arrive, but let's just go ahead. When we were coming to the building, we were thinking, okay, one year ago, we are on the road. Actually a lot of things have happened in the last year. Given that we are approaching Christmas, we thought it might be a good point in time to look back on 2017, what has actually happened, and to recap and give a first outlook for next year. Don't worry, I won't read the disclaimer to you. Just overview on trivago. I guess most of you know us very well, but for those who don't, last 12 months numbers, slightly north of EUR 1 billion in revenue, 700 million qualified referrals. More than 2 billion visits in the last 12 months.
On the supply side, we exceed now 1.8 million hotels and 400 different booking sites on our platform, and we are active with 55 platforms which cover even more countries. How do we fit into the overall travel universe? The way we are thinking usually, looking for the hotel to stay in, where to stay, and where to book. Undecided, and they need support in the overall search. There you've got generalist search, which is obviously Google and specialized players like us. Underneath the majority of our Airbnb and the hotel chains. Then you've got the ultimate accommodation where you want to stay, the individual properties, that give you then the hotel experience. What is really changing in this industry and what is really going on?
I guess one thing that is very clear and that the whole sector has benefited from is the offline-to-online migration. Latest Phocuswright numbers, I think are at approximately one-third online penetration, and that has obviously gone up. That you have from our perspective, is that the user behavior is becoming more and more heterogeneous. What do I mean by that? What that means is that you are undecided not only where to stay, but also what kind of property you want to stay in, and you have more clear idea what kind of experience you're looking for. Is it a business trip where it is all about the right location? Is it a weekend trip with your significant other where it should be romantic, nice restaurants around, et cetera.
In your search and research process, you then decide whether you want to go for an hotel experience or a boutique hotel experience or for an alternative accommodation. The impact this change has on the overall industry is from our perspective, that it is getting more and more difficult to navigate through these different options. Because you are open to more options at the beginning of your search process, and as a consequence, that the value that you can create by providing search tools is going up. Yeah. Now coming to 2017. That was just a more general intro. What has actually happened in 2017? To answer that, I think you need to look back a bit more.
Starting in 2015, what we've done here is we basically, we have plotted the share of our, for some time, largest advertiser, and used that as a proxy for the commercialization of our platform. When you look at 2014 and 2015, pretty stable around the current share on the platform. Then in 2016, you actually can see a change happening. That advertiser decided to increase very significantly the activity on us as a platform. You can see that the share as a consequence, has gone up a lot. In 2017, what you can't see on here, but what obviously has an additional impact, that advertiser had a low relevance assessment score, which made him pay even more, which if you would've had a good score, you would've resulted in even higher share.
As a proxy for commercialization, you would need to make that mental adjustment. In Q3, and also in Q4, you see basically return to the old levels. Why does that really matter, and why is it very important to understand this year, but also last year's performance? We increased our growth, in the third quarter 2016 very substantially. We are at around 40-ish% in the first half of 2016, and then used the additional commercialization that you can clearly see on this slide to reinvest into the business, and then went up to 70% growth in the fourth quarter. In a way, that growth was partially fueled by the reforms.
In Q1 and in Q2 we did the same, again, we had very, very high growth rates fueled by additional profitability coming through this additional pressure on the marketplace, which then in Q3 and in Q4 reverted and obviously made the growth slowing down. In a way, the Q3 2016 to Q2 2017 were particularly good and when you compare against them, then it is not a like-for-like compare one year where again they had particularly strong commercialization, which was not a like-for-like comparable. I think this is really, really important to understand the numbers and the dynamics in the business, that the yearly comps that we obviously show are not really comparable. Sorry, just to add to that, going forward, we plan with the structure as it is today, obviously.
I don't think that you can hope for a return to the elevated levels. That's really loud. I'd like now to go through the various areas of the business and just point out a couple of things that we believe were very important that happened in 2017. On the marketplace side, we are rolling out and continuing tool of advertisers that are using that tool has gone up by more than 100%. It is really an onboarding advertiser-by-advertiser process, very time-consuming. On the Express Booking side, we've been even more successful and increased the number by 250%, more than 250%. On the marketplace as such, we started to make significant changes end of 2016, the direction of these changes are really to allow more access of advertisers and making access more flexible.
More advertisers, more rates on the platform, and trying to come up with rules how to deal with this enlarged access. On the product side, it is a bit more difficult to look back at 2017 because the key focus in 2017 was really on the back end. Redesigning the back end, making sure that it can actually support a lot more growth, for the growth for the years to come. From the outside, obviously, that is nothing that you can really see, but we believe it was very important and we are now in a much better position than we were at the beginning of the year. Phase is continuously developing. We introduced tabs for the slide out.
We have an improved rating scale there with more sub-ratings in the product, and Boundless Maps we use for television advertisements as well. I think it's a very big improvement and technology not that straightforward, so that you basically reload the map automatically when you're scrolling. For users using maps, it's a big improvement in the overall product. On the branding side. More than 800 TV spots were tested through our internal tool. Not all of them, obviously, aired and produced. That's the idea behind that. On the SEM side, we are now at more than 745 million bids per day, continue to roll that out as well. On the brand awareness, the biggest change is, if you remember, the last time we disclosed these numbers, which was around the IPO, is obviously the U.S.
A very significant improvement in the U.S., 10 percentage points in brand awareness. That gives us comfort that our marketing activities actually pay off. In Europe, obviously, we are very high already, there you don't see big changes anymore. One of the key topics on the marketing side has obviously been the new attribution model. We talked about it quite a lot already. Just to recap, what does the new attribution model do? The new attribution model focuses our performance marketing activities on booking value. Yeah, on value that we deliver to our customers and not on our own revenue, which would be click optimized.
