that query for us. We're doing all the work. And we're working -- you may have seen that Expedia is working on their own chatbot. I think it's called Romie that's in early stages of testing, but that's a place where we are powering content for them. So Perplexity is one, Romie is another. Now we obviously want to continue to license our data. But I think there's really 3 layers to value in this entire ecosystem. The first is training data, which I think is the lowest order value. One thing that has definitely emerged as a theme for folks who are building out their own LLMs is high-quality human text. And one of the beauties of Yelp is it's user-generated content. We're constantly generating more of this text. So it's a great source of value to us, but the lowest order of value is for us to sell that as training data. The next step up, of course, is be able to do something with that and that's where you're using a large language model as an input with that underlying set of data to improve an experience. So we have something called review snippets. That's where we're pulling out little pieces of text from the reviews to really summarize what's going on with a particular business. So I'd say that's the second order. What we're talking about with Expedia is another example where that AI API, you're processing the query and returning the results. The highest order value, though, is something like Yelp Assistant, where you actually have this interaction where both the consumer gains value, you gain value and the value gain you're able to pass on to the Service Pro. And that's where we're really focused. We're really focused on the second level and this third level of value as opposed to this first level of just selling training data. especially since we think that we have a very unique position with the breadth of data that we have, the quality of data we have and our ability to generate insights from that data.

So moving to Restaurants, Retail and Other, which is now the minority of your revenue. You mentioned this earlier, just some of the headwinds facing the category. I mean we've heard from a number of QSRs about kind of pressure on the lower end consumer. Question for you is how broad-based is what you're seeing today? Like is this just happening at larger restaurant chains? or are there other categories impacted as well? And maybe if you could just give us some examples.

Apologies, Cory. Let me just go back one last point to make on services, and then I'll come back to Restaurant, Retail and Other. Just as another example of the way that we are really focused on acquiring off-Yelp traffic. So I talked about paid search, I talked about what Expedia is doing. I talked about Perplexity. You may have seen it. We're working with Mozilla and Firefox. So we're returning search results as a person types in a query that can link them to Yelp. And then just in the past few days, Apple has rolled out a feature in Apple Maps, which is called Quote. And that's for services queries, so not a restaurant or a hotel, but they're actually looking to say, for a plumber, electrician or [ small ] check. And when you press Quote that takes you into our Request-a-Quote flow. So just this theme of broadly accessing consumer traffic off of Yelp to connect them with great local businesses through these other platforms is a big theme for us. Now just turning to Restaurant, Retail and Other. We have seen this as very broad-based. It is not unique to the restaurants category, where consumers appear to feel pressured and businesses also seem to be squeezed on margins. So that's definitely at the enterprise level, the mid-market level and in SMB.

Okay. Also, I wanted to ask on the earnings call, you mentioned perhaps seeing in that channel some spin to platforms such as [indiscernible], DoorDash. Could you expand a bit on what you're seeing there in terms of allocation of budgets? And then how big of a concern is this to you?