Magnite, Inc. (MGNI) Q2 2017 Earnings Report, Transcript and Summary

Good afternoon, and welcome to The Rubicon Project Second Quarter 2017 Earnings Conference Call. All participants will be in listen-only mode. [Operator Instructions] After today's presentation, there will be an opportunity to ask questions. [Operator Instructions] Please note that today's is being recorded. I would now like to turn the conference over to Mr. Nick Kormeluk. Please go ahead.

Thank you, Andrea, and good afternoon, everyone. Welcome to Rubicon Project's Second Quarter 2017 Earnings Conference Call. For those of you I've not yet met or spoken to, I'm the new VP of IR at Rubicon Project, and look forward to speaking to you in the coming days and months. As a reminder, this conference call is being recorded. Joining me on the call today are Michael Barrett, President and CEO; and David Day, our CFO. I would like to point out that we have posted slides to accompany today's presentation on our website. Before we get started, I would like to remind you that our prepared remarks and answers to questions will include information that might be considered to be forward-looking statements, including but not limited to, statements related to our anticipated financial performance, operating and strategic goals and plans, expectations regarding new initiatives including our acquisition of nToggle, our relationships and business with buyers and sellers using our platform, competitive differentiation, pricing and take rates, capital investment, organizational development, competitive position, market condition, trends and growth expectations, including growth in header bidding, orders, mobile and video. Forward-looking statements involve risks and uncertainties and assumptions and actual results may differ materially from the results suggested by forward-looking statements for various reasons, including, without limitation, if such risks or uncertainties materialize or assumptions prove to be inaccurate. Further, we may adjust our plans and expectations in response to market conditions or other factors. Reported results should not be considered an indication of future performance. A discussion of these and other risks and uncertainties and assumptions is set forth in the company's periodic reports filed with the SEC, including our 2016 annual report on Form 10-K and quarterly reports on Form 10-Q for the first and second quarters of 2017 under the headings Risk Factors and Management Discussion and Analysis of Financial Conditions and Results of Operations. We undertake no obligation to update forward-looking statements or relevant risks. Our commentary today will include non-GAAP financial measures. Reconciliations between GAAP and non-GAAP metrics for our reported results can be found in our earnings press release and the financial highlights deck, which we have posted at the Investor Relations website at investor.rubiconproject.com. At times, in response to your questions, we may offer incremental metrics to provide greater insights into the dynamics of our business. Please be advised that this additional detail may be one-time in nature, and we may or may not provide an update in the future on these metrics. I encourage you to visit our Investor Relations website to access our press release, financial highlights deck, periodic SEC reports and webcast replay of today's call to learn more about Rubicon Project. I will now turn the call over to Michael. Please go ahead.