With this model, we can fully optimize for customer value in our own marketing activities, and we've rolled it out, relatively speaking, quickly in Developed Europe in the last weeks of July, where we saw in the rollout, a drop in qualified referrals and a significant improvement in booking conversion. Much higher quality, lower volume. In a way, exactly what we anticipated. On the SEM side, it has been more challenging than we originally anticipated. We originally were planning for a quick rollout, realized that the technical implementation in our bidding tools is taking more time than we thought. What we are looking at there is a soft launch in this quarter, in the next couple of weeks, and then a full rollout in the first quarter.
Having said that, looking back at this year, we've learned that high volatility is obviously not good for our business. We will, on the SEM side, be more gradual in the rollout versus what we've done on the DA side, where we rolled it out very quickly. On the advertiser relations side, we obviously continue to add more inventory and more advertisers, as I said on the first slide. In the last year, we managed to increase the overall inventory by more than 70%. We are now at slightly north of 1.8 million properties on our platform and continue to grow that, which is a lot of hard work, because obviously the incremental additions are becoming smaller and smaller unless you enter a new category. That is obviously one of the bigger changes and bigger achievements, I would say, of this year, alternative accommodation.
What is our approach to alternative accommodation? The first thing is we think that it should be part of the same search. As I said before, if you haven't decided yet where you actually want to stay, then you need one tool that gives you all the options and allows you to make a decision. Do I want to stay in the family suite in a hotel for whatever, EUR 500, or am I going for an apartment for EUR 200? Or if the hotel is only at EUR 200, perhaps I might decide differently. That's the value that you need, or the transparency you need to give to the user, and that's why it has to be in one product. If you take that as a starting assumption, it has to be in one product.
The difference of alternative accommodation is actually not that big to small hotels. Now, if you've got a small bed and breakfast somewhere on a Greek island with three bedrooms, that is not so different to a large villa that is rented out in Mallorca the whole year or so. The challenge is that you have less data available, that you tend to have less content, less description, and just less frequency on the property. The approach that we took here was we initially tested with some alternative offering and mix it into the hotel offerings and really look at the site, how does it convert, do we need to make any adjustments to the way we present it, the way we integrate it in the search results, the marketplace dynamic, our marketing activities? What impact does it have as a first small mix?
The second step that we took was to integrate HomeAway, which we've done now a bit more than a month ago, which gives us obviously a lot more inventory and the opportunity to gradually increase the number of alternative properties on the platform and relation properties on the platform. Does the website still work? What changes do you need to make? What happens to the marketplace, et cetera. When we are done with that test, ultimately we will basically try to, like we did on hotels, we try to integrate every provider of alternative accommodation to the platform. The fourth quarter is almost over. For 2017, we continue to believe what we said on the last earnings call. We confirm our guidance for 2017, growth 36%-39% and positive Adjusted EBITDA. There is no change.
I guess the more interesting part is now how to look at 2018. There's been a lot of volatility in 2017 and partially in 2016. The comps are very difficult. How should you think about 2018? What are the key drivers and what are the trends behind them? We'll not give official guidance for 2018. We'll do that on the Q4 earnings call, but we thought it is very helpful for you to have some discussion around the key drivers and how we think about the first half and the second half for next year. On the outlook. First half 2018 will be very tough. On the revenue side, we said before flat to negative. That continues to be true. It will be very challenging to reach the same revenue number at a lower commercialization.
On return on advertisement spend, in particular the first quarter, but also second quarter, were very profitable, fueled by the additional commercialization that we had beginning of the year. There you will see a negative trend. We don't think that we will hit the same ROAS numbers in the first half of the year. On the OpEx side, we have significantly increased our overall cost base, and that's predominantly obviously people this year. The starting point is obviously much higher than beginning of the year, so no surprise the OpEx will go up overall. On the revenue per qualified referral to be there in the first half of the year compared to the first half in 2017. We now have a more balanced marketplace dynamic, so we have a lower commercialization addition to that.
As a consequence, revenue per qualified referral we expect to be negative comparing first half 2018 versus first half 2017. Overall, we expect the first half of 2018 to yield a negative Adjusted EBITDA, with negative revenue development, negative ROAS development, increase in fixed costs. Those are basically the drivers for that. Second half, we are more optimistic. On the revenue side, we expect in the second half to return to our fundamental growth rates in the various regions. Positive growth there. On the return on advertisement spend, we expect a positive development in the second half as well. There we've got pretty fair comparison that is pretty much like for like. As a consequence, we expect the upwards trending ROAS development to continue that we've seen the last couple of years.
It's likely to lead to a negative overall development because the rest of the world, obviously, is growing faster than Americas and Europe, and has a lower RPQR. On the Adjusted EBITDA, we expect second half to be positive, and show, I guess, a like-for-like comparable results. What does that mean for the overall year? Overall, I think, we will grow, and we think that the consensus is more or less in the right range, path. What does that really mean? In a way, what this year has shown us is that to grow faster, quicker, if anything, is strategically pretty important. If we have to decide, are we investing into Sorry. As a consequence, we'll not be profitable anymore. That is something that we would consider for next year, which I guess is some change to our philosophy before.
That doesn't mean that that has to happen. Okay. We're almost through, and open for Q&A. There are possible challenges for 2018. Testing of bidding strategies on the advertiser side, that has started to happen, in Q3. These tests could obviously lead to a change in bidding strategies, or change in profitability targets. The reaction to our continuous optimization of the marketplace on the advertiser and the user side could potentially be a challenge. The effectiveness of our advertising could be a challenge. The overall use of metasearch, obviously. Regulatory attention to the sector has increased and could pose a challenge to the sector overall and to us. The focus on the lifetime value and on the booking value could actually not be successful.
