推特 (TWTR) 2014 Q3 法說會逐字稿

Good day, ladies and gentlemen, and welcome to the Twitter third-quarter earnings conference call.

(Operator Instructions)

As reminder, this call is being recorded. I would now like to introduce your host for today's conference, Krista Bessinger, Senior Director of Investor Relations. Ma'am, please go ahead.

Thanks, Danielle, and good afternoon. Welcome to our Q3 earnings call and thanks for joining us. We have with us today our CEO, Dick Costello; and CFO, Anthony Noto.

We will begin with approximately 15 minutes of prepared remarks followed by Q&A. During the Q&A, we will take questions asked via Twitter, in addition to questions submitted from conference call participants. Questions submitted via Twitter should be directed to @TwitterIR using the hashtag TWTRearnings.

We would like to remind everyone that we will be looking forward-looking statements on this call such as our outlook for Q4 and 2014 and our operational plans and strategies. Actual results could differ materially from those contemplated by our forward-looking statements and reported results should not be considered as an indication of future performance.

Please take a look at our filings with the SEC for a discussion of the factors that could cause our results to differ materially. The forward-looking statements on this call are based on information available to us as of today's date, and we disclaim any obligation to update any forward-looking statements except as required by law.

During this call we will discuss certain non-GAAP financial measures. Reconciliations to the most directly comparable GAAP financial measures are provided in our earnings release. These non-GAAP measures are not intended to be a substitute for our GAAP results.

And audio replay of this call will also be available via Twitter and on our website in a few hours. And with that, I would like to turn the call over to our CEO, Dick Costolo.

Hi, everybody, and thanks for joining us.

I want to spend some time giving you a view of our strategy to build the largest audience in the world, a strategy we've started to lay out for you on the last two calls. But first, an overview of the quarter.

We had another very strong quarter financially; revenue growth of 114% year over year totaling $361 million. Adjusted EBITDA rose to $68 million, up from $9 million just a year ago.

We continue to make progress in user growth, adding 13 million net new users in the quarter, including another 3 million net new users in the US for the second straight quarter. That brings us to 43 million new monthly active users for the year and 284 million monthly active users in total.

We've talked last several months about the size of our total audience and the additional reach of our ad exchange. So I want to put all the pieces together for you today, and what you'll hear will echo some of my comments from last week at Flight, our mobile developer conference.

You should think about the size of our total audience as a series of geometrically eccentric circles. At the core, you have our monthly active users. They are our most engaged users, contributing and consuming the vast amount of great content on Twitter. Those contributions are the fuel that powers the entire system.

In the circle behind that, we have the logged out audience on our owned and operated properties, hundreds of millions of users that come to Twitter every month but don't log in. We've talked about the size of this group of users as another 1x to 2x the size the core, and that remains case.

In the third circle beyond this are the people we reach in syndication via embedded tweets and timelines across the web. And now across the mobile app ecosystem through our new mobile developer platform fabric, which I'll talk more about in a minute.

We think about everything we do in the context of this set of geometrically eccentric circles and think it's useful for you to as well. Then I would say we have three objectives that we're focused on across this total audience.

One, strengthen the core. We have to continue to grow our monthly active users and make Twitter an increasingly daily use case for them.

Three of the priorities to strengthen the core include improving the new user experience and getting new users a high-quality timeline the moment they sign up for the service, second, providing much better rich media creation and consumption tools and experiences to drive more breadth and depth of content, and lastly, adding functionality to our direct messaging service to enable users to move fluidly between the public conversation and private conversation, all on Twitter.

This quarter saw several launches along those lines. As one example, we rolled out a brand-new web onboarding flow that dramatically simplifies the sign-up process and that automatically creates a timeline based on what topics users choose, as well as their browsing history.

We get them in a fully populated timeline that does not require them to immediately know the language of Twitter or who is on Twitter, et cetera. Expect to see this on mobile in short order.

Not only is the breadth and depth of content expanding, but that content is some of the best across the digital landscape. Just looking at television and music, we hosted the season premiere teaser of Showtime's hit show, Homeland, and one of ABC's new shows, Selfie, and exclusive videos in music from such diverse musicians as U2, One Direction and a Michael Jackson video, all exclusive to Twitter.

Our second objective is to reduce barriers to consumption to help build the largest total audience in the world. We will continue to innovate on ways to better organize our content to deliver the right experiences at the right time for all types of users logged in, logged out, and on syndicated partner properties.

And the third objective is building new applications and services. Our announcement of Fabric, a week ago at our developer conference, is the most recent initiative in this area. Components of Fabric are already in use by tens of thousands of developers around the world, collectively reaching well over 1 billion iOS and Android users.

We believe Fabric can be the one SDK that any global app developer needs to embed in their application. We provide crash reporting, beta testing and analytics from the moment a developer is getting started.

When developers are ready to launch their production apps, we provide identity services in the form of Twitter login and Digits, our next-generation native mobile sign-up service. Digits provides developers with multiple competitive advantages over other sorts of identity services, including building their own native mobile sign up. And finally, developers can easily turn on the MoPub kit, our mobile ad exchange, when they're ready to monetize their apps at scale, harnessing our best-in-class ad mediation and exchange technologies.