Thank you, Nick, and good afternoon, everyone. Welcome to our Q2 2017 earnings call. Last quarter, I noted that I would share more details regarding the strategic plan that we are executing against at Rubicon Project, and so I will spend most of my comments today focusing on that. 5 months on the job, it's become clear to me that header bidding has fundamentally reshaped the role of exchanges and it has had a powerful impact on Rubicon Project. To succeed in header bidding, you need to source tremendous volumes of inventory, manage a highly efficient platform and excel at win rates. We are pleased with our progress in volume and pricing and are optimistic about improving win rates, especially with the addition of Boston-based nToggle, whom we're thrilled to welcome to the Rubicon Project family. Let me give you further details on our performance in these 3 critical areas. Rubicon Project is moving from a modest volume, higher margin business to a high volume, lower margin business. Today, we are exceeding historic inventory levels due to header bidding. Rubicon Project now has access to 50% more inventory than we did a year ago. Our objective is to have all inventory that can be bought or sold programmatically available on Rubicon Project's global add exchange. We are already working to source all formats and channels of inventory. This might sound elementary, but to go out and inquire inventory across all formats and channels and incorporate that into a marketplace is not easy and has become less of a focus for our key competitors. We have seen solid inventory growth across mobile and video, especially related to mobile app. In Q2, our mobile app inventory nearly doubled since last quarter and increased by more than 90% since Q2 2016. Our video inventory for mobile apps grew by more than 40% year-over-year with video ad spend in mobile inventory growing 90% year-over-year. We're also excited about net new inventory opportunities, including our integration with Amazon Publisher Services' Transparent Ad Marketplace, we refer to as TAM, as well as our onboarding of British Telecom and AccuWeather. BT is one of the world's leading providers of communication services and solutions, and they are using Rubicon Project's global ad exchange for the sale of their mobile in-app, mobile web, video and desktop inventory in open and private marketplace environments. AccuWeather is the global leader in weather information and digital media with more than 1.5 billion users worldwide, and they are using our exchange for the sale of their mobile in-app inventory. We also continue to make strides in growing our emerging media business, both in the U.S. and globally, especially for digital out-of-home. In Q2, we signed TouchTunes which delivers music and advertisers messages across more than 63,000 venues in the U.S. and in all 210 DMAs, with more than 38 million unique listeners each month. And Adshel, the leading digital out-of-home media company that reaches more than 92% of commuter audiences in Australia and New Zealand. Let me give you an idea of the potential that digital out-of-home offers. According to eMarketer, nearly $4.5 billion is expected to be spent annually on digital out-of-home advertising in the U.S. by 2019, an increase of approximately $1.2 billion from 2016. And PricewaterhouseCoopers predicts that digital out-of-home will grow at a rate of 15% a year for the next 4 years. Rubicon Project has a strong foothold in this emerging channel with key strategic partnerships with Clear Channel, TouchTunes, DOmedia, Adshel and others, and we expect to on board more digital out-of-home partners, especially in the international markets. Next, we made a concerted effort in Q2 to build a more efficient marketplace by significantly lowering our total take rates in order to be more competitive and enhance the dynamics of the marketplaces that we power for our clients. David will discuss the take rate in more detail, but as you can see from our earnings release, our total take rate for Q2 2017 was 21% and we exited the quarter slightly lower than 19%. And today, we are already beginning to see the positive effects of those pricing decreases and better win rates in header bidding auctions. Buyers have told us they want to integrate with fewer exchanges, not more, and that they need help managing their total cost of acquiring inventory. By lowering total take rates, introducing greater transparency into auction mechanics and improving win rates. We are working to become the lowest total cost per transaction provider for buyers and sellers of programmatic advertising. To be clear, being the lowest total cost per transaction provider encompasses all of those factors, not just take rate, and we will use all of these to become the most efficient marketplace. Lastly, let's discuss our efforts to improve win rates. Win rate drives our ability to monetize inventory available to us. There are 3 main factors impacting our win rates. First, the continued movement of traffic from the legacy waterfall in Smart Tag technology to header bidding. Second, the need for continued optimization of our machine learning algorithms for header bidding implementations. And third, the inability for many of our buyers to see all of the inventory and effectively respond. We continue to make progress gaining access to inventory as it shifts from the legacy waterfall in Smart Tag technology to header bidding as I noted. This shift will naturally drive down win rate due to increased competition. However, we believe that will be offset by greater inventory and ongoing improvements to our platform. We are rolling out new algorithms and mechanics that are optimized for both client side and server-to-server header bidding solutions. And thanks to our acquisition of nToggle, we are well underway with our strategy of helping buyers cope with the influx of bid request and inventory being made available. Since the introduction of header bidding, the number of bid requests received by demand-side platforms, referred to as DSPs, has increased by as much as 5 times. The infrastructure costs associated with that growth have put a huge burden on DSPs, resulting in many of them arbitrarily throttling the number of requests they see. Meaning, that they miss out on accessing inventory that may be valuable to them. nToggle's traffic shaping technology makes it easier for DSPs to more effectively identify and target their key audiences, while also significantly reducing their infrastructure cost by compressing inbound queries per second, often referred to as QPS, by as much as 80%. This technology uses proprietary machine learning-based software, paired with DSP self-service tools to more effectively target the traffic and impressions that are most valuable to a DSP: reduce the number of duplicate bid request and prevent irrelevant bid request from chewing up DSP resources. The most meaningful benefits to Rubicon Project are expected to be a greater share of DSP wallet or market share, increased win rates and greater CapEx efficiency. nToggle's traffic shaping technology has the potential to directly impact win rates in a meaningful way upon the completion of its integration with Rubicon Project's tech platform. To illustrate the leverage we can gain from nToggle's technology, net revenue should improve on approximately a 1 to 1 basis with every percentage point improvement in win rate. Thus, if nToggle's technology were to drive an improvement in win rate of 10%, it would result in about $17 million in additional annual revenue based on Q2's revenue run rate, all other factors being equal. To clarify, this is not what we are modeling for in our financial results, but just an example to demonstrate how an increase in win rate transfers to revenue benefit. As you can see, this type of impact, plus share gains in CapEx efficiency, explain why we are extremely excited by the potential that the nToggle technology offers in combination with our platform. Tom Kershaw, our Chief Technology Officer, and the technology team at Rubicon Project are working diligently to fully integrate the traffic-shaping technology into Rubicon Project's platform with full deployment to our DSP clients expected within the next 6 months. Initial feedback from our DSP clients has been overwhelmingly positive as this acquisition illustrates our continued commitment to investing in our buy side technology and helping to meet the ever-changing needs of buyers in an accelerated fashion. We also believe the payback period for nToggle will be within 3 years. It's clear that the world of programmatic advertising and what it takes to win in this new world has changed dramatically with the introduction of header bidding. All in all, we have made great progress in Q2. We are on our way to becoming the exchange where all inventory that can be bought or sold programmatically can be found. We are working toward becoming the lowest total cost per transaction provider for buyers and sellers. And we are continuing to improve win rate with the addition of nToggle's technology to our platform. Although our financial results will lag these efforts, we believe that our strategic initiatives have us on track to drive year-over-year growth by the end of 2018 and deliver long-term shareholder value. With that, I will hand the call over to David to give an update on our financials. David?

