Criteo SA (CRTO) 2016 Q2 法說會逐字稿

Good day and welcome to the Criteo second-quarter 2016 earnings conference call. All participants will be in listen-only mode. (Operator Instructions). Please note today's event is being recorded.

I would now like to turn the conference over to Edouard Lassalle, Head of IR. Please go ahead, sir.

Thank you, Rocco. Good morning, everyone, and welcome to Criteo's second-quarter 2016 earnings call. With me today are CEO, Eric Eichmann, and CFO, Benoit Fouilland.

During the course of this call, management will make forward-looking statements. These may include projected financial results or operating metrics, business strategies, anticipated future products and services, anticipated investment in expansion plans, anticipated market demand or opportunities, and other forward-looking statements.

These statements are subject to various risks, uncertainties, and assumptions. Actual results and the timing of certain events may differ materially from the results of timing predicted or implied by such forward-looking statements. We do not undertake any obligation to update any forward-looking statements contained herein except as required by law. In addition, reported results should not be considered as an indication of future performance.

Also we will discuss non-GAAP measures of our performance. Definitions of these metrics and the reconciliations to the most directly comparable GAAP financial measures are provided the earnings press release issued earlier today. Last, unless otherwise stated, all growth comparisons made in the course of this call are against the same period in the prior year.

With this I will now turn the call over to our Chief Executive Officer, Eric Eichmann.

Thank you, Edouard, and good morning, everyone. I am pleased to report another strong quarter of profitable growth. We continue to execute consistently on our plans for growth and operating leverage.

Before diving into the quarterly earnings, let me take a few moments to explain Criteo's vision in a fast-changing advertising landscape. We believe that data-driven people-centric marketing that is held accountable to performance metrics is the way all advertising will be done in the future. The era of not knowing which half of a marketer's advertising spend is working will be a thing of the past.

Three big trends are accelerating this transition. Number one, mobile is driving the digitization of off-line activities, making off-line intent data available and off-line sales trackable. Two, one-to-one marketing at scale based on accountable metrics is growing fast. Three, increasingly, marketers are demanding optimization and coordination of their marketing activities across channels and devices. We are well-positioned to help marketers make this transition, and finally, make all advertising work.

Through our solution, consumers experience more relevant ads across channels and devices and clients get superior performance. Our continued innovation in core technology, mobile, and cross-device are driving higher value for clients.

Now let's turn to Q2. We exceeded revenue ex-TAC and adjusted EBITDA guidance for the 11th consecutive quarter. At constant currency, we grew revenue ex-TAC 35% to $166 million and adjusted EBITDA 61% to $39 million. We continue to make progress toward our long-term operating model, demonstrating the scalability and leverage of our business.

We performed well across all areas. We delivered on our innovation roadmap. We continued to expand our publisher relationships and we added a record number of new clients across all regions.

Technology innovation drives more client sales every quarter. Q2 2015 clients generated 14% more revenue ex-TAC at constant currency in Q2 2016 in line with our expectations. Keep in mind that this metric is based on clients that were live any day during the last 12 months to Q2 whereas advertisers that were live in both Q2 last year and Q2 this year generated 20% more revenue ex-TAC at constant currency. 77% of our business comes from uncapped budgets helping drive same-client growth.

Three areas are worth mentioning in technology and product innovation. Number one, we continue to perform well on mobile which represent over half of our business. Commerce on mobile devices is growing for our clients and represents 40% of their e-commerce transactions. Our full solution positions us well to take advantage of the rapid shift to mobile commerce. Mobile app commerce continues to enjoy strong momentum with advertisers, generating conversion rates of up to 3 times those of the mobile web. Our app business grew almost six fold compared with last year, accelerating from last quarter and now drives a meaningful share of mobile growth.

Two, client adoption of our Universal Match technology remains strong. As commerce becomes more and more fragmented across devices, advertisers increasingly need seamless solutions across consumer touch points. 66% of our advertisers share anonymized CRM data which enables us to grow our cross-device graph. This pooled cross-device graph become stronger as we scale, benefiting from network effects. Matched users represented 47% of revenue ex-TAC.

Three, we continue to innovate across the core platform. We are making good progress in building the infrastructure to leverage our large-scale cross-device graph within the recommendation and prediction layers of the engine. We are also developing new versions of the engine, one to optimize our clients' gross margin and the second one to allow them to bid on a target cost of sales.

