Trivago NV (TRVG) 2018 Q1 法說會逐字稿

Good day, and welcome to the Q1 earnings release 2018 conference call. Today's conference is being recorded. At this time, I would like to turn the conference over to Matthias Tillmann. Please, go ahead.

Thank you. Good afternoon, everybody. Welcome to trivago N.V.'s Financial Results Conference Call for the first quarter ended March 31, 2018. I'm pleased to be joined on the call today by Rolf Schrömgens, trivago's CEO and Managing Director; and Axel Hefer, our CFO and Managing Director.

The following discussion, including responses to your questions, reflect management views as of today, April 25, 2018, only. We do not undertake any obligation to update or revise this information. As always, some of the statements made on today's call are forward-looking, typically preceded by words such as we expect, we believe, we anticipate or similar statements. Please refer to today's press release and the company's filings with the SEC for information about factors, which could cause our actual results to differ materially from these forward-looking statements.

You will find reconciliations of non-GAAP measures to the most comparable GAAP measures discussed today in our earnings release, which is posted on the company's IR website at ir.trivago.com. I encourage you to periodically visit our Investor Relations site for important content, including today's earnings release.

Finally, unless otherwise stated, all comparisons on this call will be against our results for the comparable period of 2017.

With that, let me turn the call over to Rolf.

Welcome everybody. Many thanks for joining our first quarterly earnings call for 2018. It's the 3rd of 4 challenging quarters that we had expected in our guidance mid of last year. Just to remind everybody, we grew revenues nearly 70% in the beginning of 2017, and competing with these numbers with a very low commercialization this year was not an easy task. A new challenge comes also with new experiences and learnings for the organization. The last quarter meant to show a lot of discipline to use our resources even more wisely than we did.

It's impressive to see how fast everybody adapted to that mentality, while the overall culture of the company did not change. We have seen a lot of ups and downs in the history of trivago, but we believe value is created over longer time frames. And for that, it's important to stay focused and to consequently execute on our mission.

As a two-sided marketplace, we have 2 customer groups that we want to make happy. On one side, we have our travelers that we want to give the best overview and lead them to the ideal hotel. And despite the strong comp set made difficult for us to ramp up our advertising spend, like in the previous years, we were growing globally our brand recognition and significantly increasing our visitor base on a year-over-year basis.

Our tech team worked hard to rebuild the infrastructure of our search to make a real matching between user needs and hotel profiles possible and to build a basis for continuous improvement of personalization through machine learning. We are proud that we were able to bring initial tests of that new back end live on our web platform during this quarter.

The second group of our customers are advertisers. We want to be a profitable channel for them to reach the target group. Looking at the last 9 months, we have to conclude that our commercialization did suffer. But this also has an upside. Because on the other side that means that our advertisers can on average be significantly more profitable on the lead volume that we provide to them. Looking forward, we will always stay committed to both, deliver great value to our users and to our advertisers.

So let's have a look at the overall numbers.

With EUR 259.4 million, our total revenues stayed relatively stable compared to a super-strong quarter with 69% year-on-year growth, beginning of 2017. We see that that's a success, keeping in mind that we estimate a double-digit percent lower commercialization. Additionally, we have to face major headwinds in terms of euro exchange rates. We estimate that the strengthening of the euro had an approximately 15% negative impact on our revenues in Americas and a 10% negative impact on the revenues in our the Rest of the World markets in Q1.

Against all these trends, and although we have focused on improving our traffic quality, we were still able to grow our qualified referrals by 7% to 189.5 million compared to 177.2 million in the same quarter last year.

Revenue per qualified referral went overall down 9%. The positive effect of our traffic quality initiative was unfortunately overcompensated by lower commercialization and the exchange rate effects.

The trend to a more equal distribution of revenues between the 3 segments continues in the first quarter of the -- sorry, in the first quarter, the Rest of the World segment already contributed 21% of overall revenues.

To provide some more color on what happened mid of 2017, we want to show you the quarter-over-quarter qualified referral growth rate development for the last 3 years.

