Ken Tinsley - Corporate Treasurer and VP, IR Doug Merritt - CEO Dave Conte - CFO

Raimo Lenschow - Barclays Walter Pritchard - Citi Brent Thill - UBS Michael Turits - Raymond James Melissa Gorham - Morgan Stanley Brian White - Drexel Kash Rangan - Bank of America Merrill Lynch Jesse Hulsing - Goldman Sachs. Matt Swanson - RBC Capital Markets Keith Bachman - Bank of Montreal Ed Maguire - CLS Taylor Reiners - Piper Jaffray Jason Velkarvh - Robert W Baird Greg McDowell - JMP Securities Karl Keirstead - Deutsche Bank

Good day, ladies and gentlemen, and welcome to Splunk Inc. Third Quarter 2017 Financial Results Conference Call. At this time, all participants are in a listen-only mode. Later, we will conduct a question-and-answer session, and instructions will be given at that time. [Operator Instructions] As a reminder, today’s program is being recorded. I would now like to introduce your host for today’s program, Ken Tinsley, Splunk Corporate Treasurer, Vice President of Investor Relations. Please go ahead sir.

Great. Thank you, Jonathan and good afternoon, everyone. With me today on the call are Splunk’s CEO Doug Merritt; and CFO, Dave Conte. Our press release was issued after close of market today, and is posted on our website. This conference call is being broadcast live via webcast, and following the call, an audio replay will be available on our website. On today’s call today, we will be making forward-looking statements, including financial guidance and expectations for fourth quarter and fiscal 2017, market opportunity, competitive position, and momentum, our business including cloud offerings. Products and services mix, expected growth in our international business and planned investments and expectations. These statements reflect our best judgment based on factors currently known to us, and actual events or results may differ materially. Please refer to documents we filed with the SEC, including the Form 8-K filed with today’s press release. Those documents contain risks and uncertainties that and other factors that may cause our actual results to differ from those contained in our forward-looking statements. These forward-looking statements are being made as of today, and we disclaim any obligation to update or revise these statements. If this call is reviewed after today, the information presented may not contain current or accurate information. We will also discuss non-GAAP financial measures, which are not prepared in accordance with Generally Accepted Accounting Principles. A reconciliation of GAAP and non-GAAP results is provided in the press release and our website. With that, let me turn it over to Doug.