Thanks, Michael. As Michael mentioned, we made very meaningful strides to improve our competitive position, both strategically through acquiring nToggle and with our aggressive pricing investments as part of our plan to grow market share. We have made the majority of price investments we plan to make at this time and exited the quarter at a take rate just below 19%. The reduction in take rate represents an almost 30% decrease from peak rates to the current level. We also had solid growth in some select areas, notably mobile app and video. For the second quarter of 2017 we generated $204.4 million in advertising spend; $42.9 million in GAAP revenue, which is the same as non-GAAP net revenue in Q2 because of our exit from intent marketing in Q1; $3 million of adjusted EBITDA; and a loss of $0.10 per share in non-GAAP EPS. Total advertising spend increased 7% in Q2 versus Q1 2017 and declined 21% year-over-year, driven primarily by desktop ad spend, which was lower than prior year by 31%. Total mobile ad spend increased 26% versus Q1 2017 and increased 1% year-over-year, driven by mobile app. Mobile app ad spend, which is an area of focus for us, grew nicely versus prior year. Ad spend was composed of 42% mobile and 58% desktop for the second quarter, representing a significant shift from Q1 of 2017 when mobile represented 35% of total ad spend, and is reflective of our continued shift to mobile and specifically, mobile app. This also represents a significant increase over Q2 of 2016 when mobile made up 33% of total ad spend. Non-GAAP net revenue for the second quarter declined 5% sequentially from Q1 2017, despite higher ad spend and 34% year-over-year, which was greater than ad spend decline due to much lower take rates. As a reminder, because of our exit from intent marketing in Q1, we now report all revenue on a net basis. Our take rate was 21% in the second quarter of 2017, a decrease of 430 basis points from the second quarter of 2016 and a decrease of 270 basis points sequentially. Take rate is defined as non-GAAP net revenue divided by total ad spend. As stated earlier, the year-over-year decline in take rate was primarily due to our aggressive investment in price reductions to improve our value proposition. In addition, mix shift also had an impact on take rates. For example, the shift from RTB to orders, with the latter carrying a lower take rate. Also, the decline over the previous year in intent marketing culminating in our exit from that service in the first quarter, contributed to the lower take rate as that solution carried a higher take rate than any of our other revenue streams. Operating expenses for the second quarter of 2017 were $54 million, down from $69 million in the same period a year ago. On an adjusted EBITDA basis, operating expenses for the second quarter were $39.9 million, down from $46.7 million in the same period a year ago, reflecting the impact of our cost cutting initiatives from the last several quarters. Net loss was $11.6 million in the second quarter of 2017 compared to net loss of $2.7 million in the second quarter of 2016. The change in net loss year-over-year was due primarily to lower GAAP revenue, partially offset by reduced operating expenses and a lower tax provision. Adjusted EBITDA was $3 million in the second quarter of 2017 as compared to $18.4 million reported in the same period 1 year ago. The decrease in adjusted EBITDA was driven primarily by a decrease in non-GAAP net revenue, partially offset by lower cost, as previously discussed. Diluted GAAP loss per share was $0.24 for the second quarter of 2017 compared to diluted GAAP loss per share of $0.06 in the same period in 2016. Non-GAAP loss per share in the second quarter of 2017 was $0.10 compared to net income per share of $0.28 reported for the same period in 2016. Capital expenditures, including purchases of property and equipment as well as capitalized internally used software development costs, were $3.8 million for the second quarter of 2017. We closed the second quarter of 2017 with $192.6 million in cash and marketable securities, an increase of $4.6 million from the quarter ended March 31, 2017. As a reminder, we used $38.5 million in cash just after the quarter-end for the purchase of nToggle. Free cash flow for the second quarter of 2017 was $6.1 million as compared to a negative $2.5 million during the first quarter of 2017 and $17.4 million during the second quarter of 2016. We calculate free cash flow as net cash provided by operating activities, less capital expenditures, including capitalized software development costs. As we discussed last quarter, we are not providing formal guidance at this time. We are continuing to provide our general business trends for the remainder of 2017. Ad spend, we believe, has begun to stabilize at levels consistent with the past 2 quarters. We have made these macro pricing changes that we plan to make at this time but, of course, as always, we'll monitor the impact and adjust as necessary. We believe take rates will be flattish to slightly down in Q3 relative to the exit take rate in Q2 of less than 19%. We expect adjusted EBITDA operating expenses to grow on a sequential basis for the remainder of 2017 as we integrate nToggle and continue to reinvest in engineering and product initiatives, which will offset some of the savings from our prior restructuring activities. We expect CapEx spend levels in 2017 to increase $4 million to $5 million over 2016, driven by our nToggle acquisition and rollout. As we look forward, we expect to continue to see growth from mobile, specifically mobile app, video and see DSP market share gain resulting from our purchase and integration of nToggle. Our balance sheet remain strong, even following the use of cash for a very strategic acquisition and our significant investment in pricing. We will continue to prudently monitor our cost structure, while keeping our focus on the important investments and initiatives that we believe will help us return to growth in the future. I'll now turn the call over to Michael for some closing remarks.