For 2015 and and 2016, that's the blue and the red line in the chart, you can see that the seasonal development of qualified referrals was pretty comparable throughout the year. For 2017, then you see a very weaker growth rate between Q2 and Q3, where most of the drop in commercialization appeared. On the other side, you can also see that for the last 2 quarters, we are back to the normal seasonal pattern.

Next slide. Looking at the share development of our advertisers. We have seen a lot of volatility in Q4. For the current quarter, the development remains relatively stable and the share of small advertisers remains strong.

So let's take a deeper dive into the development in brand marketing. We have to consider that we already established a very high brand awareness globally. This is not only, but specifically true across our most established European markets. Our new ad creators are now more focused on advancing the users' understanding of our product features in order to further increase usage. In nearly all European markets, we are the second strongest brand in terms of regular usage. We aim to even continuously improve our position in the European markets on this KPI over time.

In newer markets, like the U.S., there is still a lot of headroom to grow. To illustrate this, we want to show you here the development of the unaided brand awareness and top-of-mind share. For unaided brand awareness, the independent panelists asked to actively name a site, where you can research or book a hotel. The trivago brand was recalled in 28% of all answers. This is an extremely good value for any brand. In 15% of all cases, trivago is even named first. This represents a strong increase to the previous year with 22% unaided brand awareness and 11% top-of-mind share in March 2017.

Nearly doubling our numbers over the last 2 years in one of the most established and competitive markets represents a substantial increase of our brand value.

Thanks for listening. I will now hand over to Axel for a more detailed look on the financials.

Thanks, Rolf. Our overall total revenue was broadly stable, as Rolf mentioned, at EUR 259.4 million, down from EUR 267.6 million, facing significant headwinds through the strong euro and the drop in commercialization.

As our key focus for the quarter has been our top line development, we, to a large extent, absorbed the negative effects of the dropping commercialization resulting in a drop in profitability.

Our adjusted EBITDA came down from a positive EUR 19.3 million in the first quarter 2017 to a loss of EUR 21.9 million in Q1 2018. Net income turning from a EUR 5.2 million gain to a $21.8 million loss. As percent of revenue, our adjusted EBITDA margin came down from 7.2% positive to a negative 8.4%, while the net income as a percent of total revenue came down from 1.9% positive to 8.4% negative.

Our return on advertisement spend was reduced from 121% to 108% on a global basis, driven by the lower commercialization.

So if you look at our KPIs on a global level, we managed to grow our qualified referrals by 7%, while at the same time, we improved our traffic quality through our marketing and product initiatives. The RPQR in euros came down from EUR 1.49 to EUR 1.35, which represents a 9% drop. Key drivers of this drop have been the significant currency effect, the significant drop in commercialization, while on the positive side, the traffic quality has improved and partially offset those negative effects.

The ROAS dropped from 121% to 108%, again driven by the lower commercialization.

If you look at Developed Europe, qualified referrals came down from 73.6 million to 70.1 million, while at the same time, traffic quality improved. And the RPQR in euros was slightly down from EUR 1.54 to EUR 1.49, driven by lower commercialization and positively impacted by the improved traffic quality that did not fully compensate the drop in commercialization.

On return on advertisement spend, we saw a change from 139% to 128%, again driven by the lower commercialization.

In Americas, we managed to grow our qualified referrals against a very strong first quarter in 2017 by 8%, up from 55.5 million to 59.9 million. The revenue per qualified referral in euros was down 12%, negatively impacted by the currency and the commercialization and partially compensated through positive traffic quality development. Translated into U.S. dollars, the RPQR was up by 2% from $1.96 to $1.99.

Looking at the return on advertisement spend, we saw a decline from 118% to 104% through the significant drop in commercialization compared to the first quarter in 2017.

Rest of the World experienced healthy growth in qualified referrals of 23% coming up from 48.2 million to 59.5 million. The RPQR in euros decreased by 10%, again driven by strong currency movement and the drop in commercialization and partially offset by the improved traffic quality. Looking at the RPQR translated into U.S. dollars, also in Rest of the World, we saw a positive development up 4% from $1.08 to $1.12.