We released a new prediction model driving efficiency around post click sales that resulted in meaningful uplift in revenue ex-TAC. And we deployed new dynamic creative features for advanced image management and image cropping to create even more engaging ads for consumers.

Shifting now to publisher relationships, we added over 1,200 publishers, bringing us to over 17,000 direct publisher relationships in our network, providing us with a strong advantage. Around 6,000 advertisers are now live on Facebook dynamic ads on both mobile and desktop with another 500 ready to go live. The dynamic ads product is performing well for our clients and we continue to see further potential to optimize it in partnership with Facebook. In addition, many advertisers are now live on Instagram, a new source of social inventory for us.

In native, we continue to enjoy very positive traction as native inventory is growing quickly among publishers. Revenue ex-TAC from native outside of Facebook grew close to 50% quarter over quarter. Our strength in flexible integration and dynamic creative capabilities positions us well to drive further native growth. We started to roll out our direct-to-publisher native ad solution and expect a larger scale deployment in the coming quarters.

We signed and expanded partnerships with several native platforms including Yahoo! Gemini. We also extended our partnership with Taboola and now buy several ad slots at the same time compared to only one previously.

We further expanded our mobile app reach in particular in app-heavy Asian markets. We launched Twitter's MoPub across APAC and included mobile app in our partnership with Baidu in China and mobile with Kakao in Korea. We launched eight new RTB platforms and are excited about the strong pipeline of new app inventory.

Now moving to client additions. With over 900 net new client additions, we set a new record ending Q2 with close to 12,000 clients and we maintained client retention at 90% for the 20th consecutive quarter.

We signed new large and midmarket clients across all regions. Midmarket contributed to over three quarters of client additions this quarter. Midmarket remains a significant opportunity for us, with the penetration of less than 15%. To accelerate the launch of midmarket campaigns, we continue to roll out a set of automation tools. Our feed module, which allows for automated integration of our clients' product catalog with our servers, went live in Q2. By early 2017, we intend to deploy the complete range of self-service tools from registration and tagging to product feed, creative, and payment.

Turning to regional performance, we delivered consistent execution across all geographies. The Americas grew revenue ex-TAC 38% at constant currency and was the largest contributor to our global growth. Midmarket growth remains healthy across the region at close to 70% year-over-year despite a demanding hiring environment for salespeople. Brazil remains a tough geography for large clients as a result of the difficult political and macro context in that region.

Revenue ex-TAC in EMEA grew 25% at constant currency. Established markets continue to post solid growth in particular Germany. We signed several new large clients including Telecom Italia and a global apparel group based in Spain. Travel clients further expanded their business with us across many markets in EMEA.

Finally, APAC revenue ex-TAC grew 50% at constant currency. Momentum is very solid across markets. New business was strong in particular in Japan and India. We opened our legal entity in India in Q2, and signed several large advertisers. These include India's largest e-commerce marketplace, India's fashion the e-commerce giant Jabong, and MakeMyTrip, India's largest online travel agent.

Growth across Southeast Asia was close to 80%. We also continue to make progress developing our domestic business in China and we hired Yvonne Chang, a high-caliber leader with strong industry expertise, as head of APAC.

In addition, we further strengthened the global leadership team with Tom Aurelio as Head of HR and Elie Kanaan as Head of Marketing, both with a strong technology background.

Looking to the second half of 2016, we remain focused on a clear set of priorities. First, continue to innovate on the core platform, in particular building the infrastructure to leverage our cross-device graph within the Criteo engine. Second, expanded into further sources of inventory especially mobile, social, and native. Third, strengthen our position in APAC in particularly in Southeast Asia, India, and China. And then finally, develop new products especially in paid search.

We will be happy to share more about future growth opportunities and progress on new product initiatives including search at our upcoming investor and analyst day in San Francisco on September 15. We hope to see you there.

In closing, I am pleased with our strong Q2 performance, delivering high growth and improving profitability. We are executing in line with our plans and have exciting new products in the pipeline. As advertisers demand more relevant, accountable, and seamless marketing, we believe we are best positioned to address their needs in the coming years.

With that, let me turn the call over to Benoit Fouilland, our CFO.

Thank you, Eric, and good morning, everyone. I am also pleased with our strong performance in Q2. Fast growth on expanding profitability remains a key attribute of our business model. I will walk you through our quarterly financial performance and our guidance for Q3 and full-year 2016.