Finally, the cost to comply with Sarbanes–Oxley , given that we now exceed the revenue threshold, we need to be SOX compliant. That obviously comes at an expense, and adds some implementation risk, and that could be a challenge for next year as well. Good. Now we are open for questions.
A couple of questions. Just quickly on that last point you just made about the SOX compliance, obviously, this is something that everybody has to do, and I'm sure you've been planning for this. Is there any stair step in the G&A that you believe has to occur because of compliance? Is there consulting fees or accounting fees that you didn't have last year, for this year, I mean?
Yeah, absolutely. This year, we started to work on SOX compliance, end of last year, beginning of this year. In our OpEx in the first three quarters, you have significant costs in there already. If you look at the year-on-year comparison 2017 versus 2016, obviously, but the whole being public costs are in the OpEx. Increased lawyer fees, consultants. Obviously, we ramped up our team significantly, both to address the material weakness on the voluntary reporting side, but also to prepare for SOX compliance.
What you expect just for the stairstep from SOX compliance will be in G&A? Incremental dollars?
From now to the future, I don't think there are additional costs.
There's no inherent new. Oh, okay. That's fine then.
It's in there already. That's what I tried to say.
Oh, okay.
We invested this year, obviously, looking on being compliant end of December. We should be fully ramped up.
Okay, too much time on it. Virtually irrelevant for us. Going back to the earlier thing about the growth rates and what happened to your key partner or largest customer and what happened to the revenue share. One, did you see this in Q4 of 2016 and you were going public, did you see this as over-earning? Did you see this as something that this was occurring that you thought wouldn't be sustainable?
Obviously not.
Okay, good. Secondly, as you see this, as you go forward in, what you're bringing up here is this is very specific to your platform on how they have changed their behavior. From one of your competitors, it's clear they've changed their behavior on all meta platforms. Is this not a return to a new level, when you're saying it's a return to a new level, that it's a return to the original level, the 2015 level. This was a period of them overspending. Is this a more of a strategic change on their part than this is implying and this number could actually go lower?
If you look back, so the increase from the current level to the elevated levels, and when at that point in time, we thought, okay, it is the largest player in the industry, and in a way, the share that they had before didn't feel right. I mean, it was a bit too low. We thought it was more, okay, an adjustment to the natural share going up. I think the reason why it went down now, and you should ask them directly, but what we think has happened in terms of learning between the different data points is that when the relevance assessment score of this advertiser improved at such a high share of the business, the impact on us overall was pretty significant. Yeah. As a consequence, we adjusted our own performance marketing activity.
In a way, that's a very rare situation where you can see a change in one channel, and you see a reaction on another channel. In a way, what is the cross-channel spill of your activity? That is something that you normally can't see. Say that what the right bidding level is because there's obviously, let's take an extreme example. If it would be 100%, yeah, to get 100% on the platform, you would have to bid, I'm making it up, let's say EUR 10 per click. At that level, we could basically dominate all other performance marketing channels because we would have so much money. That obviously wouldn't make sense. Yeah. If you go from this extreme example down, obviously, there is a point where that is not the case anymore.
I guess the insight of certain advertisers was that if you're around 50%, you see some of this dynamic that I just exaggerated, so that you have a negative impact on your large performance marketing channels from an increased spend on the smaller channels. In a way, there is something like a right level of spend, yeah, so that you don't miss out on market opportunity and that where you don't fund your own competition on other channels. From our perspective, at the current level, that is the right number or in the right range. Could it be 5% higher or lower? Difficult to say, yeah, without having full transparency on the overall profitability. Yeah, I think you could argue that 50% wasn't the right level anyway.
Final question, I'll pass it on. Sorry, guys. During your IPO, one of your main thesis that you pushed was that you guys had figured out that although you were performance marketing, that playing the SEM game was a losing battle, that you couldn't win just buying Google Search. You expressed how your extreme diligence and kind of mathematical marketing in brand, and how you drove your business through TV. Come around 2Q, 3Q, 2Q, really, you're over-earning because of what's going on at that time. Growth is enormous. You piled a lot of that money back into SEM. Why not put that money into brand like you had proposed is the long-term winning strategy?
We did. I mean, we increased our brand spend as well, but the brand spend is not as responsive. In most markets, depends, there's a difference by market, but in most markets, it takes some time before you can economically buy. You need a couple of months lead time. If you decide to ramp up your spend, performance marketing is obviously most responsive. You can do that tomorrow. The TV spend takes a bit longer. We've done that as well.
Basically, you were so over-earning as your expectations that you had the excess money so quickly, you couldn't put it back into brand fast enough to reinvest fast enough. That's why you increased your percentage of SEM.
That's correct. Yeah. It takes a longer time to move in, and it takes a longer time to move out. That's correct.
Well, for the recording, it's Nat Schindler at Bank of America Merrill Lynch.
Thanks. It's Kevin Kopelman from Cowen. Just had a couple, just kind of focused near term first. On your Q3 call, you gave some specifics about advertiser market share in October. As we look at the slide that shows the share in the fourth quarter, is that similar? Was November similar to October in terms of the way the major advertisers played out on the platform?
You mean the slide five of the earnings release where we gave the October to date, basically quarter to date. This is obviously updated. Try to read the fine print. When exactly that was. That's in the last couple of days. The number hasn't moved that significantly between October and November.
Then one other short-term question. As you reiterated the full year 2017 guide on revenue growth, there is a big range in terms of what that implies for the fourth quarter, all the way anywhere from down a little bit to high teens. Can you give us any more color about where that's falling out so far?