In closing, I'm happy with the strategy and the quality of the work we're doing. But given our significant aspirations and the breadth of the opportunity in front of us, it's more critical than ever that we increase our overall pace of execution.

And with that, I'll hand it over to Anthony to go deeper into the financials.

Thank you, Dick, and good afternoon, everyone. I'll discuss our financial and operating performance for Q3 and provide our guidance for Q4.

Q3 was another strong quarter for Twitter. Total revenue reached $361 million, up 114% from the year-ago period, and $21 million above the high end of our guidance range.

Ad revenue reached $320 million, up 109% year over year. Growth was strong across all channels, with the DSO channel being the largest contributor, while the MMS and small-medium business channel growth rates outperformed.

Looking by product, the vast majority of our year-over-year revenue growth was driven by promoted tweets, up 136% year over year, as we saw meaningful contribution from new ad formats, including mobile app downloads, website cards, promoted video ads. In Q3 US ad revenue grew 88%, while international ad revenue grew 164% year over year and accounted for 35% of total ad revenue.

International growth outperformed the US, driven by strong growth in both APAC and EMEA countries. In Q3, we expanded our sales efforts to 12 new markets in Europe. We also expanded the reach of Twitter's self-serve ad platform to 12 new countries.

We now cover over 45% of our consumer footprint with the ability for advertisers to use our self-serve platform. We now have a sales presence in 60 countries around the world and see significant room for continued international revenue growth, as we further expand and grow each channel around the globe.

Looking by platform, our mix of revenue for mobile also continues to grow, with 85% of total ad revenue now generated from mobile devices, up from approximately 70% from the prior-year period.

Data licensing and other revenue contributed $41 million in the quarter, an increase of 171% year-over-year. This growth was driven by both mobile ad exchange business and a full quarter of the Gnip data licensing business.

Moving on to costs and EBITDA, in Q3 total expenses were $341 million, up 85% year over year. The increase was primarily driven by head count and related overhead costs, as well as infrastructure investment.

We continue to invest in our workforce to scale our business, and drive continued product innovation and begin to drive broader trial of the product. We ended the quarter with approximately 3,600 employees.

Adjusted EBITDA margin for Q3 was 19%, more than three times that of Q3, 2013, and up 200 basis points sequentially. Adjusted EBITDA totaled $68 million compared to just $9 million in the prior-year period, and above the high end of our expectations. Adjusted EBITDA significantly outperformed our expectations due to the outperformance in revenue and lower-than-expected hiring in the period.

Non-GAAP net income was $7 million in the third quarter, up from a non-GAAP net loss of approximately $17 million in the same period of a year ago. This includes approximately $600,000 in interest expense from the convertible notes for the partial period, as well as unrealized FX loss of $8 million from the mark to market balance sheet accounts. Excluding the cash interest expense from the convertible notes, and the unrealized loss FX adjustments, non-GAAP net income would have been $15 million.

For modeling purposes, we expect the quarterly interest expense to be roughly $20 million to $25 million on a GAAP basis and $4 million on a non-GAAP basis. Our GAAP net loss in the third quarter was $175 million, which includes $170 million of stock-based competition expense.

Before returning to metrics, I want to cover a few items related to cash and CapEx. We successfully completed a convertible note offering, raising approximately $1.7 billion, net of the convertible hedge, at very attractive rates.

We ended Q3 with roughly $3.6 billion of cash and marketable securities. Cash used in operations was $88 million, CapEx was $101 million, $62 million of which was financed through capital leases.

Now I'd like to turn to our operating metrics, first on users. Average monthly active users reached 284 million for the quarter, reflecting 13 million net additions in the period. We added 3 million users in the US, and 10 million internationally.

Net adds in the US were in line with Q2, while international net adds slowed a bit sequentially, due primarily to the implementation of increased authentication measures, which negatively affected users in a number of APAC countries. In addition, Q3 did not have the benefit seen in Q2 from elections in India and news events in South Korea.

Timeline views increased to approximately 181 billion, up 14% from a year ago time period. Timeline views per MAU were down 7% versus the prior year to 636, as expected.

The year-over-year decline primarily reflects the changes we've been making to allow users to more efficiently access our content. We will continue to focus on driving a better product experience, which may pressure timeline views per MAU over time. That said, based on current trends, we expect timeline view per MAU in Q4 to be flat on a year-over-year basis.

Switching to monetization, ad revenue per 1,000 timeline views continues to show strong growth, reaching $1.77 in Q3, up 83% year over year and 11% sequentially.

US ad revenue per 1,000 timeline views reached $4.28, up 66% year over year. And international ad revenue per 1,000 timeline views reached $0.84, up 132% year-over-year. We continue to see steady improvement in monetization as we expand geographically, roll out new ad products and improve targeting, attribution and ROI. Given this, and the expected seasonal fourth-quarter strength in advertising demand, we expect to see sequential growth of approximately 28% to 30% in ad revenue per timeline view in the fourth quarter.

Moving on to cost per ad engagement. In Q3 ad revenue grew 109% on a year-over-year basis despite a decline of 17% in cost per ad engagements.