Revenue was $407 million, up 36% or 35% at constant currency. Revenue ex-TAC, a key metric we use to monitor our business performance, grew 36% or 35% at constant currency to $166 million. This was driven by the impact of the largest quarterly addition of new clients as well as sustained growth of existing client spend in line with our expectations. Revenue ex-TAC margin was 40.8%, consistent with prior quarters.

We grew midmarket very fast and the share of midmarket clients continues to increase. At constant currency, average revenue ex-TAC per live client increased over 10% and 13% for large clients and midmarket respectively.

Compared with guidance assumptions, change in ForEx had a positive impact of $0.7 million on reported revenue ex-TAC, mostly driven by a stronger Japanese yen. Compared with the prior-year period, ForEx represented a tailwind of 140 basis points to reported growth in revenue ex-TAC, also largely driven by the stronger yen.

Turning to expenses, other cost of revenue comprised of hosting and data costs was $20 million. Non-GAAP other cost of revenue grew 49% to $11 million, mainly driven by increased hosting capacity across data centers.

Operating expenses were $128 million. Non-GAAP operating expenses grew 27% to $116 million. Headcount-related expenses represented over 75% of non-GAAP OpEx. We added 110 net new employees and closed the quarter with over 2080 employees, a 27% increase compared to June last year.

On a non-GAAP basis by function, R&D expenses grew 60% to $27 million, largely driven by the 46% growth in headcount to approximately 470 employees.

Sales & Operation expenses grew 16% to $65 million also largely driven by a 21% increase in headcount to over 1,250 employees. Quota-carrying headcount grew 21% to 570, with 70% of the growth coming from the midmarket. G&A expenses increased 32% to $25 million while headcount grew 28% to 360 employees.

Adjusted EBITDA grew 66% or 61% at constant currency to $39 million. Adjusted EBITDA margin for the quarter was 9.6% of revenue or 23.6% of revenue ex-TAC. As a percentage of revenue, this is a 170 basis point improvement. This was driven by a significant 270 basis point leverage in sales and operations while investing 100 incremental basis points of revenue into R&D for future innovation.

During the first six months of 2016, we improve profitability by 150 basis points of revenue compared with H1 2015. This demonstrates the scalability and leverage in our model and I am pleased that our improved profitability is tracking well in line with our midterm operating model.

Financial income improved by $2.5 million, close to nil. This increase was driven by the conversion of intragroup debt with our Brazilian subsidiary into equity and the removal of cash hedging costs related to that position, in line with the indications provided in our last earning calls.

Going forward, we will not incur any hedging costs in the financial income related to Brazil.

Net income increased 240% to $13 million, driven by 128% growth in income from operations and the significant improvement in financial income over the period. The effective tax rate for the quarter was 25%, based on our estimated annual effective tax rate, which includes the recognition of deferred tax assets in the U.S. Adjusted EPS on a diluted basis increased 106% to $0.33.

Cash flow from operations grew 61% to $19 million, despite the significant negative impact of working capital changes and an increase in income taxes paid over the period. While we maintain DSOs in a healthy level of the mid-50 days, we continue to be prudent with regard to credit risk in certain geographies, given the breadth of our global footprint.

CapEx increased 22% to $22 million. This represented a sequential increase of 85% mainly driven by new data center equipment in line with our investment plan for the year. Free cash flow improved 51% to negative $3 million despite the negative impact of the change in working capital on increased CapEx over the period.

Finally, total cash and cash equivalents were $377 million at the end of June, up $24 million from December 31, 2015.

Last but not least, I will now discuss our guidance. The following forward-looking statements reflect our expectations as of today, August 3, 2016. We expect Q3 2016 revenue ex-TAC to be between $170 million and $174 million. This would imply a growth at constant currency of between 27% and 30%. We do not expect changes in ForEx to materially impact our reported growth in Q3. And we expect Q3 2016 adjusted EBITDA to be between $42 million and $46 million. ForEx assumptions underlying the Q3 2016 guidance are included in the earnings release we published earlier today.

Consistent with the more stable guidance framework for the full-year that we discussed over the last two earnings calls, we reiterate our financial outlook for fiscal year 2016 as provided on February 10, 2016.