Thank you, David. Before I wrap up, I would like to note that there have been a lot of significant changes at Rubicon Project since the beginning of 2017, and I am incredibly proud of how the team has and continues to execute. We have made some difficult decisions that have significant impacts on the business in the short term because we believe that these changes will be critical to Rubicon Project's success in the long term, including pricing and transparency, strategic partnerships in server-side header bidding, onboarding net new clients and putting our balance sheet to work with strategic acquisitions like nToggle. Our 23 new team members up in Boston are already proving to be great additions and we are already seeing some positive effects from our efforts in pricing, server-side header bidding and investing in growth drivers, including mobile, video, orders and emerging media channels. We have more hard work ahead of us, but we have the right team and focus to get back to growth in 2018. Thank you for your time. And with that, let's open the line for Q&A.

And our first question comes from Kerry Rice of Needham.

Two questions, if I can. Can you talk a little bit more about what drove the growth in mobile or as a percentage of spend it was 42%. Was that related to just having more mobile app impressions? Mobile web impression? Being able to sell those more, was there some - was there any new partner for publishers? Anything else you can provide there. And then maybe a little more context around the progress on the server-side header bidding solution. I know last quarter you talked about integrating the open source header bidding solution, the pre-bid technology. Are you - all the way, is that done? Are we halfway there, any other update on that? And then, I guess, maybe a little bit of context as to how does nToggle complement that. Does it sit on top of that since it's an open source? Or how do you integrate both of those together? Thanks.

Great. Kerry, its David. I'll start with the mobile growth. Yes, you're exactly right. We've had a very significant increase in available inventory. And in particular, our mobile app has grown significantly. Spotify is a partner and that's been a very significant driver and we got a handful of other specific clients that we've had increase in activity with. Some of our other larger partners include InMobi, Interactive and a few others, but definitely seeing the results of greater access to inventory.