Thank you, Ken. Good evening everyone and welcome to the call. I am pleased to report a solid Q3. I want to first thank our employees and partners for their many contributions. Of course, as Splunk is all about our customers and the best part of the call for me is having the opportunity to share a small sampling of their successes. This is reflected in our total Q3 revenues of $245 million, up 40% over the same quarter last year. Our growth continues to come from the combination of new customers, our existing customers expanding their deployments, and increased adoption in the cloud where we again exceeded our internal plan. Q3 also allows me to acknowledge everyone who joined us at dotcom, our seventh annual user conference. We had more than 4500 attendees and the richest set of used cases ever presented. Two of my favorites were Cox Automotive who uses advanced analytics and machine learning where Splunk and IT Service Intelligence or ITSI to ensure their revenue is flowing by keeping their auto options up and running, and Dunkin' Donuts, who had aligned out the door for the session and it wasn’t just for the free donuts. They use Splunk Enterprise to analyze the preferences of their 5 million loyalty program members, and gain real time insights into the effectiveness of their marketing campaigns. Splunk helps Dunkin' Donuts increase revenue and customer loyalty. Dotcom has also unique opportunity to share the keynote stage with our partners from AWS, Accenture, and Insight Engines. Together with these important partners, and hundreds more, we are driving to become the ubiquitous machine data fabric for our customers. I’d like to call out a few specifics. Let me start by thanking Mike Clayville, Head of Global commercial sales for Amazon web services, who joined me on stage to speak about the power of Splunk plus AWS to drive innovation and help customers move to the cloud with confidence. Customers across virtually every industry and geography are benefiting from the agility of AWS and Splunk together to get the most from their cloud deployments. Thanks again to AWS for being a lead sponsor dotcom, and we are announcing at Reinvent our 5.0 version of the Splunk app for AWS. The free app that leverages machine learning and delivers a prebuilt set of dashboards and reports containing key operational, security, and billing insights. Our strategic relationship with Accenture continues to strengthen. We work together closely at a number of our largest clients, and also enable Accenture’s intelligent application management and security solutions. As global leaders in our respective fields, the Accenture and Splunk relationship brings unique value to our clients. It is also a pleasure hosting Grant Wernick, CEO of Insight Engines on stage during my keynote. Grant demonstrated the power of using natural language built on top of Splunk technology, which greatly expands the number of users who can harness the power of the platform without being SPL experts. Insight Engines acts as a search assistant that empowers users to quickly interact with data using common English phrases. I encourage you to watch the video of my keynote, and see Grants demo. While not on stage for the keynote, we had another strong quarter of momentum with Cisco, which named Splunk as global ISP partner of the year for a second year in a row. Splunk was also recognized by Cisco as the U.S. West Outstanding Solutions Partner and the U.S. public sector ISP and consultant partner of the year. I was also very proud to unveil the Splunk pledge, our new philanthropic program through Splunk for good, targeting non-profit organizations and educational institutions. The pledge commits us to donate a minimum of $100 million in software licenses, training, support, education, and volunteerism over a 10-year period. At dotcom, we showcased our latest product releases, including new versions of Splunk cloud and Splunk Enterprise 6.5. These releases deliver powerful advances in machine learning, simplification and data analysis, and further reductions in TCO. Another big change in 6.5 is the removal of metered enforcement or allowing customers to exceed their license amount on a temporary basis. Our customers increasingly rely upon Splunk in mission critical situations, and we are ensuring that the power of Splunk is available when they need it most. This customer focus change fits into our growing customer success initiatives, including our enterprise adoption programs. We want to make it easier for our customers to index larger volumes and varieties of data, which is key to enabling them to realize greater insights and value from Splunk Solutions. Our goal has become the standard for machine data and every account. We also introduced the newest versions of our premium solutions, Splunk IT Service Intelligence, Enterprise Security and User Behavior Analytics. It was the best conference to date and really demonstrated the passion customers have for the Splunk family of products. More broadly, we continue to believe that we have a tremendous market opportunity. Digital transformation is disrupting business models, technology stacks, and how organizations operate. Machine data is the lifeblood of the secular change, and Splunk is the leading technology that enables customers to realize value from this data, but the move to digital, e-commerce, mobile and social organizations are now increasingly using machine data to provide critical context to the transactional data stored in ERP and CRM systems and relational databases and data warehouses. They are doing this so they can make data-driven decisions in real time. Splunk continues to be a leader in operational intelligence with our customers indexing, searching, and analyzing orders of magnitude more data daily than ever before. Splunk’s platform provides the best solution to enable customers to easily and cost-effectively collect, analyze, and get maximum value from this largely unstructured and voluminous data. Our customers and prospects have diverse and demanding technological needs, and their landscape remains highly heterogeneous as it has been for decades. Recognizing this, Splunk is designed as an open platform and increasingly works with and leverages third party and open-source technologies like HCFS, Kafka, Elastic, Plume, and Spark to enable our customers’ success. Increasingly, Splunk is evolving towards becoming our customers’ ubiquitous machine data fabric, solving their big data and analytics challenges and IT operations, application delivery, security compliance and fraud, business analytics, and the Internet of Things. We are focused on these use cases and markets, because each is going through a shift to an analytics based approach where Splunk continues to deliver for our customers. So moving on to execution in Q3. We continue to see strong traction in our core use cases of IT operations and security, a great example of that is the velocity we're seeing in our premium solutions, ES and ITSI which serve these core used cased. ES is now well established and we have been recognized as a leader in Gartner’s Magic Quadrant for SIEM for four straight years. And while this is our first full year in the market with ITSI, the parts are showing the same upside potential as ES. Infact, IDC has declared that Splunk is the global market share leader for the emerging segment of IT operations analytics. Q3 notable wins included Rackspace, who extended its use of Splunk Enterprise in ES and now uses Splunk at its core decision analytics engine across operations, security, compliance, app development, and business analytics. Thanks to our partners at Kinney Group for their work on this opportunity. A large healthcare company purchased Splunk Enterprise and ITSI to increase visibility into its claims process, which was lacking. With ITSI, the Company will be able to quickly track SLAs around claims processing, troubleshoot any ongoing issues, and help its claims team’s process faster and smarter. ZenDesk also extended this quarter, and in addition to using Splunk enterprise as a platform for IT ops, did ES for compliance and replaced legacy SIEM. A large global financial institution sent an EAA for Splunk Enterprise, which they used for a wide range of use cases, including protection against insider and external security threats, fraud detection, ensuring regulatory compliance, and improving customer experience across all of their channels. Other notable wins in the quarter include Emirates Airlines, University of Illinois, Exostar, U.S. Department of State, and BMW. Our cloud business continues to grow, tripling versus a year ago, continuing the momentum we experience in the second quarter. Consistent with our on-prem business, our cloud customers typically start with a single use case in IT operations for security, and then expand across multiple data sources and use cases with Splunk technology delivered as a service. The sampling of our cloud wins include Dow Jones, a Splunk customer for seven years, who chose Splunk Cloud and ES to replace his Legacy SIEM to gain a 360 degree view into the machine data, to speed up security investigations and to reduce their costs. Educational Testing Services or ETS is the world's largest non-profit testing company and it expanded its use for Splunk cloud, for application monitoring across the lifecycle of the testing process. Other cloud wins include University of Miami, and Georgia State University. On the people front, I am very pleased that we have recently strengthened our executive team, welcoming Susan St. Ledger as our Chief Revenue Officer, Richard Campione as our new Chief Product Officer and Brian Goldfarb is our new CMO. Susan has been onboard six months now, while Richard and Brian just recently joined. I look forward to their great contributions to Splunk. Lots of good progress in Q3 that said, there are few areas I’d like to talk about. As we discussed on last quarter’s call, our cloud momentum is strengthening. And as we discussed in the past, for any given use case, a customer will typically make an either/or decision on cloud versus on-prem. I think it's obvious to reiterate, the absolute size of cloud orders are generally smaller compared to on-prem perpetual license sales. One other area worth highlighting is large orders. While we saw a healthy growth and large transactions, on a year-over-year basis, we grew six-figure transactions by 30% and drove an almost 40% increase in the number of seven-figure transactions, the average order size for a seven-figure transaction was lower than last year's Q3, which was a particularly difficult compare specifically because of big deal over performance in last year's quarter. In summary, it was a good quarter, with some variability in bookings composition. Most companies recall 40% revenue growth, 34% license growth and 3x cloud growth are a runaway success, but we set the bar high for ourselves. Our business is scaling fast, and we recognize that we are in a classic tornado market. By the nature of the term, tornado markets have tremendous opportunity, but also test the company's going through them. Splunk is no different, and this is part of the fun of the fast-moving world that is Splunk. We have a once-in-a-generation opportunity, that's immediately in front of us Splunk is uniquely positioned to capitalize on this opportunity, and we are pursuing it aggressively. We are early in our journey, and we are investing for scale and growth. We are delivering high value to our customers, who are expanding their adoption of Splunk as their platform for machine data analytics, both on-prem and in the cloud. We look forward to an in-depth discussion about our plans for next year at our upcoming analyst day. This is a good place as any to hit the pause button and it to turn over to Dave for a deeper dive in the numbers.