No. It's a slide.
Fair enough. Just a couple of questions on the guidance. Have you gotten any more clarity on how big the relevance assessment surcharge impact was in the first half of 2017, or is it even possible to quantify?
It is very difficult to quantify because it is one dimension. What we said before is that we think that it is at least double digit.
Double digit as a percentage or double digit contribution to growth?
Growth. Yeah.
Okay. The third quarter of 2018, is that a fair comp versus the third quarter of 2017, given-
Yeah
the spend level from the largest advertiser was still elevated.
Yeah. That's a very good point. When you look at the third quarter, it is basically the first two months are more like the second quarter, the third month is then heading down. There is still some unfair comparison effect in the third quarter. The true like-for-like comparison will be in the fourth quarter. Third quarter obviously will already be much better than first and second quarter because you have partially the normalization in there. Plus you have a fair relevance assessment comparison. Yeah.
I'll ask one more and pass it on. Can you talk about your relationships with the large hotel chains and how active they've been on the platform and what you're working on to get them even participating more fully on trivago?
Yeah. We've got good relationships with all significant chains. Pretty much all relevant chains. In a way, they've got a lot of potential on the platform. We discussed that in the past. What is really under our control? What is under our control is that we can give hotels preferred visibility. If there is a live connection, there is more visibility on the hotel rate. Because that's positive for conversion. That's on the product side. On the tool side, it is really what I mentioned before. It is automated bidding and Express Booking, which aims to optimize the bidding strategy and to improve the booking conversion. That is really what we can work on to roll that out to more and more hotels, hotel chains. That's only our part of the equation.
To what extent the hotel chains are really pushing us as a channel, become more aggressive, improve their own booking conversions, et cetera, that is something that we cannot control directly.
Do you see the bigger hurdle as being the hotel chains improving their conversion rates from the platform, or the hotel chains decisions in terms of how much they're willing to spend for a lead or ultimately for a conversion on your platform?
The conversion is the bigger challenge because that is something where you really need to optimize, and that takes time and effort, and resources, obviously. Whereas the decision to change your profitability target, you can implement immediately. I mean, you take a decision and time to implement it. When you look at the market, the competitive landscape for them on the booking technology, they're obviously competing with companies that do only that. They only do online hotel bookings. There it is obviously tough to compete, and to narrow the gap. On the profitability target, it's everybody's own decision.
Can you comment on the profitability of Germany specifically? More importantly, any kind of commentary that you could provide us on what the growth of that market would look like if you were to significantly cut back marketing spend? Maybe a better way of asking is, what would growth look like or how much would you need to spend to have 0% growth in that market? Just trying to understand how strong the brand is in Germany. Any kind of comments you can provide on that.
We can't comment on individual countries. It's difficult to test. I guess moving away from Germany to more in general terms. In the past, we've done it frequently. A couple of years back where we basically went offline in a market for half a year, three months, whatever. Okay, what impact does that actually have? We've done something similar in one market, but only for two months, this year, where we saw that the main impact was that the growth basically came down obviously to no growth. I'm not sure that that is really 100% fair because obviously if you would not do it for two months, but for six months or for 12 months, which is really what you're asking, it is likely that the effect will be greater.
It is difficult to say without really investing significantly in a very large test.
If you were to look at this model and you were to put a similar margin or similar spend ratio to what we see the OTAs doing, which they're spending 30%-40%
Of their dollars in the marketing channel. I'm just trying to get a feel for what the growth of this business model looks like in a more normalized marketing spend or a more steady state model. That's why I asked about Germany, because I know it's much more established, but I'm just trying to get a feel for what sustainable growth of this business looks like. Any kind of comments, or maybe asking 0% spend isn't the right question. Maybe it's more, what does it look like when it's more normalized at a 30% margin spend?
I guess one data point that you can look at obviously is the relative profitability there versus the other markets. When we discussed it during the IPO, we were at a contribution margin of 25% in Europe. Then obviously the question is, which part of the overall overhead is triggered by Europe? Which is under proportional, obviously, because you've got fixed costs by market, rest of the world is a lot more complex per EUR in revenue or per USD in revenue than Europe. That would at least give you, looking at the hard data, a good starting point. Europe, that's why we will continue to invest into that. If you say at maturity of the market, everybody went into a vertical hotel search, then the profitability should obviously be higher than what you see in Developed Europe today.
That's how we came up with our 25% as well in the IPO.
All right. Thanks. As we've seen Booking.com shift from city landing pages to hotel back to city, I'm curious, they're now absorbing the relevance assessment again. Are you seeing unit economics improve a bit from where they were when they were buying hotel specific? Or are they further reducing bids just to compensate?
To that new, or the alt new, whatever. Yeah. To the regional layout. If you look at the share here, and at the share that we presented there, no significant change. That implies obviously, impact is obviously on the overall business, not as big if you are at 50% versus at 25%. But yeah, the relevance assessment score of that advertiser has come down through that change. That is correct.
Okay, I guess as you guys have shifted from a manual to an algorithmic calculation of that relevance assessment, do you see any change from customers in their ability to test it or their confidence in bidding with that algorithmic relevance assessment?
We see a change there. Yeah. Certain Expedia brands are testing the same layout currently. Which is in a way exactly why we came up with the relevance assessment in the beginning so that we could allow for more testing activity. I think it's probably fair to say that there is now more testing activity and more broadly spread testing activity than before. What is a bit too early to say is what really the impact of the move to the automated assessment versus the more manual before is, because we just finished the rollout, and it will obviously take time for everybody to react to it. That we can't say yet.