The ROAS came down from 96% to 87%, again driven by the significant drop in commercialization compared to the first quarter of 2017.

Looking at our overhead costs. We had our total cost and expenses increased from EUR 30.9 million in the first quarter 2017 to EUR 46.3 million in the first quarter of 2018 or a 50% increase. The biggest driver of this increase has been the increase in the overall headcount, which has increased by 338. This happened across the board as we invested into our hotel relations team, our technological expertise and our G&A function, at the same time successfully mitigating our material weakness and becoming SOX compliant within the first year as a public company.

In addition to the increase in headcount, we had significant increases of our investments into TV spot production, which are part of the other selling and marketing expenses. In Q1 2018, our headcount only grew slightly by 11 talents, and we do not expect a significant increase in the course of 2018.

Coming to guidance. Despite the positive revenue development, considering the difficult comps in the first quarter, we expect the overall year to show stable revenues. We expect to continue to experience negative currency effects and lower commercialization in the first half of the year, as we already mentioned on our last earnings call. We believe that since then our largest advertisers may have increased their profitability on our platform even further, a development that would have a direct negative impact on our commercialization. Although this development may or may not change in the course of the year, we have assumed in our guidance that this change will persist. As a consequence, we expect the revenue per qualified referral to decline significantly in the first half of 2018 and to remain slightly negative year-over-year in the second half of 2018.

However, we continue to expect to return to our growth trajectory in the second half of the year. In addition, we believe that ROAS will improve compared to the second half of 2017 in the second half of 2018, and that the adjusted EBITDA in the second half of 2018 will be positive. For the full year, we continue to expect a negative adjusted EBITDA, ranging between EUR 25 million loss and EUR 50 million loss.

With that, we are ready to take your questions now.

(Operator Instructions) We will take our first question from Douglas Anmuth from JPMorgan.

This is [Billy] on for Doug. I just have 2 questions. First one, could you -- you mentioned one of your largest advertisers has changed their profitability target. Is this a 2Q development that's not being shown in your, I guess, 1Q number? And in addition to that, it seems like your RPQR outlook was lower versus what you had at 4Q earnings. Could you explain what changed there? And then -- and secondly, could you give us more color on the change in the leadership structure that you guys announced during the quarter? Why and how things will change from that?

Okay. So on the timing of the change in profitability targets, that is correct. So the first quarter has only to a very limited extent been affected by that. And we expect a significantly stronger effect in the second quarter.

Okay. Regarding the leadership structure, so Peter and Malte are with company for 13 years. They're super loyal to the company. They want to stay on board. But on the other side, they also want to take a step back, and we used that opportunity to simplify the MD structure of trivago and to keep only 3 MDs. But that doesn't mean that the leadership team as a whole changes. We want just to be more flexible in the future to also promote new members into the leadership team and to be able to have more flexibility there. But Peter and Malte will remain with the company. They will just go into other roles and not stay part of the leadership team.

We will now take our next question from Lloyd Walmsley from Deutsche Bank.

This is actually Seth on for Lloyd. I was just wondering if you could talk about what's happening to profitability in your oldest markets, and how this might affect any geographical expansion going forward? And then also a question on short-term rental supply. I realize it's early days, but I was wondering if you could talk about bringing on the short-term rental supply to the platform. And how has the advertiser and customer feedback been for the 350,000 units now on the platform? And also the final thing is just would you be open to additional short-term rental platforms or would that dilute the value of the current advertisers on the platform?

Okay. So on the development of the profitability of individual markets, I mean, we do not comment on individual countries. Broadly speaking, I mean, the drop in commercialization has not been in certain markets, but across the whole platform. So the drop in profitability was widely across most of the markets or all of the markets. To come to your second question in terms of alternative accommodation, I mean, as we said in the past, I mean, we think it is a very big opportunity. But what is also very important is to be disciplined and not make radical changes to the overall mix on our platform. So we are exactly on track and in terms of testing and gradually increasing the visibility of alternative accommodation, and that also answers your third question. At the right point in time, we will, obviously, onboard more and more advertisers. But we think that it is important that, that you don't rush into it, but that you gradually make the changes on the platform that you need to make in terms of marketing, in terms of marketplace algorithms because there is a significant difference in terms of inventory between alternative accommodation and hotel inventory.