All right, thanks Doug. Good afternoon, everyone. Thanks for joining the call. Our Q3 performance was solid against the very strong comps. Third quarter revenues were $244.8 million, a 40% increase over Q3 of last year. License revenues grew 34% year-over-year, totaling $139.7 million, and we recorded 483 orders greater than $100,000 in the quarter up 30% year-over-year. As Doug described, we had another good quarter with our cloud business, which tripled over the prior year and once again exceeded our plan. We achieved these results despite having fewer of our large orders come from cloud transactions in the quarter. In Q2, we achieve for the first time an annual billings run rate greater than a $100 million. We also detailed the overall impact of cloud over performance on license revenues as well as billing growth, given the dollar weighted average duration of cloud orders. Q3 weighted duration was consistent with Q2 at about two and a quarter years. Similar to our on-prem business, the number of large cloud orders we closed in any quarter will continue to impact the variability in our cloud results. At the high end of all of our large orders, those worth over a million dollar for either on-prem or in the cloud, we saw average order size lower by approximately $500,000 on a year-over-year basis. Factoring this change over the total number of seven-figure orders in the quarter represented approximately 20 million in total potential billings. The size and timing of large orders remains highly variable and we expect that variability to continue. For instance, remember, that at the end of the last year, I reported a sizable spike in ASP's due to the impact from our largest orders. Even considering the variable average order size, overall ASP's for the license orders was consistent year-over-year at about 70,000. This large order size fluctuation quarter-to-quarter is especially impactful in the back half of the year where we booked the majority of our largest transactions. We're excited about customer adoption and eager for more customers to take advantage of EAA programs, something Doug highlighted at Users Comp. But with that, we have modeled our fourth quarter considering our actual ASP levels year-to-date and reflected that in our guidance. Now, additional details for the quarter, international revenues were approximately 22% of the total in line with previous levels and comparable on a year-over-year basis. In the near term, we expect overall international revenue as a percentage of the total to continue to be in the 20% to 25% range, increasing and eventually contributing 30% to 35% in the next few years. Our education and professional services represented 9% of revenues in the quarter. Including our cloud transactions, the rattle mix for all Q3 software bookings was 41%, similar to last year. Now turning to margins and other metrics, which are all non-GAAP. Q3 overall gross margin was 84% consistent with the first half of the year, and reflecting the expansion of our cloud business, as well as services around our core market solutions. For context, we model that our cloud offering will reduce total gross margin by 46% [ph] this year and next as we reach higher levels of scale, coupled with continued adoption of Splunk cloud. Despite all the complexity of our transaction composition between cloud, perpetual, large orders, etcetera, we outperformed on the top line for Q3, which translated the better-than-expected bottom-line performance as well. Operating income was $16.7 million, representing an operating margin of 6.8%. Q3 net income $16.3 million and EPS was $0.12 per share based on a fully diluted share count of 138.4 million shares. Cash flow from operations was $45 million, while free cash flow was $32 million and we ended the quarter with just over $1 billion in cash and investments. Now looking forward to the remainder of fiscal 2017, we expect Q4 total revenues of between $286 million and $288 million with an 8% to 9% positive non-GAAP operating margin. This translates to full year revenues of between $930 million and $932 million up from our prior guidance of $910 million to $914 million. When combining these top-line expectations with our continued investment in market groups, product teams, the field and Splunk Cloud, we now expect that non-GAAP operating margin will be between 5% and 6% for the full-year, up slightly from our prior expectation of 5% flat. Since we expect to be profitable on a non-GAAP basis for Q4 and for the full year, for your EPS calculations you should use a fully diluted share count of approximately 140 million shares for Q4, and 137 million shares for the full year. Considering our Q3 actual results around large order composition, including the $20 million impact from seven-figure order sizes, and updated Q4 outlook, I am now expecting our cash flow from operations to be closer to 21% of total revenues for fiscal 2017. Regarding CapEx, I am pleased to report that we're in the final stage of our South Bay campus build out, and we expect to move in next month. And I know that's great news for the employees listening. The new facility will complement our San Francisco HQ building, and provide a unique Bay Area prominence and workplace experience that will attract and retain top talent in the region. The completion of the South Bay project will drive the bulk of our estimated 25 million to CapEx in Q4, yielding full-year CapEx of about $53 million which is on the low end of the $50 million to $60 million we previously expected. We look forward to make it easier for our partners to sell and our customers to consume our offerings as we shift towards our next stage of growth. We’ll be taking a deeper dive on our short-term and long-term plans at our analyst meeting, including how we expect our model will translate to revenue, fillings [ph] and cash flows, and we plan to provide our initial views for next year's revenue outlook at that time. We'll also cover Susan's plans for the field organization, and how we will continue our market group focus. The new rev rac [ph] standard is coming fast so we’ll also spend some important time discussing that interesting topic. In closing, Q3 reflects continued momentum around customer adoption, cloud consumption and margin improvements. I’m enthusiastic about our outlook for the fourth quarter as well as their prospects going forward. We're all looking forward to spending time with you at our upcoming Analyst Day. So with that, we appreciate your time and interest and will open up for questions. Jonathan, we are ready for Q&A, please. Jonathan, are you ready for Q&A please?

Can you hear me? Great. Our first question comes from the line of Raimo Lenschow from Barclays. Your question, please?

Thank you. It's funny, the operator doesn’t go off mute. I'm calling from the AWS conference here, so maybe let us start with a cloud question. Doug, if you just -- if you talk to customers at the moment, can you talk a little bit about how you see that that on-prem versus cloud momentum is moving, and also maybe comment on the 20 million billing opportunity that you got from the ASP [ph] change on the seven figure deals? And then I have a quick follow-up for Dave.

Thanks Raimo. I hope you are having a good time down there, I got some reports about the -- the security jam that we are running with Twitter participants, and it sounds like it was a smash hit down at the conference. Really, really excited about it. So Cloud momentum, as we've talked about a number of times, workloads are continuing to play both on-prem and in the cloud, and we know the rates of transition to the cloud. [Indiscernible] offered a hybrid solution in an arena where you want your product close to the origination, source is key. I think that's why we keep getting such wonderful tracks with AWS. You've got a lot of customers that are performing hybrid approach, and we are thrilled with the way the partnership has continued to progress, and this cloud growth is so important to our long-term strategy so we can serve our customers the way that we need to. And -- we're expecting as Dave was signaling his remarks, continued expansion of that cloud growth and opportunity set. On the large order variability piece, the key element for us was just a tough comp versus Q3 of last year. If we look at the core orders that came through, that $0.5 million delta in average order size was really attributed to a handful of big orders that were a wonderful surprise in Q3 and we've been talking about large order variability for a while. And, and we just saw a different variable rate this quarter.