Okay. Then specific to your SEM spend on Google, as they shift more traffic to Hotel Finder, can you just give us a sense for your early tests bidding on Hotel Finder in, I think, one country. Are you seeing any good performance from that? Are you ultimately paying more per click there versus core AdWords and then the downstream.
Yeah
booking? How does that look on the two different models?
I think it's a bit too early to say. Yeah.
Had to try.
Planning on going direct on TV in many other markets. I think they said 30 by the end of this year. What kind of impact are you baking in when you are talking about the 2018 guidance? Is that already baked in or should we see that as a potential risk?
I'm not sure I fully understood the question.
Booking has talked about Booking.com is planning on expanding the TV ad rollout. They're going to 30 markets by the end of this year, they said. If we look at your 2018 outlook, what kind of impact are you baking into your outlook from that, if any?
Yeah. Okay. Thanks. On TV, the reality is there is always somebody else on TV, in any market, it's not that we have online travel monopoly or had that in the past. I guess we are one of the players that has most focused on it and has been most consistent in the TV investments continuously obviously being present for a very long time period. Now with more players being more consistent as well, or that's at least what they To whom and how you make clear how your product differentiates Because the value proposition of the different players in online travel is different. Our value proposition is ideal for somebody who hates to miss something. This small boutique hotel that you can find on us, but perhaps not on other platforms that you want to consider as well.
If you feel that you're missed out on this special opportunity, you're more inclined to use us than an OTA directly. Same with prices. If you want to make sure that you have full price transparency and that you know exactly what the price structure in the market is, then you're much more likely to use us than going to an OTA directly. If that is not important to you because you travel for the business and don't have to pay yourself or are more focused on convenience, et cetera, then you might be more likely to use somebody else first. In a way, the market is developing and like any market that is developing, the more developed, the more Message. That, I think, is much more important than for the effectiveness of the advertisement than how many players are really active.
Meeting, or is that a potential upside that can materialize?
You're referring to the product side? Are we referring to the attribution rollout?
The attribution rollout ultimately driving booking conversion.
We obviously plan. If there are absolutely breakthrough innovations that we will launch that are beyond the normal improvements, then that could provide some upside. As of today, the outlook that I'm presenting is basically what we are currently thinking with what we are knowing today.
Just as a follow-up to the marketing question, can you actually speak, just holistically speaking, about your marketing mix plan for 2018 to support the financial outlook that you laid out relative to what you did in 2017? I'm particularly thinking about search in particular. Google seems to be the ultimate winner of this whole thing going on. How aggressive will you be on that platform, since obviously the other guys are going to be pretty aggressive as well? I have a follow-up.
Our competitiveness on Google obviously depends on our own profitability target and then our ultimate conversion. There, we are obviously working on further improving that going forward. That should less well, at the same profitability target, that would mean that we are not as competitive as in the first half of 2017.
Does that mean that you're actually spending less in aggregate on Google or you're spending more, but it's just that the unit economics on Google are not as attractive?
We don't comment specifically on spend developments, but what I said was that the ROAS will go down or we expect the ROAS to go down in the first half. If you assume that that is across all the different channels, that would mean that for the same revenue, we would have to spend more. Then depends a bit where we come out, but let's make it simple. If we hit exactly the same revenue and the ROAS has come down in all channels, that would imply that we would have increased our spend.
Okay. Then lastly, can you help us just quantify or envision the alternative accommodation opportunity for you? How long does it take you to build it into something that starts moving the needle?
How long will it take us to build something that moves the needle? On the user side, on the value proposition towards the user, I think it moves the needle more quickly, as does long-tail inventory on the hotel side as well. It's not a high and a higher percentage. That from a value proposition perspective, I think we should be able to deliver that, relatively speaking, shortly for the long-tail and ultra long-tail to be a very significant part of the overall rev. To us, the key thing is do we have a complete offering and do we have, and that's another aspect that is pretty important, not from a user perspective, but on the B2B side, do we have competition in the long tail? More niche destinations, very busy times, trade fairs, et cetera.
There obviously it can become very relevant, that's why we think it is very important.
Hey, Axel. How's it going? Hey, thanks for doing this meeting. Couple questions. Have you guys seen any.
Yep
gating factor? Obviously, one of these advertisers has a far better conversion rate than the others, and so maybe it's just pricing needs to get readjusted.
You can actually see some of that dynamic on the earnings release slide where we have the different brands in there. What you can see there is that one brand has reduced very significantly, and pretty much all the others have gained pretty much pro rata. What that means is that it is not that there is one or two advertisers that just pick up that volume, but that it is spread more evenly. If you go on the site, you can actually see that as well. You've got more diversity on the site than you had six months ago. In certain markets, a lot more diversity, which obviously is, from a user perspective, beneficial.
Back to my conversion rate topic or issue, do you see CPCs get readjusted as the makeup of the bidders changes in the auction?
Absolutely. Let me just go back to the magic slide. A lot of your competitors, very consistently, if you then revert that, it obviously has an impact on the average bid as well on the negative side. That's clearly the case. Particularly, if you have somebody who has been the marginal bidder in such a high percentage, then that player pulling back has an impact on the winning rates. Absolutely.
As you sit today trying to provide guidance at a time when you guys haven't, it doesn't sound like you have done a ton of testing with the attribution model. Is that still a pretty meaningful variable to play out here? It sounds like that with the attribution model, we could have some near-term negative impact on referral volumes, but a long-term benefit to-
Yeah
that may have or-
As I said, in the outlook that we provided, we consider everything that we know today. We have better knowledge on what is happening because we now obviously have the benefit of observing what has happened in the DA channel now for almost half a year. I think that's one point. We understand it better than when we introduced it, to be fair. The second thing is that we'll roll it out in SEM more steadily than we have done in DA. Whatever impact there is, that will be spread out over a longer time period. We don't expect that to have a significant additional negative impact.