Let me just also comment on that. We really see that there is a great chance for us in this segment. But what you have to also keep in mind is that there is, of course, like, difference in the commercialization of alternative accommodations and hotels just because of the structure of the offerings. And that's why you have to be careful. What we did basically during the last year is that we revamped our whole back-end infrastructure to make it possible that we have very personalized search results. These personalized search results give us a chance to show to users who are more likely to book alternative accommodation, to show them more offers and commercialize well on them, while we have less offers to those people who would never book an alternative accommodation. So I think that is something that's very important. So basically the exposure on trivago of alternative accommodations goes hand-in-hand with the personalization approaches that we have done.

We will now take our next question from Naved Khan from SunTrust.

Can you comment on the impact of changes you are making to the marketing attribution model? Any changes that you see in terms of your mix between TV and other channels? And then, in terms of just the traffic quality, I think you talked about how the quality has improved. Are you seeing any increased usage, repeat usage by the consumer base? And also are you seeing higher conversions? And then I had a quick question on GDPR. Are you baking in any kind of impact on your second quarter performance from -- as a result of GDPR rollout?

Okay. So, on the attribution model. We are progressing as planned. So the majority of the -- on the DA channel, we rolled out a new attribution in July last year. So that has been on the new attribution for a long time. And that is part of the effect when we say we improved our traffic quality. We see a significant improved traffic quality in that channel for quite some time already. And at the end, we see the majority of the effect of the new attribution in Q1, but to have it 100% rolled out will take some time. So we are gradually working on that. And also there, you see an improved traffic quality and, obviously, a negative effect on the overall volume compensated by the higher quality. In terms of repeat usage, we have limited visibility because we don't have a high share of logged-in users, like as any search business. And that's why it's difficult to comment on that. And on your last question on GDPR, we don't expect a significant impact on the business, and we are obviously working on being compliant in time.

We will now take our next question from Brian Nowak from Morgan Stanley.

This is Matt Cost on for Brian. I have two. So what are the biggest changes that you'd point to in the business that led to the decrease in the guidance? You mentioned lower commercialization, and positive traffic quality on the positive side. I'm wondering if you can give some more color on that. And then, how do you think about the long-term scalability of the model? What drivers do you think are needed to get to the long-term margins set at the IPO?

So when you look at our current guidance versus the guidance last quarter, the, I mean, obvious impact is, is obviously, the development of the first quarter and also the recent developments of the second quarter. So we see that the first half will come out more negative than we originally thought. And we assume that the more recent change in advertiser profitability targets will persist for the second half. So we also see the growth in the second half slightly lower than where we originally would have seen it. Those are basically -- this is the main update. And in terms of guidance, so as I've said, we continue to be positive on the growth momentum in the second half, but to a slightly lower level than we originally thought through a lower revenue per qualified referral assumption.

Yes. I think also we want to be cautious for the second half of the year. And looking at the long-term margin development, I think that the question is like several questions. The first question is, how much is the current level of commercialization sustainable? Will it go up or go down? So but there it's not about short-term trends and short-term reactions of our advertisers, but it's rather the long-term market structure. And there we see actually some very positive signs. So there are global advertisers going into the market. We see that the different accommodations types will probably melt into each other. So there will be more players in the marketplace, and we introduced alternative accommodation in our marketplace and that will play, hopefully, also a more important role. So I think in the long-term development, I think it's quite hard to see where it will be in 2 or 3 years, but there are some positive signs about that. In addition, I think when you're looking at long-term profitability, I think, that we showed in the past, and I think we also showed it this year in this quarter, we showed that we are able to improve our profitability and improve our mechanisms, improve our marketing, improve our product. The problem is that it's currently really hardly visible, and I can totally understand that, I mean, looking at the total numbers because basically you are looking -- because it's significantly diluted by FX and by this onetime drop, yes? But there is an underlying model where we have been improving. And we have been improving over the last years and also during the current year. And I think that is the question, like how much of this, even if it would stay on this level, how much would be again be able to overcompensate or compensate over the next years, yes.