The increase was driven by total ad engagements which grew more than 150% year over year, reflecting higher-quality ads, improved prediction and targeting and media forward. Cost per ad engagement grew 3% quarter over quarter due to the increased adoption of higher-priced and higher-performing ad units.

Now I'd like to turn to our guidance. Before I provide the detail on our guidance, I want to note that in the past we've outperformed our guidance, as our guidance approach balance the opportunities and risks in setting expectations. This quarter, however, we're attempting to be more accurate in our revenue guidance relative to results, as we've placed more weight on the opportunities in the quarter come to fruition. For that reason, and based on current visibility, we do not recommend projections that deviate meaningfully from our guidance.

For the fourth quarter of 2014, we are raising the range for total fourth-quarter 2014 revenue to $440 million to $450 million, which is $23 million higher than the previously implied fourth-quarter range established after our Q2 results. We are raising our adjusted fourth-quarter 2014 expected EBITDA range to $100 million to $105 million, which is $60 million above our previous range at the midpoint.

We expect stock-based compensation expense in the range of $175 million to $185 million. We expect CapEx to be between $120 million and $150 million.

We are raising our range for total revenue for 2014 to $1.365 billion to $1.375 billion, which is $50 million above our previous range at the midpoint. The increase in revenue guidance for the full year of 2014 reflects our Q3 outperformance and our increased expectations for the fourth quarter.

Moving on to adjusted EBITDA, we are raising our range for adjusted EBITDA to $260 million to $265 million, which is $43 million above our previous range at the midpoint. The increase in adjusted EBITDA guidance also reflects our Q3 outperformance and our increased expectations for the fourth quarter.

Before we take your questions, I want to take this opportunity to remind all of you about the upcoming Analyst Day on November 12. We look forward to seeing you there.

With that, we're ready to take questions. Operator, will you please announce first question?

(Operator Instructions)

Douglas Anmuth, JPMorgan.

Two things. First, can you guys talk about the experiments that you're doing so far around content organization? And when you think about your ability to potentially move away from reverse chron or give additional choices beyond that, how do you know when it's the right time to implement that?

And then, Anthony, can you also comment on the impact and explain a little bit more around the authentication efforts that you mentioned around the international user number? Thanks.

Sure, Doug, on your first question I think the key word that you mentioned is experiments. Obviously Twitter will always be a real-time network.

That is our number one priority when we think about the user experience. But there are time periods when something incredibly relevant to you as a user could have occurred hours ago before you opened up the Twitter app.

We will see unique opportunities like that to give you content that's incredibly relevant, even if it's not based on what happened most recently. What I want to emphasize more than anything else is Twitter will always be a real-time network.

We have been experimenting with a number of different initiatives. One example would be if a user repeatedly pulls to refresh on the mobile app and there's no content to deliver to them, instead of giving them the nell set, we're now providing something that we think would be relevant to them. That's just one of the examples of what we're trying in that area.

In terms of the authentication efforts and impact on MAU, over the last two quarters we've taken a more deep view of what can we attribute to our actions within a quarter to our overall net MAU growth. And the specific comment I made about the authentication issues in Asia and that region, we saw a negative impact from a higher level authentication standard that we applied that us to lose 1 million to 2 million monthly active users because of that specific initiative.

Paul Vogel, Barclays.

I was wondering if you could talk a little bit more about your retail initiative on the advertising side as well. The experimentation on the buy button and the recent acquisition of CardSpring and how those all are going to play out together.

Thanks, Paul, this is Dick. The CardSpring acquisition is definitely related to tying that team together around our ongoing commerce efforts. I wouldn't extrapolate too much from what you've seen around the buy now button that we've launched on Twitter to any forward-looking expectations.

We're continuing to explore the way we think about in the moment commerce and now commerce, and the different kinds of opportunities we see in that area. We'll continue to play around with explorations there and look at opportunities in that area. As opposed to thinking about what you're seeing today as what we're launching permanently.

Ross Sandler, Deutsche Bank.

Two quick questions. First on the off-Twitter users, and then a comment on the guidance, Anthony. Can you just help us understand the behavior of logged-off users, specifically how often they come to the Twitter pages on average per month?

What percent of these folks are seeing content today versus a registration law? Then can you talk about new products that might increase that level of engagement?

And then, Anthony, you just made the comment that guidance is the guidance going forward. And that implies about 80% ad revenue growth in the fourth quarter versus the 110% to 130% we've seen over the last couple quarters and the comp looks to be about the same.

So can you talk about where you think ad load is currently? And is that the right way to think about the fourth quarter? Thanks.

First on the logged-out users, we haven't provided specific metrics beyond the fact that our total audience consisting of logged-in and logged-out is two to three-x the size of the logged-in and that hasn't changed.

Let me give you a little sense for qualitatively what's going on as opposed to giving you the specific numbers. We do know the specific numbers. We do study this. We're trying to create unique experience against each one of the four things I'm going to articulate.