Yes, hey Raimo, it’s Dave. In particular, those large orders -- that category of orders actually grew 40% year-over-year, which was pretty good, but again those very large handful of transactions last year made for an interesting comp. So there was potential billings that, that changed based on that, that average order size.

Okay. And then that may even lead into my follow on question. The one question I got from investors was around short-term deferred. And was that impacted due to that mix affect as well, or what drove that number?

Yes, so I mean as you know we really focus on total deferred in terms of geography between income statement and balance sheet. But when you look at large order contribution and you say -there was a different order size and it was 20 million, clearly a component of that would have landed in short term versus long. But as we continue to do adoption transactions and cloud transactions, we see a continued contribution on the long-term front, but I do think there's a correlation between those two topics.

Perfect. It makes sense. Thank you.

Thanks, Raimo.

Thank you. [Operator Instructions] Our next question comes from the line of Walter Pritchard from Citi. Your question, please.

Hi, thanks. I must be sitting right next to Raimo here. Just on the ratable, Dave the 41% I think, is flat year-over-year and you talked about strong cloud adoption, which would suggest maybe that cloud was up pretty – as you said tripling year-over-year. What do we do to read into what happened with the other ratable streams, the EAA and the term licenses, and what's driving what looks like a potentially a reduction in that or a flatness in that system?

Yes, so I think it goes back to the same thing Walter, which is a pretty interesting comp last year, so we had a couple, a handful of extremely large seven-figure orders, greater than seven-figure orders in Q3 of last year, which contributed to the ratable percentage at that time. So if I look at the body of the transactions this quarter, in terms of the mix of cloud and EAAs, we measure the percentage in terms of percentage of the software booking, so when you have the change in the comparable of the seven-figure order, you get a different numerator and denominator. So I don't -- I don't think there's an indication of slowdown in terms of adoption, rather we just didn't have a couple giant transactions in Q3 of this year.

Okay. And then just to follow up on something that Doug said around the Analyst Day, discussing the go-to-market plans. You have a new Chief Revenue Officer, and I'm wondering as you look into next year, how substantial are the changes you expect in the field organization, given it's a new fiscal year, customary to make changes, and then you have a new person in charge there?

Walter, I think it will be a continued evolution that Susan is going to be driving. As we’ve covered over and over, they have got this massive market that we’re trying to go after. The heartache that Dave and I feel on all these calls and in between the days – in between is. how do we get enough coverage out there to be effective in prosecuting all the opportunities, and that’s had constant balance of putting enough reps into the existing accounts, so that cohort that we've been talking about actually comes into play and we see the continued adoption both in data volumes and in overall revenue contribution from those existing accounts, but also carving off enough reps to focus on only net new, and we just find over and over that's hard to balance those two. So she'll be making adjustments keeping in mind not making too many shifts on account relationships and rep relationships and all the usual crossing your T's and dotting your I’s as far as minimizing field disruption as much as possible, yet doing what's right for the business so we can increase coverage. And there will be a lot more detail when we go through in January.

Great. Thank you.

Thanks, Walter.

Thank you. Our next question comes from the line of Brent Thill from UBS. Your question please.

Doug, just as it relates to the Cloud, when you mentioned some of the lower initial deal sizes. When you look at multiple quarters of data and how those customers come back, are you starting to see a trend on on the cloud and what you're seeing kind of multiple quarters in quarter you know quarters after the subsequent purchase even though they may come in initially smaller, the repeat cadence seems to pick up? Can you just give us a sense of what’s happening there?

Great question, Brent. It’s probably a still a little bit early on cloud by itself to have enough data points to be firm on the answer. But we're seeing an early cloud customers, it look a lot like our own on-prem customers and actually a big chunk for our cloud customers or hybrid customers that are on prem as well. And we just updated the cohort analysis, we do on a consistent basis, the numbers are consistent with what we've seen in past cohort analysis that we analyses that we've driven out, which is the repeat purchasing behavior of customers is very strong, which is why we consistently see 70% of revenue coming from existing customers and our job is to make sure that we're getting more and more data sources, in from each of those customers going horizontally across those customers, doing whatever we need to lunch and learns, many Splunk lives at customer events to help them see the additional use cases, so that they can understand the power of harnessing big data when there's multiple conflicting sources that are hard to structure. And that's why I think we see that growth within existing accounts. And for those cloud customers who have been on for a while, we're seeing that same type of characteristic will have better information as more quarters go by.

Okay. And quickly for Dave, some questions around the government and some large deals this quarter, could you just recap there and also just maybe clarify international that decelerated again. It looked like it was a tough com, but any color there, other than the comp on international growth?

Yes, if you look at the composition of those large transactions, again up 40% year-over-year. As you would expect, and again it ties right into the cohort and how we see the buying behavior customers are more mature customers who are on that journey to enterprise wide adoption are the longer tenured ones. Many of those longer tenure customers of course are dispersed geographically, but internationally the U.K. would be where we've got our longest tenure customers. So when you look at the overall dollars contributed from large orders, it's reflected in our international results and public sectors as well. Now I know that there is some commentary or some channel information about some eight-figure transaction and public sector. And we have some really nice transactions in the public sector, but we're not reporting any eight-figure transactions that are in our financial statements in the quarter.

Thank you.

Thanks, Brent.

Thank you. Our next question comes from the line of Michael Turits from Raymond James. Your question please.

Hey guys. Michael, the question for Dave and for Doug I guess. I want to come back to Walters question on the on the ratable rate. I understand it was a top comp year-over-year, but it was also down from last quarter. So same question, what brought it down sequentially? And just a to parse that wouldn't have had a positive impact on building since that seems to add credible to actually hurt billings in the past?