Lastly, we haven't talked about since the IPO, are you still moving into the new headquarters next year? If so, are there any costs we should be thinking about?
Yeah
associated with the move?
I hope so. I hope that the construction is ready mid of the year. That's at least what we are planning for. We think we are excited to move in there. We will have everybody together and not spread out over three buildings or now even four in Düsseldorf. The cost for the headquarters are obviously factored in. We have, clearly, at the time of the move, we obviously have rent twice. In the old buildings and in the new one, that's obviously factored in.
Next year, how different should your branded referral growth or referral growth from branded traffic sources be? Should that change much throughout the year? I know you mentioned you reinvested some dollars into brand, but it didn't have the same impact. How should that cadence change throughout 2018?
I'm not sure I've fully understood the question. How different will the growth of the branded traffic be in the first half and the second half? Is that what you're asking?
QR growth from branded traffic sources. Non-SEMs, basically.
Yeah.
Okay. In the first half, obviously, referrals driven by SEM will be down.
Yes
It will be offset by branded. Can you just help me?
I guess when you look at the overall trend and the branded trend, looks more steady, obviously. That's clear. Also on the branded side, you've got tough comps. If you make X% more money than on your profitability targets on trivago, obviously that has an impact on your marginal spending as well. That gives you more firepower to spend. We, in a way, need to optimize against that and compensate for that drop in profitability, through optimization. That optimization will then go into the compensation rather than into the further growth as it normally would do. You've got a negative effect also on the branded traffic growth. You're absolutely right, it will be more steady compared to the overall traffic.
You talked earlier about changing to more of a focus on revenue growth as opposed to profitability, specifically for 2018, given the advertiser pullback. I guess my question is, longer term, how do you guys think about a path to profitability? When should we expect trivago to start to show profitability?
Yeah, as I said, I think our fundamental view is that revenue growth at this stage of the company and the industry is much more value creating than optimizing for more profitability earlier and leaving the revenue growth on the table, giving in a way that market share opportunity to your competition or leaving it to your competition. Looking back, that's why I made that comment, I think we are now taking a more extreme position than we were before. I think they focus more on it. That's our current view, that would imply that you would have to be patient for some time.
First question on the SEM attribution rollout. Can you just maybe provide a little more color on some of the difficulties that you run into that maybe is taking longer than expected? Just want to understand.
That is bidding every day, roughly 750 million times on the 750 million different campaigns. The key challenge is to basically integrate the new data into the bidding algorithm and then linking that to the different SEM platforms. The change in the optimization in a way in our proprietary bidding tool or the technical implementation, that is the key challenge.
Got it. Then just a second question on RPQR. I know in the past you've kind of guided longer term to more kind of flattish. Maybe if you can just walk us through the, I think, three drivers that you've called out in the past in terms of conversion or given growing mobile or country mix and such.
Yeah. Basically, when you look at the key three subcomponents. You've got commercialization, you've got the booking value per booking, and you've got the booking conversion. On the commercialization, we obviously have a negative comp in the Q3 and for some time until we lap it. On the booking value per booking, that is flat to slightly upward sloping, but it's not changing dramatically. On the booking conversion, we managed to increase the booking conversion significantly in the third quarter. Is that something that you will be able to show every single quarter? Probably not. If you assume that the commercialization is, if you take out the noise of the last couple of quarters, looking forward, it's pretty much flat.
The booking value per booking is pretty much flat, and there is some upside in optimizing the side and the traffic acquisition for more booking conversion, then you would come out with a slight positive. The negative that you need to factor in is obviously an increase in multi-device usage, which is deflating the RPQR and inflating the QR version. You're right, it could also be slightly up, slightly down, but that's our expectation by region.
I think you'd said earlier that there was a double-digit impact from the relevance assessment. I just wanted to get a sense for what that means exactly. Are you talking about if you grew 60% in a quarter, that was at least 10% of that 60%, or was it 6% of that 60%?
No, 10. More than 10 out of the 60.
Of your total growth.
Yeah
Not just the advertisers.
Yeah.
Okay. Secondly, you've switched to algorithmic. Can you comment on which direction that's taken the relevance assessment? One would assume up, but who knows? Also what feeds that algorithm?
Yeah. What exactly feeds the algorithm I can't comment on. That's obviously highly sensitive. On the direction of the score, that differs by advertiser, given that we've just completed the rollout, I think there again, it's a bit too early to say. Obviously it's more efficient, the way we've now set it up. We think that it is better for testing and has higher acceptance with the advertisers, because it's quantitative and it is not a qualitative score.
You said that going forward, you think that the current revenue share of your largest advertiser is probably about right, that seems to feed into your guidance for next year. I think you said maybe it could be up 5%, down 5%. I get that there's probably a correct equilibrium level. What makes you fairly confident that this is the right equilibrium level of share?
It's a bit difficult to say, because we have incomplete data. We can only see what is on our platform, we don't have full transparency, obviously, what is happening for each of the advertisers. You need to work with assumptions. It is getting more and more expensive if you drop more and more and at a low level. Whereas if you're at, let's say, at 50% and you are break even, you don't lose any traffic by going to 49%. That is economically the biggest protection. On the strategic point, okay, to what extent do you actually have a negative spillover in other performance marketing channels? If you're one of many, the spillover is obviously not there. Because that doesn't really change our activities. If you are the marginal bidder on the majority of the traffic, obviously that is much more so the case.
I don't know. We don't have all the data to say this is exactly the right level. From our perspective, it seems to be a sustainable level, and even if there is a change at that level, the impact is obviously much, much lower than what we've seen the last couple of quarters.