The first-use case that a logged-out user comes to Twitter is when they do a search on one of the major search engines for a person. And that person clicks on hashtag, person's name and they come to Twitter to a profile page. That is the one experience that we've actually experimented the most with in enhancing that profile page for the user I'll share with you in a second some stats on what we've seen since that initiative.

The second experience is when someone does a general search, not on a name, it could be on a topic or it could be on an area of interest. They would land just on a tweet-detailed page. No other tweets, just that one tweet, no advertising, no merchandising, no collaborative filtering, just literally that one tweet.

The third experience is when someone comes to our home page or application and hasn't logged in and is looking to probably do a search, but they're looking for what's going on right now. They get nothing. They have to log in and many people do not log in and they bounce off of that page.

And the last experience is when someone sees one of our tweets on one of our syndicated partners like CNN or ESPN, and they click on that tweet and come to our application or home page. In that situation, again, they will get minimal content without logging in.

And so each one of these use cases is a unique opportunity for us to create an experience that's immersive that gives the user the value that we have in content in both breadth and depth. And brings them into Twitter in a way that they want to come to Twitter, and not in the way that we have historically provided information, or the requirement for them to log in.

In terms of stats on the profile pages, we launched a profile page and talked about it last quarter and have been measuring the impact over that time period. And what we've seen is an 83% increase in the number of profile impressions, a 77% increase in profile scrolls of the tweets on that profile page, over a 500% increase in media timeline impressions, over 200% increase in media timeline scrolls, 11% increase in retweets on profiles and a 15% increase on favorites on profiles.

This just starts to highlight how we're scratching the surface on giving these individuals unique content that meets their specific need on that specific use case in which they come. We'll continue to experiment with that user base to capture the opportunity.

And then in terms of guidance, what I'd say is this. Our load rates or coverage, as we refer to it, remains very low. On a year-over-year basis, the bulk of our revenue growth, the majority of revenue growth came from quick rate improvement, although coverage did also contribute.

And we do believe we have significant upside to monetization of our user base, as we've talked about previously, and none of that has changed. The change in the guidance outlook is nothing more than we've been a public Company for over a year.

We have a better understanding of the data that we have in quarter and its predictability, and have historically been very conservative in a highly confident number that we've brought it to the Street. And now we're just putting a little bit more of the balance towards the opportunities coming to fruition, instead of equally balanced between opportunities and risks.

The next question comes from Twitter. It comes from the Twitter account of Rich Greenfield at BTIG. He asks, as you reengage with third-party developers, why are you not syndicating ads to TweetDeck, Flipboard, et cetera?

This is Dick. First of all, let me just set some context for the comment around reengaging with third-party developers. When we announced Fabric last week, our mobile software development platform, that platform is all about providing services to developers, mobile app developers, across platforms from the day they start developing their applications to the day they launch those applications and want to measure growth, to the day they need to monetize and want to monetize usage at scale.

When we think about the opportunity afforded by being part of the foundation of the entire mobile app landscape, we know that there are going to be enormous opportunities there. Those, of course, include helping developers monetize their apps, which is why MoPub is such an integral part of the Fabric SDK. We will think about, and look for, opportunities to provide our native advertising units in syndication, both to other Twitter-based properties like TweetDeck, and across the entire mobile application ecosystem.

Mark Mahaney, RBC Capital Markets.

I wanted to ask about engagement trends. I know that in the past you've had some misgivings or have caveated our views or our looks on that views per MAU. Any updated thinking on other metrics that you could disclose in the future that may give us a better picture on what's happening with engagement?

And then secondly, if you look at engagement trends amongst different cohorts, whatever different cohorts you look at, would you see anything in there that shows a real gap of opportunity? Do you have a cohort in mind that shows dramatically higher than what your average engagement levels are? And that's where you're trying to steer the rest of the user base to?

What I'd say is timeline view per MAU remains our measurement for engagement that we disclose publicly. We look at a lot of other measurements as well. In terms of the engagement with retweets and favorites and those trends all remain positive.

What I'd say about timeline view per MAU, we spent a fair amount of time digging into the cohort analysis and it is actually relatively encouraging. If we look across 14 different cohorts, the seven most recent cohorts, which would be a cohort that just joined in the second quarter, all the way back in time across 14 quarters.

The seven most recent cohorts have lower timeline view per MAU then the seven oldest cohorts. But the trend that they show, is they ultimately get to the average of those seven older cohorts.

We're seeing a convergence to that overall number, it's just the newer cohorts start out lower and then as they become more familiar with the product, they become more engaged and they actually reach the level steady state that the older seven cohorts have. It's actually an encouraging sign.

Dick mentioned that we have an aspirational goal of having the largest daily audience in the world. We obviously do not provide the daily active user measurement.

We really don't have specific initiatives against that today. As those initiatives change and evolve over time, we could consider other measurements, but we're not doing that today.

Heath Terry, Goldman Sachs.

Dick, while I know you're focused on continuous improvement, when you think about the impact that your relatively new heads of product and engineering are having on the way that users see Twitter, how close are they to getting to what you would consider a normalized pace of development? And how would you characterize their vision for product and technology compared to what we've seen at Twitter over the past year?