Hey, it's Dave. I think you got to look at the ratable percentage in terms of absolute dollars. So if we go back to Q1 where our ratable rate was actually over 50% that looked eye-popping, but on a relatively small much smaller absolute size of the business at that time. So, the early quarters are certainly more impacted by the variability of large orders that happen to be ratable. Q2 specifically we had four of our 10 largest orders were cloud, so that drove a big component of ratability in Q2. And Q3 we didn't have four of our largest orders we had two. So when you look at order composition in the big orders it really flows through in terms of the ratable rates.

And then just an old question that gets asked but I just wanted to check in on it. How was the pricing and discounting in the quarter, anything unusual?

There was no significant trend that we saw in discounting this quarter. Michael, I go back to the comp. If I look at those, 40% of orders, there are seven-figure orders, that comp was a tough comp. So in that band we had that 500K drop on average across those deals. But we're looking at the body of total transactions. It was very similar as far as ASPs and overall discounting we've seen in the past quarters as well.

Okay. Thanks Doug and thank Dave.

Thanks Michael.

Thank you. Our next question comes from the line of Melissa Gorham from Morgan Stanley. Your question please.

Great. Thanks for taking my questions. Dave you talked about the cloud being better than expected and then the default comparable in a large deals, but license revenue did I think come ahead of what most of us were expecting and it did accelerate in the quarter. So, can you maybe help us reconcile that strength along with the headwinds that you mentioned?

Yes. Hey, Melissa. I think I had one comment in my prepared remarks that was intended to be sympathetic for everyone building a model out there with the challenges we have building our model in here, which is variability in large orders cloud mix, adoption transactions makes it obviously challenging to forecast individual line items in the financial statements. In terms of the uptake in license again in Q2 we had four of our 10 largest orders be cloud. In Q3 we had two of our 10 largest orders to be cloud. So even though we had over performance of our plan in Q3, we still didn't have the large orders that affect more of the geography on the income statement than anything else.

Okay. That's helpful. And then just shifting to the gross margin impact from Splunk cloud, as you noted as maybe a headwind overall gross margins. And I know it's included in maintenance services although we don't have that individual line items, but can you maybe just help us understand how much it drags on overall gross margins and then how it scales over time particularly as it's built on AWS?

Sure. What I quantified is it's a 4% to 6% drag on gross margin, the cloud business in terms of overall gross margins. So again, overall gross margins were just over 84% this quarter. And if you look at the trend line on gross margins over say the last you know four to eight quarters you can see what is effectively the impact of cloud specifically. So gross margins went from call it 90% overall down to just over 84% this quarter and we've got a 4% to 6% drag from cloud. What's interesting about that is you we again I said in the prepared remarks, we expect a similar drag next year. So you say well geez is actually getting greater scale and you're building out on AWS so would you have improving margins? And the answer is absolutely, we are. But as cloud as we look at cloud is a greater percentage of overall revenue next year than its margin characteristics will have a greater impact. So, the actual drag on margins looked to be about the same year over year. But we've got a lot of really great work going on in the product organization and with our DevOps theme around scalability of cloud, efficiency of cloud. I think overall the biggest thing is just absolute critical mass of cloud customers and then to Brent's question as we get to renewing, recurring, expanding deployments of cloud customers that are similar to our historic on-prem trends we'll see better economies of scale in terms of margin characteristics.

Okay. That's helpful. Thank you.

Thanks Melissa.

Thanks Melissa.

Thank you. Our next question comes from the line Brian White from Drexel. Your question please.

I am wondering if you talk a little bit about how you've been able to execute in this cloud transition. There's a lot of companies that are going through a cloud transition now. What it means is they are coming up short on revenue or not beating as much of the revenue, their license revenues are sloppy, so maybe talk about what you're doing because I think tonight's numbers are pretty amazing. And also touch on the security front maybe just an update on what percentage of businesses is and some of the trends you've seen especially after what we saw during the election? Thanks.

Okay. Thanks Brian. I'll start with security first. It was the same composition. We've seen security and IT ops both be roughly 40% and either of use cases sold in a given quarter, 20% being some combination of IoT and business analytics and its right in that same range again. We are seeing more and more traction within our accounts being turned to as a true SIEM provider, but as you said in past calls there's as much incident of us interoperating with and supporting an existing SIEM as or is customers looking at Splunk to be their primary SIEM. So, that openness and integration is important and what really comes into play as security analytics capability to support a security incident event management SOX type framework. And David I'll let you.

Yep. Hey, Brian. If you think about our transition and the introduction of cloud and how is that feather into our financial results. One journey that we've been on with you guys for a long time is this, the ratable mix and it's grown from what was four years ago in the 10% to 20% range it is now consistently between the 40% and 50% range on a quarterly basis. And that's both the combination of adoption transactions and cloud transactions/ Now, what we know is this is a very large market, Doug talked about it. Customers absolutely are thrilled with our solutions and we've been focused for a long time around adoption. So our objective is to make sure Splunk is where your data is and where your workloads are. So one thing that it's hard for us to quantify but I'll tell you from a qualifying perspective, customers continue to expand their deployments many of them are now moving to cloud which what might have been previously on-prem. So that's where and Melissa asked this question earlier about the geography of license revenue versus non-license revenue. I think that's been the more challenging piece for us just in terms of our own forecast ability is where will the revenue land versus how much is going to be ratable or not ratable. We've given guidance on the ratable mix, so I think I've missed it as many times as I've made it, because it is so variable based on large transactions, but that as a customer going to put their next workload in the clouds or they're going to put it behind the firewall that one is a bit more challenging for us. But I think overall revenue results we contemplate the introduction of cloud in our models and we have that going forward.

And Dave we should stick to this 40%, 45% ratable mix for the foreseeable future?

Yes. We're going to give you guys the greater insight at Analyst Day in terms of how we're thinking about transaction composition, what does that mean for billings, what does it mean for cash flow and all the important metrics that you guys see in the financial statements. And I can tell you we're working on. I think we've had the right metrics to help you guys, measure us and have inputs to your model and we're now looking at should we be enhancing those retiring some, adding some new ones that give you greater insight and of course when you got this hybrid model ultimately in the way you cut through all the revenue geographies with billings. So we're looking forward to talking to you guys at Analyst Day and try to give you better or I'd say a next-gen metrics for how to think about us.