Kerry, this is Michael Barrett. So on where do we stand on server-side header? It's still a Greenfield for us. Most of this success that we've had in header to date has come mostly from client side. We have our, as you know, we have our proprietary solution out there. This quarter, we began working aggressively with Amazon's offering, TAM. And we will be launching our server side solution in the coming month, and probably announcing further partnerships on the server or header sides. So we still very much believe in the efficacy of server-side header and also in the huge opportunity for us to play a leading role as opposed to fast follower as we did in the client side.

All right. Thank you.

Our next question comes from Jason Kreyer of Craig-Hallum. Please go ahead.

Hey. Good afternoon, guys. Thanks for taking my question. Just wanted to touch on client-side header bidding for a moment. So you said you had a 50% increase in inventory year-over-year. Are you back to where you need to be in terms of working with your customers, is there still more room to grow? I don't know if you can maybe put an inning on it or how do we think about much progress you've made and how much further you can go?

Sure. Again, it's Michael. I'll handle that and David can chime in. I think we're ecstatic where we are. If you recall, even when I first came on board. It was still a situation where we were playing catch-up. We said, hey, a lot of these key relationships that we had, we've been uncoupled from. And now it's time to couple them back, get the inventory flowing through the pipes and then start to improve win rate. And I think we're very pleased with the progress that we made on onboarding inventory on to the platform. As I said in the prepared remarks, we're up 50% year-over-year in terms of the inventory in the platform. As we move towards this lower-margin, high-volume business, we will aggressively pursue inventory. We see it coming in the form of server-to-server to the clients that haven't done - to our customers that haven't been client-side header. And of course, header started off as this phenomenon in North America, it's now spreading globally. So the addressable market of inventory is just enormous. And we are very excited about the position we have ourselves in today. We think that more inventory is going to flow throughout the coming quarters. And our focus, now, is as much on optimizing net inventory and increasing win rates as it is chasing more inventory.

Okay. And then just two more.

I was just going to add, Michael is exactly right. We've hit the critical mass that we think we want, of course, we're continuing to sign up new partners and we're happy with our win rates. But we do believe there is even more upside as our algorithms, our machine learning, there is great momentum and activity there, but there's even more upside there as we continue to optimize how we play in that space. And with our recent price moves, that will also help us increase our win rates.

Two more, actually, on take rates. So just wanted to understand the economics a little bit better. Obviously, you lowered the take rates at some point. The goal is to have higher impressions that offset the lower take rates. And I'm guessing that we're not there yet, but perhaps you can give a little bit of a timeline on when you think we can kind of bottom out and rebound there. And then just, second, thinking about take rates in terms of orders. Is that the same strategy as RTB where we're kind of going after an efficiency play in lowering those take rates and kind of more of a price competition there to really get more impressions and more volume? Thanks.

Sure. From a timing perspective, we're still making a lot of investments. Certainly, in 2018, at some point, we think we'll see the fruits of these activities. I think there is still a lot more to learn. And in particular, we've just barely acquired nToggle and we believe that, that acquisition will have a significant impact, but there's an integration and a whole lot of activity to go along with that. So that's from a timing perspective.

And there's no question that when we said before, trying to become and build the total - the low-cost total transaction marketplace, right. And so take rate is an element of that, but we talked about auction mechanics, transparent auction mechanics, incredibly important, as well as what nToggle is going to bring for us and that's matching. So if we are able to have very competitive rates in the market, have an auction mechanic that's transparent and buyers lean into and then come up with proprietary software that allows for greater matching of prospects with ad dollars, we feel very strongly that we're going to still share of wallet from DSP spend from competitors. So definitely, our pricing strategy has an eye towards that. Obviously, all those take a while to fall in line and we have yet to see the full benefit of it, but we're absolutely convinced in the strategy.

Okay. Thanks, guys.

Our next question comes from Matthew Thornton of SunTrust. Please go ahead.