I would say that, again, I would go back to some of my initial comments in my remarks, that I feel good about the strategy that we have in place and that we have been executing against over the course of the year, and the quality of work that I'm seeing against that strategy. As I mentioned, it's critical that we increase the pace of execution, and you just referenced that.

Let me talk about what I mean by that maybe in a little bit more detail. When I talk about pace of exclusion, I'm really talking about faster iteration from hypothesis to prototyping against that hypothesis, to experimenting against that prototype, to launching against that experiment, and then iterating across that cycle.

It's that kind of clock speed and coordination that I want to see more of from product and engineering working together. That's a combination of both, one, as a CEO I think you always want to see those kinds of things; and then, two, infrastructure capabilities we need to develop to enable the team to move more quickly.

The next question comes from Twitter. It comes from the account of Victor Anthony at Topeka Capital Markets. He asks, on the Q2 call you mentioned that there was nothing structural keeping you from monetizing at the same levels as other social networks. Over what timeframe do you see the gap closing and how are you getting there?

What I'd say is the overall statement holds true. We see nothing structurally preventing us from reaching the monetization levels of our industry peers.

We made the note about $10 of average revenue per user and getting to that level or higher. The key drivers from where we are today to getting there; today our load rate remains meaningfully below our industry peers. And we have significant upside on an absolute basis, as well.

The drivers that will ultimately result in a monetization at that level or higher that we've mentioned, would first be advertisers. Today we continue to drive nice growth in advertisers and spend per advertiser as we continue to improve the ROI that we provide for them. But importantly continue to offer new advertising products, as I mentioned, our promoted video ads, as well as improved SMB platform, in addition to mobile app downloads and then web-based cards.

These specific formats allow the advertisers to pick the specific marketing tactic or initiative they want and better suit their needs, which provides better ROI. As we continue to do that and prove out the ROI, the number of advertisers will grow meaningfully.

Where we are today versus the numbers cited by our industry peers, is we're at a very, very, very low level. And we have magnitudes of upside there. That would be number one. As that demand increases, we'd increase our coverage or load factor to meet the demand.

Additionally, the more information we have and the better dashboards we have will drive continued higher click-through rates and that will result in better ROI. It's combination of those factors that allow us to really unlock the monetization opportunity that lies ahead of us.

We continue to see nice improvements in the advertisers and the spend per advertiser. That's the bulk of what's driving it today.

Justin Post, Bank of America Merrill Lynch.

The first question is on app downloads and video ads. How much did they contribute in 3Q? And could those get big enough to stem the rate of revenue deceleration as you look forward?

And secondly, maybe a little bit more color on the deceleration in 4Q guidance? And is there any FX impact?

We're not providing the specific magnitude of revenue from ad downloads or video. We're encouraged by both the quantity of revenue we've achieved in both of those products, as well as the trend versus prior periods.

Obviously the promoted video is still in beta, and we haven't released it for general availability, where app download has now been available for a quarter.

In terms of the FX impact, what I'd say is it really depends on what your starting point is. If you go back to July 1 versus our guidance today, the magnitude of the impact of FX would be a negative of $5 million. I would emphasize it depends at what point you pick to have that reference point.

In terms of the guidance for the fourth quarter, what I'd say is there's nothing inherent in that guidance that deviates from the other drivers of our business that we see. The only comment we made philosophically is that it's a greater balance towards the opportunities coming to fruition versus an equal balance versus the risks.

Ben Schachter, Macquarie.

At a high level, three to five years from now, do you expect the majority of the revenue to be driven by the core logged-in MAUs? Or is it going to be logged-out network users?

And then more specifically, on the Analyst Day, Anthony, what are the key focus areas of the day going to be? And should we expect multi-year financial guidance?

Ben, there's not a specific answer to your question, three to five years from now. The way I would qualitatively answer it is this: We always have a very large audience and logged-out users, it's one- to two-x the size of the monthly active logged-in users. And so it's a sizable audience, one that we feel confident we can monetize once we've nailed the consumer experience.

Our approach has been to really nail the consumer experience and to have high levels of engagement before introducing advertising. Our advertising team, led by Adam Bain, are very, very much in favor of being able to monetize this universe of users because they come with a very specific intent.

Especially when they're coming from a search engine or do a search on our own properties, or they come from embedded tweet on someone else's property. We know what their interest is, and that's a key element of being able to target the advertising for them.

I think your question may be are they monetizable? And our view is yes, at the right point they are monetizable. The exact magnitude difference between one versus the other, we would tell you our goal is to have the largest daily audience in the world, and that will drive the largest value for shareholders.

In terms of Analyst Day, it's been a year since we went public and I would characterize the reason for Analyst Day in the following way. Our strategy has evolved over the last 12 months. It's a much more aspirational strategy and one that can allow the Company to be much larger than just going after a logged-in user base.

So what we want to do at Analyst Day is lay out the opportunity that sits in front of us, the specific strategy that we're using to capture it. Where we are relative to that opportunity and what are the key priorities that lie ahead of us so you can track us relative to the opportunity.

The next question comes from Twitter. It comes from Daniel Ernst at Hudson Square. And he asks, would love to know DAU as a percent of MAU on the network on any given day.