Great. Thank you.

Thank you.

Thank you. Our next question comes from the line of Kash Rangan from Bank of America Merrill Lynch. Your question please.

Hi. Thank you very much guys. Dave, just to understand the implications of the cloud impact on margins, if gross margins going to go down another 300 basis points next year, how should we look at operating margin inflection? Are we still on track for getting what you had previously implied probably 200, 300 basis points of margin expansion from this year the next year or is that dynamic change in the short term as you have the preponderance of cloud on your business model? And then I have a follow-up question. Thank you.

Hey, Kash. Let me clarify for everybody. This year we expect and measure the cloud impact on gross margin to be a drag of about 4% to 6%. Next year we think it's also 4% to 6% not an incremental 4% to 6%. So the drag is going to be consistent year-over-year.

So next year's gross margins will be roughly similar to this year's gross margins?

That's right.

Okay. Got it.

And the impact from cloud is going to be about the same. Cloud's drag on 4% to 6%, yes.

My heart rate went from 120, 240.

No, no.

I'm sorry. It went from 89 to 84 or something.

There we go. Perfect.

Okay. That $20 million quantification that you pointed out is not a slippage amount rate, it's just that you had a bit of a tougher comp by about $20 million or we trying to imply that some of these larger would could have been potential license orders are entering the cloud pipeline. I don't know how to interpret exactly what you've said?

I think that -- so let's just start with. We're really happy with our large order growth. We did 40% year-over-year large order growth. The fact is that the overall average order size was $0.5 million smaller this year than it was last and that translates to about $20 million of potential billings. It's not -- the impact of cloud is a different contributor in terms of geography between license revenue, non-licensed revenue and what's in deferred versus what is, and as Doug mentioned and you guys know a typical cloud order will be smaller than a typical perpetual order. So I wouldn't characterize the large order size differential as anything other than it's a tough comp. Deal slippage, there are always transactions that you have in your pipeline that don't close in the quarter that you forecast them. But the $20 million that we articulating is not that, it's really a tough comp against a subset of really large orders last year that makes the overall average order size smaller.

No need to apologize guys. Great results 40% licensed growth through is fantastic, also 37% billings.

Kash thanks so much.

Right. Thanks.

Thanks Kash.

Thank you. Our next question comes from the line of Jesse Hulsing from Goldman Sachs.

Thank you, guys. A question maybe for Doug, I'm curious how the adoption of EAA are trending versus your expectations and maybe have a how they were trending in the in the first half of the year? And along those lines are you doing anything to incentivize your customers or your sales reps to sell or adopt EAA at a faster rate?

Thanks Jesse. Yes. I think that 40% growth and seven-figure transactions is those scope of deals are generally tightly clustered with adoption-oriented transactions. I did make a very explicit statement at dotcom to the entire audience which was 4500 of some of our best customers highlighting the unlimited term opportunity for those customers because I was not a 100% confident that our average rep was bringing that opportunity to our customer the way that we would like them to. And it's amazing the correlation that 30-plus customers that I've gone out visited since comp literally everyone has asked me questions about how they get some type of an unlimited fine vehicle, how they really go wall-to-wall with Splunk and not make the data volume and issue they worry about. So I think rolling that program that was important. I think making sure they gets communicated to our accounts is key and I think that 40% increase in seven-figure deals is an indicator that more of it is leaking to accounts, but we've got to keep that cadence very, very strong, so that pricing and pricing variability at least on the upper end when you're really trying to use Splunk the way supposed to be used with tens to hundreds of data sources and tens of thousands of use cases on top of that single logical index that we really shine through with that unique differential power there.

Got you. And then a quick follow-up for Dave, did you -- last quarter that you said over a 100 million of your billings came from cloud. Did you break out what it was this quarter?

Now just to clarify the last year I'm sorry last quarter we achieved for the first time an annualized billings rate for cloud of over $100 million. So it was an annualized number, so that was a milestone. And it was similar in Q3.

The run rate.

The run rate, yes.

Thanks guys.

Thanks, Jesse.

Thank you. Our next question comes from a line of Matt Hedberg from RBC Capital Markets. Your question please.

Thanks. This is actually Matt Swanson on for Matt. Congratulations on the quarter. You talked a little bit about in the next couple of years being able to scale that international revenue as a percentage of the total up into that 30% to 35% range. Just in the context of what we talked about earlier in the call around the sales force coverage, where does this rank as a priority and how important is it to the long-term growth?

Hey, Matt it's Doug. How important is salesforce coverage to long-term growth.

And I guess in the international market specifically how much of focus investment is up?

Got it. Yes, okay. When I go back to the market opportunity in front of us, the data that we managed is exploding and the love that the customers that we touch have for our product and that love is dependent upon them understanding hey, I got to put hundreds or thousands of data sources in here so I can drive all those used cases. That account coverage dilemma comes back to how do you spend enough time with your current accounts and spent enough time on net new non accounts so that you can cover the market. And if I look at the United States 18 of our 52 states have no rep -- 50 states sorry I was counting a few. Few other territories -- my wife is Canadian and that was that option, but that's half of our 50 states, 18 of our 50 states don't have reps that -- that are covered by reps they don't actually live in the states. So just a density even with the United States is not where like it to be the same thing on a country basis we've been disciplined at trying to put critical mass of reps into the countries that we currently have footprint. And so from a criticality to our long-term success continued to drive coverage both our own reps, partner, resources and online interactions is an absolutely key component to driving that value of Splunk to meet the opportunity.

Hey Matt its Dave. I'll just anecdotally tell you that based on a lot of conversations I had with Susan. She would argue our focus isn't high enough on the coverage.

She would like more.

She would like more, that's for sure.

Yes. And then kind of just building off that same topic, you talk about the strength with Accenture, wondering if you kind of just comment on how SI are driving the growth and how much that can supplement some of the coverage?