We will now take our next question from Mark May from Citi.

Okay, hopefully, you can hear me. Sorry about that. Can you speak as you -- question is around the new attribution model. I know you that you're still early in this process. But what impact -- we haven't seen a big impact on qualified referrals yet. Can you speak to kind of the timing and magnitude of when you would expect attribution model really to get rolling and what sort of impact we should expect on QRs?

You know I think what Axel said is right. So we have basically 2 channels majorly affected by that, it's the display channel and the SEM channel. In the display area, it was like last year already where we have seen the full effect of that. And we have seen the majority of the effect on the SEM channel already. So it's always like you're going from like high-traffic keywords to low-traffic keywords. So the majority of the effect is already in the numbers. That you cannot see it in the qualified referral, is because we overcompensated it. So what you would expect is with a change of the attribution model, you would expect revenue per qualified referrals going up and the qualified referrals going down. This effect is happening. So when looking at our numbers, you would always have to really, like, factor that in, right. So the qualified referrals effect is in there, going down, they are not growing as strong as basically our user base or visitor base growth. And on the other side, you having the revenue per qualified going up, you don't see that because of the commercialization dip and because of the FX effect. So that is basically the scheme that you should think again.

We will now take our next question from Heath Terry from Goldman Sachs.

Couple of questions. Rolf, on the last earnings call, you talked about a tendency among hotels or some hotels developing around undercutting OTAs on price or using that as a strategy of pricing below OTAs in the metasearch to generate clicks? Curious, if you're still seeing that or to what degree that has evolved? And then actually as we look at the increase in advertising as a percentage of sales this quarter, is there a level where you feel like that begins to flatten out? And how much of that would you attribute to like-for-like increases in advertising or ad unit costs, whether in search or other performance-based channels?

So I think it's on a quarter-by-quarter level. It's really, really hard to comment on your first question regarding like how hotels do undercut, if it's more or less. I think that is a development that usually, like, takes years rather than quarters. So looking at it from a quarter-on-quarter level, I think there's no significant change. I think the tendency still holds to be true, yes.

So on your second question on the increasing advertising expense as a percent of revenues, by deciding to not adjust the bidding and in a way fully absorb the drop in commercialization, that is something that we had expected. And also because for the same qualified referral, obviously, we generate a lot less revenues. And that is definitely the main effect that you see on the advertisement side. On the increase of costs on the spot or QRs, et cetera, that in the first quarter is not a big driver.

Rolf, just to follow up on that, is there a way that you would look at the HotelDirect opportunity, what you're seeing with these direct booking initiatives from the hotels in terms of an opportunity for trivago, both to diversify revenue or your revenue sources and then also potentially create some opportunities for the company within that metasearch model to use those direct booking prices?

We see this as a big opportunity. It's a really big opportunity, but it's a long-term opportunity. So I think that's the issue here. So it's definitely a big opportunity. And if you're looking at like a time horizon for the next couple of years, there will be -- definitely, there will be a change and we see this change coming, but just the base is still quite low. And so the majority of this effect, it will not happen in the next couple of months, but it will have a significant effect in the long-term view. And that is basically also kicking in when I speak about like a long-term -- where is the long-term commercialization, where is the long-term margin of this business? I think these are all long-term effects that will kick in, but it takes time. I think we have been very early in approaching hotels directly. We are also very consequent now in exploring this opportunity. So also we shifted a lot of resources that we had in our hotel relations team from our product that you know from a pro product, for which we basically get monthly revenue to onboarding hotels directly into our price comparison. So that was also significant change that we did where we again said, no, we want to go for the long-term opportunity. So we are working on it. And we are definitely making progress on it and it's a very significant progress. But the base is still quite low and we are growing continuously. And I think the effect that you will see an impact on the marketplace, there will be an effect this year. But the majority of those effects, they will kick in over the next couple of years.