We get this question on every conference call, and what I'd like to clarify is a couple things. When the question is asked on what your DAU to MAU depends on what time of the year it is, and what portion of the overall user base you are focused on. So want to provide more specificity on this call, and then just be very clear that we're not going to continue to update DAU to MAU until we have a specific strategy behind driving this measurement.

If you look at our top five markets in the second and third quarter, because there is seasonality differences, you'll see that in our top five markets, the DAU to MAU ratio is in the low 50% range. If you look at our top 10 markets during that same time period, 2Q and 3Q, you'll see that our DAU to MAU is in the high 40%s to low 50% range. And if you look at our top 20 markets, which make up 80% of our overall user base, you'll see in the second and third quarter that our DAU to MAU is in the high 40% range.

It's pretty consistent with what we've polled in the past, but I want to make it clear that certain markets are above 50% and certain markets are below them, depending if they are an emerging market or not. And it also depends on what time of year you ask the question.

Eric Sheridan, UBS.

Maybe two. One, I wanted to know if we could get a little bit of color on self-serve ad products? The countries and regions that have launched it? What you've seen to date as those have launched? And how that will inform your view of that going forward with self-serve?

And then, Anthony, maybe I could take a swing one more time at the Q4 ad revenue. You're implying in the guidance, or the comments you made on the call, that engagement actually gets a little bit better per timeline per MAU in Q4. I wanted to dig in on the Q4 ad comment, to whether that implies either ad load or price per ad might moderate in Q4?

On self-serve, what I'd tell you is we're very encouraged. We're not going to go through every country that was released in the most recent quarter, we mentioned 12 of those, and the footprint that it now covers. What I would tell you is the number of SMB advertisers that are coming to Twitter, using our advertising product, accelerated in the quarter. That's a direct result of the adoption of the self-serve platform, and we're very encouraged by that.

In terms of Q4 guidance, what I'd say is our guidance reflects a combination of top-down and bottoms-up. And you're coming at it from a bottoms-up perspective. The top-down perspective is one driven by what we see already having been accrued in revenue in the quarter, what we see in booked revenue and what we see in a pipeline revenue.

We're not constrained by supply. So your question implies that there could be some constraint there. What I'll tell you is demand is the key driver of our revenue. And demand is obviously a function of ROI and advertisers' current businesses. That's what's reflected in our guidance.

Mark May, Citi.

I think most of it's been captured, but wondering if you could talk a little bit about how the adoption of self-serve, which I think is still kind of early in the process, has helped improve coverage or sell-through rate or [adalid], whatever you want to call it, and where we are in that process. Because we know from other similar businesses, the adoption can be quite quick. Want to get a sense of where we are in the process and how that's been impacting coverage.

And then a question around content. Just wondering what role or exclusive or unique content plays in the value that your users get from the service? What, if anything, that you guys are doing to secure the rights to more exclusive and unique content from celebrities and other brands?

On your first question in terms of self-serve impact on coverage. Our self-serve business is relatively small compared to our overall business. We're still a predominantly direct sales organization, with that accounting for the bulk of our revenue. The trends we see in self-serve are very promising, but it is very early days. It's not really having an impact on how much we have in load or coverage.

The answer to your second question, Mark -- this is Dick -- is that that kind of unique content, and I would say unique regional content around the world, plays a very important role in the value we can deliver to our users. And we have a global media team led by Katie Stanton, who focuses specifically on securing that kind of unique content from all sorts of verticals, be it film, TV, music, politics, et cetera.

Anthony DiClemente, Nomura.

First for Anthony, on international, can you help us with the timing of the international expansion of those 12 additional markets in the quarter in Europe that you talked about in the release? And then, if I may, what proportion of new MAUs in the quarter may have come from those 12 new markets versus the existing?

And then for Dick, on video, you talked about exclusive video content in your prepared remarks. On the advertiser side, you have promoted video in beta. Are you in experimentation mode with video at this point? Or should we start to see Twitter roll out video ad products more aggressively from here?

Anthony, on the international comments, the roll-out of our SMB markets internationally didn't have an impact on MAU. I would tell you the initial performance financially from a market that's opened intra-quarter doesn't really have a significant impact on the overall SMB revenue number.

I think your MAU question I would answer in the following way. We've done a number of different initiatives in the third quarter and second quarter that we've gone back and have been able to tie the specific results of those initiatives against outcomes.

In the third quarter, if you think about our MAU growth, and the 13 million MAUs that we announced, about 3.3 million of those came from specific actions that we took and we can attribute their adoption to the platform from. And they would fall into categories like discovery and notification.

Which leaves you roughly 9.8 million that came from, what I would call, organic growth. The actual number for organic without the negative impact that I mentioned from the authentication bug, would have been 11.8, but it turns out to be 9.8 because of that bug.

We drove 3.3 million net MAU growth sequentially from specific actions we took in the categories of discovery and notification. And that was little bit better than what we saw in the second quarter when it was 3.1 million.

Hey, Anthony, it's Dick. Regarding your question about rolling out video products more generally speaking, I would say, yes, I referenced in my opening remarks that we want to focus on communicating through media and delivering both creation and consumption experiences that are more media-centric. And that applies to both images and video, absolutely.