Hey. Good afternoon, guys. Thanks for taking the question here. A couple, if I could. Maybe starting with the take rate, again. You guys talked about exiting the quarter at slightly less than 19%. Michael, I think you said previously, you thought that the steady state was somewhere in the 15% to 20% ballpark. Just wondering if there's any update to that number? And secondarily, around nToggle. Obviously, there should be some impact to infrastructure cost as it relates to efficiencies there. Should we see that manifest in gross margin and is that more likely in 2018 as opposed to later this year? And then maybe just a third question here, probably for Michael. Can you just maybe give us a state of the industry as it relates to header bidding, client-side wrapper versus server-to-server. Where is the momentum and how do you see this kind of playing out over the next 3 years or so?

Sure. So the first question was take rate. And we have said before, David and I - that we believe that the industry will settle in a 15% to 20% range, and we're certainly in that range right now. We said in the prepared remarks we see stabilization right now. It's really difficult to appear ahead over a horizon and say, 12, 18 months. But the modeling that we have done in the - and marketing intelligence that we get back from our sales team is that there is an awful, awful, awful lot of inventory that exists out there. Back in the waterfall world, we are very satisfied with going and then copying direct to publisher or to web app and finding a way in the food chain where we got to see inventory that wasn't - that was not seen by the - that wasn't sold by the direct sales team, that wasn't handled by, say, AdX, Google's Ad Exchange. Now at the top of the head, you get to see everything and the amount of inventory is simply astounding. And we are very, very pleased with the strategy of trying to go high volume with lower margins. The state of the industry is it's definitely moving to server. I think that you'll see - we had commented, I think, last quarter, that you'll see a lot of momentum around server. You're certainly seeing with Amazon, Google's product is out there, our product is coming to market, AppNexus product is coming to market and they're all server side. And I think the only delay you'll see is just the seasonal adjustment of Q4 where most Chief Information Officers don't want to touch websites during that period of time. So I think the big whoosh comes in 2018 when they free up resources to be able to switch over to header. But everyone we talked to in the marketplace that works with us wants to see our header solution and/or wants us to partner with some of the other folks I just mentioned, but all on the server-side basis.

I was going to shift to the nToggle infrastructure cost. But if you have a follow-up to Michael, go ahead.

No, that's actually where I was going.

Okay. Great. Yes, from an infrastructure perspective, initially, there'll be some initial CapEx cost, $4 million to $5 million, maybe the remainder of this year. Because our initial integration is for sort of speed to market as it were is through hardware. Over the longer term, that integration will shift to more of a software-type integration. And one of the reasons that we acquired nToggle, first and foremost, was for the revenue lift. But as a secondary reason, we think that it will actually decrease our per unit cost over time. And so it will be more efficient and effective for us over time, and it will also decrease the cost of our partners on the buy side or DSPs.

Got you. So I guess, when we take a couple of years out as volumes, as ad spend starts to recover, we can think about prior gross margin levels, but this probably gives the incremental lift, all things equal to that gross margin, is that fair?

That's correct.

Okay. Perfect. And just maybe one more for me. When you think about increasing access to inventory, improving win rates, you talked about perhaps taking share. Do you see yourselves as taking share from some of the other SSPs that maybe don't have the demand side to their network? Could you see that's taking share from some of the bigger exchanges out there, any color there? And then maybe if you could just talk to, relatedly, just the uniqueness of the demand that Rubicon is bringing right now relative to other players out there, SSPs and other networks out there? Any color there would be helpful. Thanks, guys.

No, I can jump on that. So there's no question that we - part of the strategy is to take share. The buyers are all telling us that they are going to plug into fewer platforms, not more. And that stands to reason, because they're seeing a lot of the same inventory across multiple platforms, and in a world where our inventory is more at par or at parity with our competitors, if you recall back in the waterfall days, Rubicon had access to proprietary inventory, and that attracted buyers. In the header bidding world, just give me far and far fewer examples in exchange with proprietary inventory. So therefore, you better build a value proposition that isn't surrounded upon getting exclusive access to inventory. And that's our whole strategy as it relates to cost of the pricing, as it relates to marketplace dynamics and transparency, and as it relates to this nToggle acquisition to try to cut down the cost for the buyers and increase their matching rates. So that combination of the 3, we're pretty certain that a lot of our competitors are going to have a real challenge with that. They may be able to do 1 of the 3, 2 of the 3, but it's going to be real hard for them to do all 3. So when the dust settles and they are only working was a handful of exchanges, we fully anticipate to be one of those exchanges and to gain a disproportionate share of wallet.