We will now take our next question from Kevin Kopelman from Cowen and Company.

So I just had some follow ups on RPQR decline, the recent development. Did you say that was -- can you say if that was one major advertiser or both major advertisers? And then, can you quantify the size of the impact on revenue growth from those recent developments, for the immediate impact? So how it's going to impact the second quarter?

Yes. So I think that we are currently in a situation of very limited visibility, also because, you know, at the end of the quarter, there was the Easter break and it was also the end of the quarter. So also advertisers adjusted -- might be for them the reason to adjust their profitability targets. So the visibility on our side is quite low right now. That's why we rather assumed a rather conservative approach for the rest of the year. And I think that is something that we have to do cautiously right now, and I think it's really hard for us to really quantify the effect, but we were rather like saying, okay, we just assume whatever happened, it will stay. And we think that is rather conservative, but that's what -- that's our assumption, that's why we adapted our guidance.

And just to add to that. So underlying our guidance, we assume a significantly weaker revenue development in the second quarter compared to the first quarter, driven by a significantly lower year-over-year RPQR trend, Q2 over Q2 '17 compared to the Q1 that we just reported. And so the quarter... go ahead.

Was it both major advertisers?

No. So I guess if you have a significant impact on the overall marketplace, then that, I guess, answers your question. There is a significant impact on the overall performance.

Okay. And then, on TV commitments, do you feel that like you overcommitted on your TV buys for the year, given the recent RPQR developments?

No. I mean, I guess, we will see at the end of the year. But from what we are currently planning for, we don't see an overcommitment. We -- even in the markets where you need to commit for the full year, we always leave some buffer in there. So I guess, what you are referring to is the position that we were facing end of last year. As of today, we don't expect that Q4 2017 situation to come back in Q4 2018.

Okay, great. And then just one clarification on the guidance slide. For the second half of 2018 on EBITDA, are you expecting that to be positive EBITDA in the second half of the year or are you expecting it to be better year-over-year than the second half of 2017?

Both. Positive. And then by being positive, better than last year.

Okay, so not EBITDA losses, but EBITDA profit in the second half of '18?

That is correct, yes. Greater than 0.

We will now take our next question from James Lee from Mizuho.

My first question is a follow-up on the new attribution model on search, specifically. Can you just explain, maybe conceptually, how you target keywords differently than before? And what have you learned so far? And which region specifically do you expect the most improvement and why? And my second question is about landing page assessment. You guys have done it for 1 year already. Just curious what kind of improvement in conversion are you seeing here? Is it worthwhile for you to continue to make that initiative specifically?

So the first question was on the attribution model. So first of all, looking at the markets, we don't distinguish between markets and so forth. So I think there is no expectation that through the change in the attribution model, there will be a change in the profile of the market. That will definitely not happen, not in overall volume or profitability, that's not what we're expecting. And conceptually what we're doing there is that in the past, you have to assume that while trivago -- we looked at our revenue that we generated. So when we're looking at profitability of campaigns, we looked at the revenue that we were generating to measure profitability of those campaigns. And with us getting more and more visibility on the overall value that is created, e.g. through a booking at the end of the time on the advertiser's side, we are able not to optimize for our revenue, but we are able to optimize for the value that we generate to our advertisers and that gives us potential to optimize. And this is resulting in better traffic quality because traffic quality is designed by value that we generate to our advertisers. That means that we generate usually -- we generate or you would expect to generate less qualified referrals with a higher quality. That's conceptually what we changed there. So in total, we have potential to optimize our marketing and to create more value with the same budget. Basically, that's what we did over the last year.