Hi. Have you noticed any difference in behavior of your second largest advertiser in all of this? You mentioned that they've maybe started experimenting with landing pages. Are you surprised they haven't stepped up advertising spend given maybe a more competitive or uncompetitive
Have we seen changes in behavior from our other large advertisers? I wouldn't say so. Increased testing activity, that was anticipated with the automated relevance assessment. That's part of the idea. Should they have stepped up more? I wouldn't say so either. When you look at the competitive dynamic, the other players have stepped in, pretty evenly, and in a way, that is a very positive thing from our perspective. If it would have only been one or two advertisers that would've picked up the volume, that would be a sign of a weaker marketplace. If it is spread out pretty evenly, that shows that a lot of players are competitive on part of the inventory. I'm not sure that that answers your question. Our relationship with our shareholder is in a way special because we run completely independently, and that has been very important since 2013.
Not only for us, but also for the business. If you are not running completely independently, then that hurts your credibility towards the competitors, but large customers of us and competitors of our shareholders. I think there it is really, there's no special treatment or anything. The advertisers are making their decisions, whether they are under the same shareholder structure or a different one. To us, that doesn't really make a difference. I would assume that it is the same as the case on their side. The first part of your question, was it a mistake to bid up so aggressively? I don't know. It depends on what you're trying to achieve. Was it up to a level that might not have been sustainable, from a profitability perspective, from a user value proposition perspective, and from a healthy marketplace perspective? Perhaps, yeah.
I had a question going back to your comments about growing faster while reducing profitability, maybe. Going back to that, does that mean you have other opportunities that you might not have been chasing and this is something where you might be, let's say, risking some profit to go after? Can you kind of qualify what you mean by chasing after growth while reducing profit?
What I was trying to say was that in the past we said very clearly that for us it has been very, very helpful to have a baseline which was basically zero profit, by self-funding the whole business. For this year, our objective that we communicated in the IPO was, okay, the existing profitability to use that as a new baseline. What I tried to say is that in the current situation, we think that that is worth reconsidering and that it is possible that we would deviate from that historic behavior.
Then this will be across all your geos, like Developed Europe, Americas, and Rest of World?
Can you say that again?
Will this be across all your geos, like Developed Europe, Americas, and Rest of World?
It would be for the overall business, and the growth opportunities obviously differ by region. There is plenty of growth opportunity in Rest of World, whereas in the more developed markets, obviously, there is less growth potential. The comment is basically on the overall business, we will continue to allocate our spend to where we think it has the highest impact.
Okay. Separately, going back to higher level, you said you have about 1.8 million hotels on your website. Are there still opportunities to grab more there? How penetrated are you in that regard?
It's a good question. We try to find a reliable source for the number of hotels on this planet when we did the IPO, and we couldn't find any. I'm sure there is still upside on the hotel side. I mean, that number includes all properties for us. In terms of overall numbers, obviously the opportunity is much greater on the long tail, so on alternative accommodation. There are still hotels, there have to be still hotels that are not on trivago. I couldn't comment on how many exactly.
Thank you.
Can you take us through the evolution of trivago Express Booking, where we are today, where that is for you in the future, any conversion benefit that you can detail for us, and anything that you can ease the friction from a technology perspective on the hotelier side?
As I showed, we are rolling it out. It is not a very significant part of the revenues yet. If you just use the site, you probably have seen. The benefit for the advertiser depends very much, obviously, on his own conversion on his site. I mean, the weaker his site converts, the bigger the benefit is from a de facto outsourced booking funnel. That is basically the key driver. I think in the Q2 earnings release, we had some examples from specific hotel chains, what the impact can be. Overall, we have high demand for the product, and so a lot of advertisers are interested in moving to it for exactly that reason, because they expect that their conversion will go up.
The ease of use part has today a lower impact because there are not that many registered users where it would be automatically pre-filled.
Does everyone have to go through a third-party provider at this point to move into the system?
Mm-
I mean, do I have to have you plugged into my CRS?
Basically submit your properties, and you submit daily bids. For the trivago Express Booking, you need to have that direction, but we also need to have an interface that is going in the other direction, where we submit actually the booking. It is something that the standard. That's basically the technical challenge.
Thanks. Kevin Kopelman from Cowen. Just have a few follow-up questions. The 2018 kind of commentary that you've given, can you talk about, I have a few kind of components here, but what kind of benefits you're looking at from the SEM attribution rollout are assumed in your commentary on revenue and EBITDA? Are you assuming some benefits? I think you're planning a Q1 rollout of the new attribution model.
Yeah.
Are you assuming that you get some uplift there? As you're doing your internal projections, is that?
Yeah. For 2018, we assume the revenue impact to be neutral. In the long term, obviously, we expect a positive revenue impact.
Are you expecting any benefit from the vacation rental initiative in your commentary for 2018?
Yeah, that's a bit more difficult to quantify. I mean, as I said before, it is ultra long tail, so the revenue share coming from those properties can't be very high. I mean, it's like when you look at. I think the main effect is really in having a better user experience and a more complete offering, and that you can't really incorporate into a financial model. I mean, that is something that obviously over time should have a positive impact, but it's nothing where you can say, okay, that is in month six, 3% in revenue uptick. I mean, that's very difficult.
Would you expect vacation rental to make it into your TV advertising in 2018, in terms of featuring your vacation rental selection on television?
It's possible. The question is, what is the right point in time to do that? I would say that we are today still in a testing phase, but it's a larger scale test than what we've been running for the last couple of months. I would say it's more likely to do these kind of things once you completed the testing phase. It's also possible to do it in the testing phase. Yeah.