We expect revenue ex-TAC for fiscal 2016 to grow between 30% and 34% at constant currency. We anticipate changes in ForEx to have a negative impact of approximately 115 basis points on our reported growth for the full-year. And we expect fiscal 2016 adjusted EBITDA margin as a percentage of revenue to improve between 60 basis points and 100 basis points compared with fiscal 2015.

In closing, I am pleased with our continued strong performance in Q2 combining fast growth with increasing profitability. I am particularly satisfied with our progress on operating leverage while we continue to invest in future innovation. I am excited and confident about the outlook for 2016. I look forward to providing more details of our financial profile during our investor and analyst day in San Francisco on September 15 and hope to see many of you there. Please remember to register with our IR team.

With that, let me now turn the call back to the operator to take your questions.

(Operator Instructions). Doug Anmuth, JPMorgan.

Great, thanks for taking the question. I just wanted to ask a few things, guys. First on Facebook, can you just talk about how you are managing the transition there from desktop to mobile and then in particular, a little bit more color around the recent Instagram DPA ramp? Also the new travel ad there as well?

And then secondly just on sales and ops, it looks like you got a decent amount of leverage there. I think you said sales and marketing hiring is up 21% in terms of headcount. Are you able to hire the way that you would like there or would you like to be able to push on that even faster to drive the topline? Thanks.

Thank you, Doug, for those questions. This is Eric. On Facebook, obviously Facebook has announced that FBX will be no more starting in October. So as we have done to date, we are managing the transition for our clients into DPA and DPA now for those clients is not only a mobile solution, it's also a desktop solution. So when you think about transition from mobile and desktop, it really is just putting all the clients on DPA. As long as the clients are on DPA, I think we are covering both areas.

As you heard from the call, we are at 6,000 clients. For Facebook, we have another 500 ready to be deployed. We continue to transition clients into our DPA solution. We are very excited about the progress we're making there and we are also excited about continuing to work with Facebook on improvements to the solution. The solution is good. The solution is delivering sales as we would expect but we believe there are still areas of improvement.

Dynamic travel ads, it's too early for us to talk about that but that's an area we will be working with Facebook on.

Doug, with respect to sales elaboration, you're right. We got some good leverage this quarter but that is something -- this is a continuous leverage that we have been gaining now for multiple quarters. It is true that the situation with respect to hiring particularly in midmarket and more specifically in the US has become more competitive and we are making a renewed effort there to make sure that we have the right sales capacity in place but this is clearly an area which has become slightly more competitive than what we've seen in the past.

I forgot to mention, Doug, you also asked about Instagram. Instagram obviously now is part of DPA for Facebook and it's an area where we have transitioned quite a bit of clients that are on Facebook onto Instagram. So we're excited about that. Just keep in mind and you can probably see this from the public information from Facebook that Instagram is a property that is about 1/10 of the size of Facebook. So you should expect that type of impact for us as we transition clients into Instagram.

Okay, thank you.

Mark Kelley, Citigroup.

Thanks for taking the question. I think you noted that your clients see about 40% of transactions on mobile. Just curious to see if you can bridge that roughly 10% gap between your mobile revenue and what your clients see. Is that purely the CPC difference between mobile and desktop?

And then second just curious to get your thoughts on header bidding. We heard from a peer last night that who you know noted they underestimated header bidding. Considering you guys continue to add publishing partners, access to inventory doesn't seem to be an issue but anything you can offer there would be helpful, thanks.

Great. Thank you, Mark, for those questions. Yes, as you mentioned there is a bit of a difference between what our partners are seeing on mobile and the revenue that we are generating on mobile. Hard to identify exactly why that is so a strong hypothesis would be the fact that we believe we have a strong position in mobile. We were once developed early on a position in mobile and a solution that sort of cut across all markets and so that probably gives us an advantageous position in mobile and we are probably taking a bigger share from that.

So on header bidding, I think in general we are technology-agnostic as long as we can drive value and get access to all impressions. So and a lot of the demand that we drive comes from direct unique relationships that we have with partners. What is true is what we are seeing now is that -- and this is early impressions -- but what used to be in some cases second price auction dynamics with header bidding are becoming first price auction dynamics. And so that might have an impact in the ecosystem in particular for sort of high-value users.

Now we believe and we've seen changes in the publisher ecosystem over the years and we obviously have a technology and a product team that just is dedicated to thinking through these things that we adjust to changes quite rapidly and quite well. We are a sophisticated buyer and so though it might -- and we haven't really seen it that much but it might have an impact in the short term we believe in the medium and long-term it will be a positive for us.