I think what we have seen traditionally and here at SI is that they are really helpful and helping you succeed within accounts where you have rep coverage and they're not as powerful as far as opening doors and helping drive transactions across accounts without rep coverage. So really, really important partners for us to have and Accenture in particular we keep having them is not only are they building practice round Splunk but they are baking Splunk into a series of their solutions that they rolled directly to market. So that is a service that where they can help us more specifically with accounts we may not have reps, but that's where augmenting the SI strategy and a technology partner strategy with companies like AWS, Cisco, EMC with the classic distributor and VAR architecture is important is ultimately what we all would like to see at Splunk is more partner resources being able to do everything from the initial door knock, chat response, phone call, all the way through to implementation or at least contract delivery.

All right. Thanks for the time, guys.

Thank you, Matt.

Thank you. Our next question comes from the line of Keith Bachman from Bank of Montreal. Your question please.

Hi. Thank you very much for taking the question. I wanted to ask you about cloud gross margins, and first off thanks for clarifying your comments for the impact next year. If you look at the dollar run rate of billings it's still a fairly meaningful impact gross margins this year and if cloud is growing next year and the impacts are same then presumably your cloud gross margins are improving next year and I just like to hear A, is that the right way to think about it. And B, what's needed to get the cloud margins up particularly since you're leveraging the fixed infrastructure of AWS I would think would be a fairly steep ramp, but I wanted to hear your views on what's necessary to improve the cloud gross margins over time?

Yes. Hey Keith it's Dave. Your interpretation is correct which is the margin characteristics will be consistent year-over-year and that by definition or mathematically means that the margins must improve next year on cloud because it's going to be a larger component of the overall business so that is absolutely accurate. I think one of the key elements is really just scale and therefore economies of scale. So the way I describe our cloud business is it's like creating a company inside the company going from 10 years almost 10 years as an exclusively on-prem software provider is actually might my view more difficult from a execution perspective and a cost structure in perspective than if we just started from inception as a as a cloud only company. So as you know we continue to have great success. We're over-delivering on our own plans in terms of cloud adoption and it's interesting because a lot of folks and said, while so new customers are coming to us via cloud. And the answer is that's true but actually half of our cloud business, cloud orders are coming from on-prem customers because guess what? Everyone is moving workloads to the cloud so our ability to deliver data analytics regardless of whether your information of your workloads is on-prem or in the cloud is a huge differentiator for us because we're the only ones out there that can do it.

And just to emphasize and is it about both Brian White and Kash's question as well is data has gravity, right there's a whole bunch of physics around it and it's really risky to move data around if you don't need to but it's also really expensive. So the any good CIO or technologist want the data indexing and storage and retrieval source be as close as the generation to have that data sources possible which is why we predict that will always be a hybrid company and there always be some percentage on-premise some percentage on cloud, who knows it's 50/50 or 40/60 or but I'll won't see it swing into 90/10 at any point in time. The improvements that we are continuing to drive had taken a product and continuing to tune it for AWS or for cloud architecture. AWS is a pioneer but to other clouds regulating all the great work that they have done and some the on-prem cloud architectures cloud foundry et cetera are mimicking those as well. So tuning the products that actually are able to be more microservices containerized and a last kind of approach. so it consumes it's less expensive to run and continuous to run that product with a higher and higher degree of automation so that you don't have to put people in the middle of any process that you're doing with that cloud offering is the core to drive our cloud margins down but then conversely in a hybrid world it dramatically helps our on-prem customers because our CIO Sarah got the exact same problems that any cloud vendor does as well.

Got it. Okay. I was just going to sneak in. Thank you very much for that answer, Dave. Any comments or any parameters you'd like us to think about next year for CapEx as we're looking at our models to tune a bit.

Yes. We'll obviously we expect CapEx to be significantly lower than it has been the last two years and I can deliver on these Bay Area facility build-outs. We're going to get more color at Analyst Day about how to model CapEx for next year, but again I think about it more from a prior to this fiscal year the prior fiscal year we were in the call it mid upper teens in annual CapEx and that was on the site, all driven by employee headcount so his employee headcount goes up of course I think that's going to ratchet up an absolute dollars but it's going to be notably lower than the last few years.

Okay. Many thanks, guys.

Thank you, Keith.

Thank you. Our next question comes from the line of Ed Maguire from CLS. Your question please.

Hi. Good afternoon. I was wondering if you could comment on the progress in the market groups and also whether you could provide any color in terms of what you're seeing whether customers are using Splunk to displace existing applications or what the mix of custom applications is that you that you're seeing?

Hey, Ed. I thinking in that one of those other questions that was I think my Brain again in the cloud part of our success I think comes back to our more solution centric approach while ultimately we get this opportunity this horizontal data layer for our accounts the that focused approach on going after buying centers, security, IT operations, AppDev etcetera I think is a huge help for our customers to focus their attention on where they can gain value and then the development of our premium solutions enterprise security as a SIEM like product on top of Splunk enterprise, ITSI as a next-generation analytics driven systems management application on top of Splunk enterprise and UVA [ph] as an insider threat, application on top of Splunk enterprise are really powerful elements as well to help our customers gain value much more quickly. So that solutions orientation I think is key and we get to Analyst Day we'll definitely spend more time at walking through with you and the others or anyone that attends how we can continue a packaging pricing and promotions approach to ensure that we are allowing our reps and our partners and the entire community to be able to go after these specific buying centers and use cases on a easy clean affordable basis. On the ES as a SIEM yes that's increasingly being viewed that way and there are more situations where in a direct compete against existing SIEM vendors time. With ITSI either there's a lot of communications a lot of back-and-forth between us and customers on what tools going to retire. The delta between our Splunk solutions and those existing solutions is the flexibility of our platform and the heavy adaptability and analytics orientation what we do, so it's rarely ever a one-for-one replacement. It is a -- you're going to get a whole different set of value when you use us as a SIEM or as a systems management technology which for most customers means there's a migration over time to be able to retire this other tools, but we do we do help our customers with tool bake-off just to see if there's a cost-saving as they roll in Splunk on a more immediate business.