To come to your second question on the experience so far with the relevance assessment, so overall, we think that it was positive to introduce the relevance assessment end of 2016. And we are continuously optimizing the algorithm. I mean, we went to an automated version end of 2017 and they're really focused on making it better and better. So overall, we think very positive experience. On your question, how much the relevance assessment has improved conversion? I mean, the -- as I said, we think it's overall positive, but it is not possible to quantify the exact effect because it is basically live on the overall platform. And as we said, we had significant improvement in traffic quality, and to isolate the effect coming from the relevance assessment is not possible.

If I could ask a follow-up question on OpEx specifically. Looks like G&A and tech and content is a little bit higher than expected. Is that the kind of new base we should model going forward? Number one. And number two is, when can we start seeing leverage of these 2 items specifically?

Yes. I mean, so in G&A, I would say, broadly speaking, the level that we are having right now is a level that, I mean, that shouldn't significantly change during the year. There's, obviously, always some seasonality in it, in particular in the TV creative expenses. But broadly speaking, the level, I think, is a good level. There are -- in the G&A, there are expenses in there that we would hope over time to get rid off. I mean, I commented on the SOX implementation and generally our first year as a public company and mitigation of all material weakness. And that incurred significant expenses that, I would hope over time, we can internalize and reduce. But broadly speaking, for this year, the level that you are seeing right now is more or less what you would need.

Let me just give some comment on the overall overhead development there. Because I think what you can see there is that -- we showed you one slide where we showed the development of headcount. And basically, the headcount is a major driver for our overhead cost. So when you're looking at our overhead costs, basically the headcount, it's a little bit the creators in -- for our brand marketing, but that are basically the main drivers, all other costs are somehow related to that. What you can see there that during the last year, we still grew our headcount because -- of course, we were going into the year with 70% growth, second quarter, 70% growth, and so on or close to 70% growth. So we were going into these quarters and so basically there we've grown the headcount. But now you can see that from Q4 to Q1, there's also no headcount development anymore. So it's basically -- and perfectly stable. So we were able to bring that down now that we also didn't expect through the -- with commercialization we expect not that strong growth in the first quarter, negative growth in the first half of the year. So we expect this overhead to be -- overhead costs to be relatively stable over the year besides the seasonal effects that we have in there. And I think then -- I think the leverage that you can see then is probably down in the second half of the year when we're going back to growth and you can see the leverage of these costs. And then in 2019, you will automatically see leverage because you will not directly ramp up our overhead costs there. So you will see a leverage in 2019.

I also noticed that you're currently doing a cloud migration. Can you help us understand what the OpEx cost is, maybe, for this year? Is that a headwind that we should model into our financial forecast at this point in time? And when should we expect, I mean, some sort of synergies in the future?

Yes. So we moved already parts of our infrastructure into the cloud and we did that over the years. We definitely know how to do it with our core application. We don't think that there will be a significant cost involved that will really be significant for the overall business. We don't expect that. We -- also we see on one hand, and of course, cloud is more expensive; on the other hand, it's way more efficient. So we don't expect that we see like a significant cost factor coming in there.

(Operator Instructions) We will now take our next question from Tom White from D.A. Davidson.

Just one for me. A follow-up on revenue diversification. Last couple of quarters, you guys have talked about trying to reengage kind of the HotelDirect advertiser and some of the smaller OTAs globally. It didn't look like your other advertiser, as a percentage of revenue, I think it ticked down in 1Q. I guess, just long-term, how should we think about the ability of those other partners to monetize relative to the large global OTAs? So for a given property on your site, how do those 2 different kinds of groups monetize? Obviously, the HotelDirect has kind of more margin to work with, but maybe suffers on kind of the conversion rate side. Just trying to understand how to think about how that plays out.

Yes, so I guess, when you look at the comparison Q1 versus Q4, you need to be a bit careful because the Q4 has very different trends in the first half of Q4 and the second half, and one interim data point where we disclosed the quarter-to-date numbers in October on our earnings call. So that gives you more detail, more granularity on the trends. But generally speaking, there is a -- I would say, there is a lot of room to improve and increase the share of the all others and the competitiveness of all others. And I think we talked about on the last earnings calls as well. I mean, we've got various tools that empower our advertisers to become competitive in the marketplace. We have automated bidding, which mitigates the different size in terms of bidding. We've got Express Booking, which offers an optimized checkout process to, in particular, smaller advertisers, and we are continuously working with them to really improve the platform. Having said that, there are a lot of advertisers, obviously, and it is a lot of work with each one of them. And so it is really something that you will see -- as Rolf said earlier, that you will see over years and not over months.