Having said that, we really are not sort of looking at that as a big factor today.

That's helpful, thanks a lot.

Heath Terry, Goldman Sachs.

Great, thanks. I was just wondering if you could give us a sense of how your vertical exposure performed this quarter. Obviously it seems like it has been a strong quarter for e-commerce sort of across the board, wondering if that was reflected in your numbers as well?

And then if you could just give us an update also on sort of publisher exposure and where -- what percentage of either impressions or revenue Facebook and Google currently account for?

Yes, so for your first question on verticals, we really globally have not seen a change in terms of distribution of revenues across verticals. It continues to be about 60% retail, about 20%, so 65% -- to give you the precise numbers -- 65% retail, 16% travel, 10% classified, and 9% other, very stable. It's quite stable. What we have seen and this is I think more seasonal than not is in Europe travel performed quite well for us. And so that's probably a distinguishing thing that we would note. But other than that, it's quite stable for us.

In terms of publisher exposure, it hasn't changed dramatically --

-- but Facebook is still a single-digit, mid-single-digit figures in terms of share of our revenue ex-TAC. With respect to Google, we have generally not disclosed Google but it's been very stable as well and that remains a very large publisher, the largest publisher for us.

Just in terms of revenue ex-TAC by publisher and we've given you these numbers in the past, about 62% are RTBs and direct publisher is about 33% or 35% so that gives you a sense for the general distribution of publisher revenue ex-TAC.

Got you. And any thoughts on how you see yourselves working with Snapchat as that grows as an advertising platform?

I think so. Obviously any time there's a new advertising platform that is capturing users' attentions, we are always interested in serving our ads on those platforms. What we generally see is the publishers go through a number of phases before they fully make their platform open to all kinds of advertising and so our expectation is we're hoping to work with Snapchat but there's nothing at this time that we would be ready to announce.

Great, thank you.

Brian Nowak, Morgan Stanley.

Thanks for taking my questions. I have two. The first one, just go back to header bidding, you mentioned what used to be second price auction dynamics is now first price. I guess maybe a simple question. What does that mean and what could that do to the overall model? How should we think about the ramifications of that near-term, long-term?

Secondly if we talk about kind of access to impressions, I think you guys disclosed the impressions delivered or purchased on an annual basis. Can you just help us -- how fast have your impressions purchased grown year-to-date and how do you think about that in the guidance for full-year this year? Thanks.

Thank you, Brian. On header bidding, let me tell you a little bit more so with header bidding, we believe that some high-value users may become more pricey as a result of the change in the auction pricing, so there's going to be more people sort of at first if you will in first look environments willing or able to bid for those users. However over time, the scale that we have, the technology and data advantages allows us to have a better view on what those users are worth and so our expectation is that even though there could be a little bit of a blip in the short term in terms of the value of those users going up or the price of those users going up, that over time that would be something that would be to our advantage.

We as I said in the prior question, we haven't seen significant impacts for this but this is something -- from this, but this is something that we are tracking closely and obviously our buyer and technology teams on the publisher side are looking at this and are looking at strategies to take advantage of the new environment.

Brian, to answer your question with respect to the growth of impressions, the growth impression that we have seen during the quarter has been very consistent with prior quarter. We were around 20% growth in our number of compressions that we bought during the quarter at $220 billion of impressions during the quarter. We see that trend continuing on a pretty much consistent pattern over the course of the next quarters, which implies obviously an increase of CPM, which is around low double-digit increase in CPM.

Great, thanks.

Ross Sandler, Deutsche Bank.

Thanks, guys. I had two questions. First, can you talk about how the Criteo Universal Match IDs compare to the identity data that's being used by Facebook or Google for mobile targeting? Are you guys allowed to use the first-party data that many of your advertisers give you when constructing these Match IDs?

Second question, Benoit, any color on the growth rate for the Americas? The quarter-on-quarter growth was a little bit lower than previous years. Was that the Brazil commentary or any other color there would be helpful. Thank you.

Great, Ross. Thank you for the questions. So Universal Match ID, just to give you a quick sense for that, over two-thirds of our clients are participating in it. They provide anonymized CRM IDs. Those CRM IDs in many cases are email addresses that are anonymized and when you think about that information, that information is a pooled information because in many cases those IDs are the same for several advertisers, so if there's a login on one device with one advertiser and on another device with another advertiser, as long as that CRM ID the same, we make a match. And so we have been able to grow this.