So you will see us invest in both on the user side of that and on the Amplify side of that, Amplify being our professional video brought into the platform. I guess I'll leave it to Anthony to comment on the advertising services related to video that we're providing.

So today, on the video side of the business, we have two ad products. One is Amplify, which we launched last year and remains a very important advertising vehicle, where we partner with key content owners like the NFL.

The second product is what we just released in beta, which is promoted video, which has gotten a very nice uptick. And as you know well, mobile obviously is benefiting from the secular advertising trend.

Additionally, social advertising is benefiting from a secular trend. And video mobile social is even better. So we're happy to have a product in this specific segment for our advertisers and so far the adoption has been really positive.

The next question we're going to take from Twitter comes from the account of Krim Delko. And he or she asks, are there any events this quarter that are similar to the World Cup last quarter that drove engagement? And what did you learn from that?

The one thing I'd like to make clear is when we did World Cup last quarter, it was an experience that was focused on monthly active users, and was not broadly advertised to non-users. And we saw a nice engagement with the product, but we didn't see an impact on monthly active users.

We've gone back and looked at the numbers a number of times now that we're in the second quarter, and want you to have the same confidence that we have that the World Cup did not have an impact on the monthly active users in the second quarter. And then I'll give you a sense for what we've done in the third quarter, as it may have been similar to the World Cup.

As it relates to the second quarter monthly active users contributed to the World Cup, what we'd say is the following. In Europe there was no change in monthly active user growth over the time of the World Cup. In fact, monthly active user growth slowed in line with expected seasonality in each successive month of the World Cup.

And Latin America, there was a slight acceleration of net ads around the World Cup. But at most, the World Cup added 600,000 new users, which is immaterial. Quite frankly, we probably saw a degradation of those users in the following month before the quarter ended.

The US actually added more net adds this quarter than last quarter, which is interesting on a sequential basis. We added 3.6 million net adds in Q3 versus Q2. In Q2 we added 2.6 million net adds versus Q1.

In terms of what have we done this quarter? We have, currently in the marketplace, hashtag NFL. And hashtag NFL leveraged the technology that we talked about would be repeatable in other tailored timelines and other event timelines.

We've pulled together the hashtag NFL initiative very late in the summer. You will continue to see it improve as it has since the season started. We currently don't have live scores. You can expect that to come over time. We've seen a nice influx of unique content from NFL teams, the NFL league as a whole, and then a number of media properties.

Quite frankly, while the experience is nowhere near where we think it can be over time as we continue to evolve it, the most encouraging thing I've seen is the amount of content that's being created by the NFL and the NFL teams themselves, that's going into that timeline. I can't remember the last time the NFL actually created content for another third-party to use at no cost.

Youssef Squali, Cantor Fitzgerald. Pardon me, Mr. Squali, please check your mute button.

Can you hear me?

Yes, sir. Please go ahead.

The next question, please.

Brian Wieser, Pivotal Research.

Just refining an earlier question, you mentioned APAC and EMEA performed well, but I was wondering if you call out any particular international markets that stood out for revenue growth? But also for net user additions, I was wondering where you thought the monthly user growth was particularly pronounced?

And separately, you mentioned in your last Q that, I guess, it was 8.5% of active users used applications automatically contacting your servers without user-initiated action, as the phrasing goes. I'm curious if there's an updated stat for this quarter?

Yes, on your second question, 11% of our monthly active users come solely through third-party clients. So that's just the third-party client part of what we disclosed last time.

And your specific question, third-party client MAUs that may be auto-pulling. I emphasize the word maybe. It was approximately 8.5%.

Really, there were two things we disclosed. The first was registered MAUs, coming solely through third-party clients, was approximately 11.5%. And third-party client MAUs that may be auto-pulling was approximately 8.5%.

And then in terms of user growth, the user growth is seasonal in different regions. We'd note that Japan continues to be a very strong contributor to growth. And we see growth in many of the other Asian markets, as well, but Japan is the largest of the strength in the APAC markets. So to give you some more additional detail.

And inside of Europe, are you seeing any particular markets that are outperforming in terms of year-over-year revenue growth?

We're not going to get into the specific detail, but Europe continues to be a nice contributor to overall revenue growth on a quarter-over-quarter basis. It was the second-largest contributor to the sequential growth in revenue as a whole, and APAC was the largest contributor, outside the United States, of course.

Arvind Bhatia, Sterne Agee.

Thanks for taking my question related mainly to the user growth discussion. I was wondering if you guys, when you think about the ratio of logged-out to logged-on users, if you see any material differences across various geographies? And then also, in the fourth quarter are there any events that you would like to call out from last year that are not repeatable that we should keep in mind? I know you had your IPO, for example.

Yes, thank you for your question. In terms of logged-out users versus logged-in, on a geographic basis, we're not prepared to share any deltas or segmented information at this point in time beyond the overall user size. And then obviously the use cases that we have today with them.

As it relates to our outlook for monthly active users, I gave a couple of pieces of information and specific intent around timeline views per MAU and revenue per timeline view. And then, of course, overall revenue.