And the problem is also -- not the problem, it is the adoption rate because basically what sounds like you empower the small advertisers, but we want to empower all the advertisers, right, at the same time. So -- and when we introduce a new tool, when we introduce, like, a new way to use our traffic and so on, so we usually open it up to everyone. And the adoption rate of the large advertisers is higher than of the small ones. The effect for the large advertisers, though, is smaller. So that is always what you have to keep in mind. So adoption rate faster, effect is smaller. And that's why it's always hard to basically see the overall effect because if you're going into the long tail of advertisers, these initiatives have a quite large effect, but the adoption rate is lower. So I think that is how we have to think about it. So -- and then that is also why you cannot like -- you don't want to avoid -- we want to make all the advertisers better, the large ones and the small ones.

We will now take our next question from Peter Stabler from Wells Fargo Securities.

This is Rob on the call for Peter. I think the last quarter, we asked about the strength of the commercialization headwind. I think you said it was high singles, maybe low double digits balanced by a roughly equal tailwind from conversion. Sounds like the headwind from commercialization became stronger. Just wanted to ask if you can give us any sense of scale there. And also the trend in conversion, was it consistent versus last quarter better or worse? And then also wanted to just follow up on the HotelDirect opportunity. Had a couple of questions there. In general, we think about it as being a positive for commercialization in terms of improving bid density, but just wanted to ask is that always the case? It seems like hotels could gain share through the blue placement, winning champions' position with the lower price versus a higher bid. Just wondering if you could talk us through some of the dynamics there as well.

Yes. So first -- I'll take your first question on commercialization. So Q1 versus Q4, we based on -- I mean, just to recap, we have approximately 50% visibility and coverage of the overall conversion of our traffic. So the rest, we do estimate, but from the data that, that we see and that the way we estimate, the overall impact, we think that Q1 drop in commercialization versus Q1 '17 has been greater than Q4 over Q4, which was expected because the relevance assessment and the positive impact last year was particularly in Q1 and Q2 and to a very limited extent in Q4. In Q2, we see a further drop in the year-over-year trend. But then, as I said, in Q3 and Q4, we are lapping the relevance assessment impact in 2017 and have smaller drop in commercialization factored into our guidance.

And so I think as I would repeat the -- or repeat the question, I think it was about, like, what do you see when, like, small at or -- like hotels put in lower rates, isn't that also a threat for commercialization? Actually that's not what we see. So when a hotel like puts in really a lower rate and advertise with that rate, and ramps up on our algorithm, usually, it goes up, which means basically, our commercialization is also better. And what we see there is that users just appreciate that you have like an offer with a cheaper rate and they have a higher conversion because they see, "Oh, there is a hotel offering a cheaper rate than the OTAs." So what we see right now with the data that we have that it's compensating the effect of that we might see on -- in the bidding. So usually, what we see there is rather success stories of that.

As there are no further questions in the queue, that will conclude today's question-and-answer session. I will now turn the call back to your host for any additional or closing remarks.

Thank you. Rolf, any closing remarks?

Yes. So first of all, many thanks for joining the call. So we knew mid of last year already that we would have 4 challenging quarters ahead of us. This was the third of them, the last 9 months, but specifically the last quarter, we had very difficult comps, negative impact of exchange rate and a significant drop in commercialization to compensate. We still were able to grow our user base to improving our product and our marketing continuously. So we know that there will be another difficult quarter, but we also are very confident about the second half of the year. So thanks a lot, everybody, for joining the call and see you next time.

That will conclude today's conference call. Thank you for your participation, ladies and gentlemen. You may now disconnect.