We started initially with a very deterministic approach meaning that we would use exclusively CRM IDs or matches that were actual matches from logins. We've expanded that view because we believe that with a very strong deterministic base, you can run machine learning models to use probabilistic matches that can be refined to be of similar quality of what you can get from deterministic. And so we started doing that and so we feel quite good about our ability to grow that.

Now how does that compare to Facebook and Google you ask. I think obviously Facebook has a very, very large deterministic base. We participate in the matches that Facebook makes and the matches that Google makes in the sense that we buy on Facebook environments, there is matching that is provided by Facebook that allows us to bid on users with a history that comes from matched ID.

So we cannot take that match outside of the Facebook environment and that's why we complements that match with our Universal Match product.

Same thing for Google. Google provides matches within the same device and so if you are in an app and then you move to a mobile browser, Google within their environment allows us to see that match. But again, we cannot take that match outside of that environment so we rely on our cross-device graph. So if you think about that, we are basically taking advantage of great infrastructure and device graph that exists within Facebook and Google and then building our own for everything outside.

Okay, so with respect to the Americas, so growth was 38%. The U.S. was growing faster because U.S. growth was above 40%. I think what is worth noting in term of color in the Americas is Brazil has been challenging especially in the larger accounts in Brazil which is a factor of I would say quite unstable macro environment there in Brazil. With respect to the U.S., we have seen good traction in Tier 1, continued traction in Tier 1.

In the midmarket, midmarket has remained very solid. The growth was at 70% in the region in midmarket. But what we've seen is from a sales capacity standpoint as I have highlighted earlier, the hiring environment has been a bit more competitive for midmarket and we have been slightly behind our plan to build the sales capacity and we see that situation continuing in Q3, but we are taking all of the right measures to address it.

Brian Pitz, Jefferies.

Thanks for the questions. Eric, you mentioned strong growth in native. I was hoping for a little color here. Is there a specific publisher that is driving a lot of this growth or is it all driven by Taboola?

And then looking at third-quarter outlook, I'm curious if there is anything worth calling out relative to higher expenses in 3Q that might be impacting the EBITDA outlook? Thanks.

Thank you, Brian. On native, I would say it's a number of things. Taboola is just an example but I wouldn't say the growth is driven by Taboola. We have a number of initiatives there. I mentioned Yahoo! Gemini and we are also sort of integrating with other players. There's quite a bit of initiatives with publishers to get native units. Why? Because they work and they very much look like content even though it is clearly advertising and so they perform quite well and what performs well generally carries higher CPMs. So that's a good thing for publishers.

In addition to that, we are launching also our own native initiative that allows us to integrate units with publishers and so we started deploying that more aggressively with all the publishers in our network which would allow us to buy native inventory from those publishers with units that we help them set up. And so that's an exciting initiative with them.

I would also say that in general, we are also starting to get access to users that are users that might be blocked by ad blockers and so that's also an initiative that though early is something that provides additional inventory for us.

So with that, let me turn it to Benoit.

With respect to expenses, in Q3 I would not call anything really material in terms of unusual items that we built into Q3 expenses. Of course we would incur slightly higher legal fees but not something very material to the overall expense line.

Ralph Schackart, William Blair.

Good morning. Just looking at the full-year guidance relative to sort of the recent beats in Q1 and Q2, I think you grew about 38% constant currency for the first six months of the year and the guidance for the year, the midpoint is around 32%. Anything outside of your general level of conservatism that you would call out for the second half that perhaps you didn't see at the beginning of the year? Thanks.

Clearly as we explained when we started the year, we've changed slightly our guidance approach for the year. Now we guide with more stable indicators over the midterm with a constant currency growth and also we are guiding on the EBITDA margin, adjusted EBITDA margin improvement, which is very much more in line with our midterm operating model.

But we are, we are consistent. Our intent clearly when we articulated that framework at the beginning of the year it would be more stable so you should not expect necessarily that we will make changes during the course of the year and we are operating according to this new model.

Okay. Thanks, Benoit.

Richard Kramer, Arete Research.

Thanks very much, a couple of questions. Eric, you mentioned the number of new sources of mobile inventory and also native and social. Can you be more specific about how you are going to secure further sources of mobile ad inventory and more importantly authenticated mobile users?