The combination of those three inputs will give you a range of outcomes for our outlook for MAU growth in the fourth quarter, and is the best indicator to how to think about it.

John Blackledge, Cowen and Company.

Two questions. On promoted video, could you talk about who the beta advertisers are and how long you expect it to be in beta? And then, Anthony, I think you mentioned the engagement cohort analysis that you did. For the oldest cohorts, what's the trajectory of their engagement? Is it growing flat or down?

On the promoted video, we're not disclosing specific advertisers. Obviously Twitter is public and so you could spend some time going through your home timeline and see a number of instances in which we have big branded advertisers advertising promoted video tweets on the platform. In addition to Amplify, which is the other form of video that we have.

In terms of the oldest cohorts, so we're anniversarying the big changes that we made in media forward last year in the third quarter, and that had an impact on last year's fourth quarter. In fact, our timeline views per MAU in the fourth quarter of 2013 were down 10% sequentially. So we actually have a quite easy comp there.

As you think about the cohorts, the older cohorts, even though they're anniversarying those changes, are relatively stable. There's a possibility that those cohorts over time will continue to increase.

But right now they're still seeing those product changes that make it more efficient for them to find our content quickly, as opposed to having to spiral through timeline view after timeline view to get to the score of the Giants versus the Royals last night. Or a conversation that took place around that.

So they are stable. Those cohorts timeline view per MAU are stable, but they are still anniversarying some of those changes. And without those changes, they could possibly have been up.

If I could ask just one more question. Could you talk about ad targeting ability of Twitter versus industry peers? I know you gave some other benchmarks for Twitter versus industry peers. How you view your ad targeting ability?

We think we have a -- Dick has mentioned the interest graph and how it gives us so many signals about an individual's interest, and that is very valuable to advertisers. We obviously do ROI analysis and attribution analysis to compare ourselves versus competitors and we couldn't feel better about the value that we provide to advertisers.

If you think about it, the toughest advertiser to win over are those advertisers at the top of the pyramid, the big branded advertisers. And we've had great success and have started with the most challenging advertiser.

And the only way to continue to drive increased spend per advertiser, which we are seeing in our biggest channel, the DSO channel, still today is by delivering that ROI. I'd tell you that's the best way to measures us, is the fact that our ARPA, average revenue per advertiser, in DSO, is still growing.

Our next question comes from Twitter. It comes from the Twitter account of Carter Mansbach. And he or she asks, are there any plans to monetize Vine?

This is Dick. Thanks for the question, Carter. I would say that our near-term and immediate-term focus on Vine are continuing to build beautiful content creation tools for all of the users of that service.

We have some really world-class content creation capabilities there that are allowing people to create these one-of-a-kind experiences that you can only get on Vine. And we want to continue to invest behind that. When I talked earlier about, broadly speaking, investing in communicating through media as a Company, that, of course, applies to Vine. And that will be our continued focus for the foreseeable future.

Thanks, Dick. And I think we have time now for just one last question. So, operator, we'll take our final question, please.

Brian Nowak, SIG.

Two, please. The first one, Anthony, I know you've made a lot of changes to improve the user experience in the onboarding. I was wondering if you could talk about rough user churn rates now? Or even directionally whether they are going up or down?

And then secondly, on the quarterly MAUs, understanding that they're lumpy and very seasonal, can you just talk to the rough size of the impact of the India elections, as well as the soft Korea news events? And then did you see a benefit from the Brazil presidential election, as well as the US midterm election that you're embedding into the 4Q guidance?

Thanks, Brian. On the user experience, one of the things you mentioned new user experience that we launched. I'd like to share with you a few stats on the performance that we've seen since we launched the new user experience.

First, after 30 days after launching, we saw the number of new users that allowed us to have access to their contacts increased by two-x. Which is an important piece of data that we can use to build a timeline for them, and to also drive other engagement factors down the road.

Additionally, we saw a 13% increase in those that have started the registration process actually making it to the home timeline and getting the benefit of all the great content we have. And then we saw a 200% increase in the number of accounts followed. Again, an important measurement that will ensure that we're delivering value to these new users as they come on board.

So in addition to reducing the number of steps it takes to be a new user from 12 steps to approximately six or seven, and building out a very strong and robust timeline, we are seeing an improvement in some of the underlying metrics. In terms of the funnel dynamics, I would say nothing's really changed significantly on the funnel dynamics. And that the change that we're making will over time improve the funnel dynamics. We're focused on both the top of the funnel as well as the bottom of the funnel.

And then on the MAUs?

In terms of your specific question as it relates to India and South Korea, it was around 1.2 million users in total.

And then did you see a bump from Brazil yesterday? And are you expecting a bump from the US midterm, as we go into the fourth quarter, in your guidance?

We don't comment on specific events intra-quarter. Use the three variables that I gave you to triangulate to an overall view for the fourth quarter.

Thanks, everyone, for joining us. We appreciate your time and we look forward to speaking with you again next quarter. Thank you.

Ladies and gentlemen, thank you for participating in today's conference. This does conclude today's program. You may all disconnect. Everyone have a great day.