Our next question comes from Kip Paulson of Cantor Fitzgerald. Please go ahead.

Hi. Thanks for taking my question. Just a couple from me. First, could you give us some qualitative color on orders ad spend? You guys were the first to go to market with premium programmatic, I believe. So just curious how ad spend here performed in the quarter relative to RTB? And second, ad spend going through FastLane has a lower take rate. But could you help us size up how much ad spend is actually going through FastLane at this point? Thanks.

Sure. It's David. Yes, we're not sharing exact breakouts on the orders ad spend, but it absolutely continues to grow. So with the concerns of inventory quality that have been out there in the press quite a bit recently, there's a lot more discussion and we think demand for the orders - products, so it's definitely strong and growing for us. And sorry, your second question?

It was on FastLane and how we should be thinking about how much spend actually goes through it at this point.

Yes. When we last discussed that, again, we're not sharing the exact percentages, but it also continues to grow. And as we mentioned, we have the critical mass of implementations that we think are important. And we think it's sort of, I don't know if cresting is not the right word, but there's certainly a huge amount of momentum as we shift to the server side. And I think you're going to see, we think, over the next short and medium term, there's going to be - that will be the next wave as it were and that client side will really shift to server side.

Yes. There's definitely been a - with all the talk of the server solution, server-side solutions coming to the market, there has definitely been a pause on client integrations on the client side. And so we think that once our offering is out there and others are out there, that it will only further accelerate the axis inventory from the header.

All right. Great. And then, finally, it looks like restructuring was $1.6 million in the quarter. Just curious how we should be thinking about that over the next quarter or two? And that's it for me. Thanks.

Yes. Sure. We should - we're done with those costs, so there shouldn't be - if there's any cost in the future, they'll be relatively small.

Your next question comes from Brian Fitzgerald of Jefferies.

Hi. This is John, on for Brian. Thanks for taking the question. Just curious, as mobile becomes a greater and greater portion of the business, what are some of the puts and takes of getting inventory online there that might be a little bit different than the desktop space? And then, also, if you could just comment a little bit maybe characteristically on kind of how you think about take rates on mobile versus desktop and the growth that's driving there?

Sure. I'll start and Michael, please add. I guess I'd breakdown mobile, there's mobile web, which is sort of subject to a lot of the same dynamics as desktop from a sort of take rate and win rate perspective. Where we're really focused is mobile app, which we feel like is a stickier and a little more unique proposition. And so that's the primary focus of our mobile activity today.

Yes. And as far as the onboarding, it's actually very interesting time. Back in the days before header, especially server-side header, we - you really had to fight in the trenches and it was all around getting your software development kit installed in the application. That was hard because you're fighting for a limited number of slots. And if you weren't born to the mobile app world, like Rubicon wasn't, it presented a long sales cycle to be kind. We see a real opportunity with server-side header, particularly partnering with Amazon as a fast way to get even more access to the mobile app inventory. The challenge then becomes, from our standpoint, helping our advertisers who are traditional brand advertisers, understand the environment, understand the targeting, understand the context. And Tom, our CTS, is doing a great job in doing some real heavy lifting, making it easier for our buyers to be able to gain access to that inventory and buy it. So I think that the supply, interestingly, will come and the demand will lag on it, just given the fact that mobile app is so much different for a DSP to buy than it is for mobile web. And we're working very closely with our DSP partners to make sure that, that inventory is both recognizable and digestible.

Great. Thanks, guys.

Our next question comes from Mark Kelley of Citigroup. Please go ahead.

Hi. Good afternoon, guys. Thanks for taking the question. The first one is on just the competitive dynamics of header bidding. I think some of the smaller players that maybe gained some early share caught some people by surprise. And I'm just curious if the ones that lack scale, you're starting to see those guys kind of go away because they can't compete with you and then obviously, we have the shift to server side which requires even more scale. So just curious what you're seeing there. And then second, do you have any idea of what CPM inflation looks like during the shift to server-to-server? Is it similar to what you saw in the client side? Anything there will be helpful.