And maybe a second question, given the rise and discussion around video advertising as a critical new medium, can you talk a little bit about how you would map Criteo's skill set onto video as that becomes more prominent?

And a quick one for Benoit. It's now been a year since the Paris Analyst Event when you laid out how many companies you were reviewing and since then you purchased almost nothing other than maybe some higher CapEx. So have you given any more thought to either a partial buyback to offset modest dilution or something else to use the $377 million you are just holding on the balance sheet? Thanks.

Thank you, Richard. Great questions. So new sources of mobile inventory, one of the areas that is growing quite fast and presents already a big part of the ecosystem is app inventory in APAC, in particular markets like China that are seeing more than 50% and this is a statistic from a quarter ago and certainly growing more than 50% of the e-commerce in those markets comes for mobile devices. A lot of that also comes from app environments. You might have seen also an article recently in The Wall Street Journal that talked about how app innovation and app consumer use in APAC, in particular in China, is driving -- is really leading the way worldwide.

So in those markets what's happening is obviously that's an important part of the ecosystem. It's becoming the most important part of the ecosystem and what had not developed as quickly was the ecosystem around the supplier base and the ability to reach inventory on a programmatic basis. That is happening and so when we talk about eight RTBs that have been integrated, a lot of them actually come from those geographies. A lot of them are actually part of the mobile app world.

You mentioned also authenticated users and mobile users, so there's a couple of areas. Obviously within the same environment, we can identify the user. It's not a problem and that is part of our solution so if you have a mobile browser user that appears several times, we have the ID for the user. We have a cookie and we can identify easily that user when they come back to that environment.

Same thing for apps. If they are on apps, we have an ID for them and when they come back to an app we can actually use the history of that user to serve ads. Where obviously a key part of the value coming out of a cross-device graph is able to match users across environments. And so if you have history on a mobile browser and people show on an app, if we have a match user, then we can actually serve an ad based on the history that happened in the mobile browser. So that's why it has been a core initiative and continues to be a core initiative for us to build the best or one of the best device graphs in the industry to be able to make that match. And over time our expectation is that will be a core asset of Criteo that will allow us to drive even more value for our advertisers.

So that's on the new sources of inventory. And then from a video advertising perspective, it's an area that we have looked at for a while and we continue to look at. In general video advertising has a couple of areas that will present distinct challenges versus what is today display or native and those challenges are one -- a lot of the inventory is not necessarily programmatic but that is evolving so that is a good trend.

The second one relates to the fact that there's still a lot of demand for that inventory, not as much supply, so generally a source of inventory that's quite expensive. And so having said that, we also buy plenty of expensive inventory in display.

I think the third one is more related to our ability to serve personalized ads on a video platform so there are existing mechanisms for you to serve parts of the video that are more like static images that are personalized, personalizing a full video even though there is technologies now that allow you to do some of that, most of those technologies don't allow you to personalize millions of videos where the video actually is quite different.

So I think that is an environment where it is a little bit different. Having said that, we are looking at this and we have been looking at this and we are probably sort of getting closer to being in a position to try things and be able to see that will drive sales for our clients, which is the ultimate metric that we go for.

Just to cover your question with respect to capital structure, we are very active and we continue to be active at creating a market for acquisition opportunities. In fact we even slightly increased, strengthened our corporate development team and product dedicated team to acquisitions.

That being said, we are very disciplined in our approach and very demanding based on the focus model that we have with respect to our assessment of acquisitions. So the capital structure with cash on the balance sheet is a great flexibility for the Company to seize opportunities of acquisitions. There is no direct plan to use the cash on the balance sheet for purchasing back shares. This flexibility is there to help us grow the Company strategically with discipline.

Now I would like to note even if it was a small acquisition, we did a small acquisition MrdDrive during the quarter, which was pretty small in terms of amount but a very interesting acquisition to build up our strategy with respect to CPG.

I would say, I would add to that, Richard, that obviously we're still in an environment that is rapidly changing so we believe that in a rapidly changing environment opportunities can arise at any moment for us to take action and drive shareholder value. So from that perspective, we don't believe at this point that it would be wise to do share buybacks.

Okay. Thanks, guys.

Murali Sankar, Boenning & Scattergood.