Well, they're putting a lot of effort into building out their supply and closing the gap in some markets with Priceline.com. Are you assuming any benefit in your marketplace from increased competitiveness between Priceline.com and Expedia as Expedia adds properties?
The more bidders they are per hotel, the better it is for us. That's absolutely correct. If some of our advertisers would be successful in adding a second connection to currently unique property of their competitors That should have a positive impact on them, on us, and a negative impact on whoever had that property unique before.
Is that in your current assumptions for 2018 in what you've told us today?
No. We plan for what we can control and whether that will materialize, how quickly that will materialize, when it will have an impact, that is impossible to forecast.
Another one on 2018. On marketing mix. In the recent past, I think by 2016, 2017, you talked about or I think suggested that performance marketing was probably a little bit over half of the marketing budget. Do you see any major change to your marketing mix between performance and brand advertising next year?
There's no change in philosophy, that's for sure. There are obviously market-by-market movements in one or the other direction, but in general, we continue to believe that the traffic mix, approximately 50/50, is the right mix, and that will guide our actions.
Just a couple more. On vacation rental, you said the initial 100,000 that you were testing, those were not from HomeAway. Were those directly? Maybe I misheard you. When you were first testing vacation rental, you started with 100,000 properties. What provider were those from or?
Today, most of the OTAs have alternative accommodation in their offering as well. We used some of the inventory that was available to us anyhow, and that we previously didn't show for the first testing. As a second step, we integrated HomeAway, which is obviously a very large provider, that allows us to scale very significantly. That is the phase we are currently in.
A follow-up on a previous comment. Did you say that Expedia was also testing regional landing pages? Did I hear that correctly or did I mishear that?
That is correct.
Okay. Last one, 2018, you've got these difficult comps through the first half of the year. Which metrics should we be focused on externally to kind of see progress through the first half of the year, given the difficult comps?
I think you can look at the KPIs, you just need to keep in mind that you are not looking on a like-for-like comparison. Let's assume the revenue would be the same, the ROAS would be the same. That would mean that we actually fully compensated for the lower commercialization through higher efficiency. That would be a big achievement. I think you can look at the metrics that we have, particularly the four quarters, Q3 to Q2, you can see a normalized trend because you then look at like-for-like comparables again. You can look at two-year growth rates, where you have, if you compare with 2015, similar marketplace structure and a better like-for-like comparison.
Thanks.
Just look at your data. The conversion on mobile is weaker than a desktop, that's for sure, as in probably every internet vertical book. Is the user loyal to us across all these platforms and devices? As long as that is the case, he will ultimately book. As long as you book ultimately through trivago, it doesn't really matter whether the conversion on your mobile phone is zero and on desktop is 100%, because it is just the way you use us as a product. That's why we have a consistent product and a consistent auction across all the different devices. I think until the cross-device attribution is really solved, the reality is you will not know how big the difference really is because you can't fully understand the user journey.
What we are doing to address that why we are less concerned about the mobile increase, one thing is when you look at our platform, mobile is growing very quickly, it doesn't have a negative impact on our RPQR. We are now above 60% revenue share coming from mobile versus 50, I think in June 2016. That's a 10 percentage point increase, it didn't hurt the RPQR, which again, makes sense. As long as the users still use us across all devices, it shouldn't hurt. Yeah, I don't expect a significant change to that trend. When we are testing the cross-device attribution, it is very significant.
Yeah, need to be a bit careful that you don't over penalize in a way mobile, because it doesn't convert, because it is important in the user journey and through tests, you can actually see that dependency, even if you can't measure it and attribute it across devices.
Okay, that's helpful. Then just between the mobile web versus the mobile app, are you seeing any kind of differences in either monetization or even user behavior?
Yes. There is obviously a difference between app usage and mobile usage because obviously, it depends a bit by market. There are markets where you use predominantly apps and there are markets where you use very little apps. If you really generalize a bit, you are more likely to download the app if you are a high-frequency, loyal user than not. In a way, if you just compare it's not a like-for-like comparison. App obviously has better usage profiles, but I think that is driven by the fact that you have an adverse selection of the users.
Which one is growing faster? Is it the web or is it the app?
Which one is growing faster?
In terms of usage. You said your mobile traffic is growing faster. Is it the app or is it the web?
That will be the case. Again, you need to be a bit careful because obviously you've got markets where app is dominating mobile usage, so you have a country mix effect in there. If you normalize for that, I'm not sure. Still, app should outgrow mobile because our app share is, relatively speaking, low. We rolled out apps, relatively speaking, late. As a consequence, the growth potential was higher.
In the last slide you showed, you talked very briefly about the regulatory attention that is being brought onto the space. Can you just elaborate on that a little bit, and where do you think the biggest risk lies?
There are various public inquiries in various markets into the overall sector. One is in the U.K., in Germany there is one, there are other markets where there are inquiries. There seems to be more focus of the regulators globally on the online travel market. That obviously is something that is very difficult to foresee. What exactly are the regulators after, and what exactly would then be the consequence of what they are trying to achieve? That is obviously something that you can't plan for.
Can you remind us what portion of your traffic today comes direct? Maybe how correlated is direct traffic to brand advertising within specific regions?
The last data that we disclosed was for 2016, which was approximately 50/50. For 2017, the numbers have not significantly changed. There is obviously a correlation between your TV spend and your growth in branded traffic. Otherwise, you shouldn't do it. There is obviously a correlation. If you spend on a good creative in the right way, that obviously has an impact and that helps your branded traffic development. Any more questions? Everybody exhausted? Good. Thank you very much for your participation and for your time, I wish all of you a great winter break, and see you next year.