So I'll address the competitive dynamics and David, you can talk about the CPMs. We haven't - to be honest, we're not aware of anyone that's gone by the wayside, given the recent market changes and the infrastructure challenges of running a header business. But with the combination of what buyers are telling us in terms of we only want - we want smaller, not larger pools of platform, the idea that a lot of them are now subjecting exchanges to a degree of scrutiny and transparency that they haven't been exposed in the past and of course are working aggressively on the rate side, we fully anticipate that it will get more difficult in 2018, not easier, for folks that are not well capitalized to compete against us in this business.

Great. And from a CPM perspective, it's interesting as you - as we got into the header bidding world, CPMs in the header bidding world are higher than actually that in a waterfall - in the waterfall world because of the access to higher premium inventory. And so we think that the CPM dynamics between client-side header bidding and server-to-server are probably similar. But if anything, the more - it'll be more driven by inventory mix than access to higher-quality inventory, more than anything.

That's helpful. Thanks a lot guys.

[Operator Instructions] Our next question comes from Sameet Sinha of B. Riley. Please go ahead.

Yes. Thank you very much. A couple of questions. If you can talk about how - as you outlined your thesis about transferring the business into kind of a high-volume, low-margin business, I heard your comments about direct orders earlier. But how does that fit in to this overall platform, because the very high CPM there and I understand the take rates are low, but it's definitely not high volume. So how does that fit in? And my second question is, thinking about as you phased out the old structure toward that high-volume marketplace, how do you think about your capital allocation? Are there areas that you need to fill? When you made the nToggle acquisition, you kind of addressed some of the pinpoints. Are there other areas where you want to address that or - and in general, if you can comment on capital allocation thinking. Thank you.

Okay. On the first question, I'm not sure if I fully understood. Can you just rephrase that? So you talked about high volume, this lower margin shift that we're getting into. Can you please repeat your question?

Yes. The orders business, the way I understand it, includes all in the private marketplaces and all those are lower volume businesses. I know things have changed pretty dramatically, but as I understood it, to lower volumes. So is that going to be - continue to be a focus for the company, because it's definitely one of the differentiating factors for Rubicon before the header bidding took on. So is that going to be a part of your - continue to be part of your focus in private marketplaces direct order?

Yes. I wouldn't say that order is actually a lower volume business. I mean, I think there's quite significant volume there.

Yes. I think the way orders were slightly different is probably a much higher in terms of - CPM is much higher; rate, lower; take rate lower but match rate, win rate, much, much higher because it's not competing, say, in an open auction where there's trillions of inventory at play. It's very strategic to us to be a big player there, largely because of the types of clients that we have. If you look at this kind of slate to brand safety in programmatic, in ad quality in programmatic, that's our platform, and that is what we pride ourselves on, this curated, well-lit environment. And our marketers, our sellers and buyers are bringing us in that direction. And so it remains to be a big area of investment for us and a big interest from both our sellers and buyers.

Yes. So that buy side demand is really going to drive a lot more growth on the supply side. And then from an acquisition or capital allocation perspective, we're taking things a day at a time, we're always looking to improve what we do and we'll continue to assess our needs. There's nothing immediate that we're talking about at this point in time.

But I think it's very fair to say that in acquisition like nToggle, the profile of - and perhaps, any future acquisitions as opposed to swing for the fences, bet the ranch acquisitions. So I think we'll be very prudent, very thoughtful about it and if it fits with our strategic vision and can accelerate that, outstanding. But we'll also be very cognizant of the fact that without - even with a strong balance sheet, we have our limitations.

As a quick follow-up. So if I were to understand the - in terms of acquisitions, you'll try to make more acquisitions like nToggle which are improving the efficiency of the marketplace or could you also look to increase the size of your marketplace maybe acquiring other SSPs, which are maybe exposed, can you elaborate on that, please?

Well, I certainly don't think we should take anything off the table. That said, I think acquiring another SSP, without going in great detail, will be much more - the profile of that acquisition would be - exceed the profile of, let's say, an nToggle acquisition. And that is not something that we are contemplating right now.

Thank you very much.

Thanks.

This concludes our question-and-answer session. I would like to turn the conference back over to Mr. Michael Barrett, for any closing remarks.

No closing remarks here. Thank you, everyone, for your time.