Thank you for taking my question. I was curious if you could go a little deeper into any changes in the competitive environment that you are seeing especially in the context of kind of two things. First, the commentary that one of your peers made on their intent marketing business and how it's essentially struggling. And the other is also with respect to legal action that has been going back and forth between yourselves and a competitor. I was wondering whether that's an indication of whether the competitive environment is becoming more intense than it was before or anything else? Thank you.

Thank you, great question. So I would say on the competitive environment, things have not really shifted dramatically. I would think that where we are seeing more changes is as we've been discussing on this call, there are changes on the supply environment with header bidding and that's something we are looking at closely to adapt for it. On the competitive environment, we haven't seen pure plays. We haven't seen new pure plays and the pure plays that are out there remain similar to what they were before. They are not a big concern for us. Our focus is really to -- and we believe as long as we do this, we will be in a strong position competitively to continue to add value to our clients by focusing on innovation and driving additional sales for our clients.

We do believe and this is to your second question that this is important for us to keep an environment where the metrics that are being used and the way the clients are reading the metrics is a fair environment and so as you know, we launched a legal action against SteelHouse just recently and in our public filings, we alleged that SteelHouse perpetrated a kind of click fraud scheme designed to take credit for online sales attributable to Criteo and other companies and also for traffic that comes into the site. So I think for us it's always been very important that there is transparency around how attribution is done. Obviously it is a client sort of ultimately responsibility but it is very important that the rules of the game are fair. And so in this particular case, we felt that it was important to make sure that that will continue. Obviously it's ongoing litigation so we can't comment further on that but we believe it's very important to have an environment that is fair and transparent in particular for clients that are trying to figure out where the sales come from, what's the right attribution. So there you go.

Tom Champion, Cowen.

Good morning, thank you. Perhaps I missed it but any change in the customer retention rate? Also existing customer spend declined to 14% year-over-year. Why was that expected? I think there was maybe some additional commentary in your prepared remarks. Maybe you could just elaborate on that.

And then last if there's time, you have been in China for a while now. You had the new data center up and running about six months. I think you signed a couple of RTBs last quarter. Was just curious, any commentary on how the business is evolving there? Thank you.

Great. So on customer retention, not really a change. 20 quarters of 90% retention rate so that's very stable.

Absolutely, 90% and that is very stable. And with respect I think to your question on existing client growth, so our existing client growth has been at 14% at constant currency in Q2 and what you have noted is that we have for the first time disclosed two metrics about that particular aspect.

Our -- the metric that we have been disclosing consistently quarter after quarter was a metric which is strictly based on the definition of our clients, and the clients is defined at Criteo as an advertiser at the parent levels that has been live any day over the last 12 months. So in other words, that metric includes a broader view of the population than the population that were just live during a given quarter. So it's the reason why we said it was appropriate to provide the second metric, which gives us probably a better view from a pure -- same client growth standpoint or equivalent to same-store sales which is basically the growth based on live clients in a given quarter. And that indicator was at 20%.

Now it's a decrease. If you take both metrics, they are in decrease compared to the prior quarter. This was expected. This growth of existing clients is driven to a large extent by innovation in our core platform and that innovation is not linear. So we were expecting to see a sales decrease this quarter but we've got as we discussed on this call, a portfolio of multiple initiatives to invest in our core platform that are going to drive growth in existing clients in future quarters.

Very good and then regarding your question, Thomas, on China, we are seeing good progress. I think first off, we consider the opportunity in China to be -- continue to be very large and we don't see any structural issues today in China that would not allow us to make progress in China. We are making good progress. Now China is our third largest country in APAC. So that's a positive for us. We have a team of about 30 people on the ground. We are connected to the 10 largest RTBs which include Baidu, Tencent, and Alibaba, and our Shanghai datacenter is up and running and fully operational.

So we have the elements in the market. However it's a new market. There's quite a bit of education that needs to be done in that market about our offering. It is a complex market and China is never simple. However, we always thought that the price if you will or the potential of the market sort of justified a long-term investment for us and for the time being we're quite happy with the progress we're making there.

Thanks very much.

Thank you very much. I think we will close the call. Thank you very much for joining today's earnings call. We will be happy to follow up with any questions. We remind you that we are hosting an Analyst Day and Investor Day on September 15 in San Francisco, so we will be happy to send you an invitation. Thank you, everyone. Have a great day.

Thank you very much.

Thank you, gentlemen. The conference has now concluded and we thank you all for attending today's presentation. You may now disconnect your lines and have a wonderful day.