Yes. And just a quick follow-up on competition whether you've seen any change at all in the competitive environment or new competition that you're that you're working against?

It remains extremely similar to what we've seen every other quarter.

Right. Okay. Thank you.

Thanks Ed.

Thanks, Ed.

Thank you. Our next question comes from the line of Alex Sukin from Piper Jaffray. Your question, please.

This is Taylor Reiners on for Alex. I was wondering we've heard anecdotally that you've been a little more flexible around both licensed pricing and data limits and wondering if you know if you've been seeing customers spend more on layering apps on top of that with higher data limits. And then I know last quarter to mention that 70% of license revenues came from existing customers, wondering if you're seeing a similar dynamic this quarter? Thanks.

Thanks Taylor. Yes, this quarter we we've consistently scene and all the course I've been here that this adoption momentum within our current account, so that 70% or more has been a very consistent factor and I think every call that we've had, its consistent again this quarter. One of those announcements that I made at Dotcom was which we've talked about in the prepared remarks was the removal of are metered enforcement where we have a license engine that actually looked at data going in. And part of taking out that metered enforcement was to re-emphasize to our customers we know that you want to put more data into Splunk. We want to be able to experiment. We still want you to comply with the overall license, but we're going to – we'll take out any guardrails that will momentarily stopped you from putting more data than you want to -- than you initially license in Splunk. And that's part of this theme of that we've seen with every customer that the cohort shows over and over which is the more data that customers get into Splunk and the more solutions and use cases that they deploy on top, the more excited to get about the product and more value they see and therefore the more data to put in. And when you combined that with all the hyperbole around the exponential growth and machine data and large data volumes, but when you're dealing with a core market that the driver is exploding then we just want to make it as easy -- we have full confidence in our products ability to meet the needs of our customers and serve more use cases. We want to make it as easy as possible for them to keep getting more data and they get more value out.

Excellent. Thank you.

Thank you.

Thank you. Our next question comes from the line of Steve Ashley from Robert W Baird. Your question please.

Hi. This is Jason Velkarvh for Steve. Thanks for taking my questions. Actually, I had a question about, Doug you mentioned in prepared remarks Insight Engines natural process -- natural language processing and I think Tables has also unveiled at comp as another ease-of-use improvement. I'm just curious is when spreading adoption to business analytics particularly is ease-of-use a barrier to adoption there more so that it would be for security or IT Ops and have you see, I know it's early any I guess reduced evidence to adopt one for business analytics sense these new features have been added?

So, thanks Jason and yes, both of those were super exciting. I love Grant's demo and the power of the solutions put in place is phenomenal and I think the Tables in a bunch of the other data analysis and prep stuff that we are doing is helpful. And it's interesting it's helpful both for existing use case proponents. It is hard for any organization to get enough security expertise, IT Ops expertise to actually be effective. So more that we can do to lower the barrier of expertise in those areas, in those quadrants helps even in our natural use cases. And then for sure as you get beyond are more repetitive use cases with the ES and ITSI there's a single app that can be deployed across thousands of customers. We've been out there working with customers that find that same, what's one app that you can solve a thousand instances with. We're getting better with some ideas around support, helping customer support, helping retail optimization, helping asset management from outside of our traditional market, so we are continuing our experimentation proof-of-concept and build out of different components in that business analytics and IoT arena. But given that there's a lot more touch of non-technical users with this data as you expand beyond IT those enhancements and advancements of things like natural language interaction easier data surfing and traversing, easier data prep and understanding I think become important just like we've seen in the track classic BI category of trying to make those tools more approachable to less technical users.

Great. Thanks. That's helpful. And then just one quick follow-up, apology if you already mentioned in this call, I think last quarter you said roughly half and half for our cloud sales to new and existing customers, is it a roughly the same this quarter?

It is. Yes. Existing customers represent about half of the cloud transactions and half the cloud transactions are new.

Great. Thanks guys. Thanks, Jason.

Thank you. Our next question comes from the line of Greg McDowell from JMP Securities. Your question please.

Thank you very much. Just one quick cash flow from operations question, I know you mentioned that this year you are now expecting 21% of revenue for cash flow from operations. And I was just wondering after being pretty consistent for the previous four fiscal years that sort of that 23% to 24% range. What exactly has changed on that front? Is it more related to the cloud, or is there something else going on the cash flow side? Thanks.

Yes, hey Greg, it's really you know we're just looking at the linearity of the quarters, and we've talked a lot about the growth and large orders, at what we saw year-over-year from Q3 of last year to Q3 of this year. And then, again where we've got our granular view on the fourth quarter, so when we put all the beans in the calculator, and think about revenue trajectory and how are we delivering on revenue relative to expectations, and then when do we expect some of these transactions to close and therefore when do you expect to collect them. That, those were the inputs to say, you know let's give a different view for the balance of the year.

Great. And one quick follow up, at the Splunk comps, Doug it sounded like you guys were doing a pretty heavy push on, EAAs not just for your -- you're very, very large customer. So I was just wondering if you could comment, is there sort of a minimum line at which you don't want to do EAAs or sort -- you're sort of lowering the bar on the side of the customer in which you want to execute these EAAs if you could just comment on that, that’s great. Thanks.

That’s a good question, Greg. There is not -- they are not tailor just large customers. We've got a couple of customers that are sub a billion dollars in revenue, where we have an term unlimited framework in place for them. That is priced differently than for the larger organizations; because we get that unlimited term structure, it becomes more factor of total revenue employee count as well as data volume. And when I look at how important big data is as a topic, and how critical it is to competitive posture, and agility of companies going forward. There's the same kind of entry-level need, department-wide need, and enterprise-wide need at small medium and large organizations, and then that's going back to focus on that new as well as focus on our existing accounts and being able to get enough market coverage. The opportunities at every business ultimately is going to become more software and data driven as a business, those opportunities all exist for a company like Splunk and we've got to be aggressive and in pursuing those on a more thorough basis.

Thank you.

Thanks very much.

Thank you. Our final question comes from the line of Karl Keirstead from Deutsche Bank. Your question please.

Thanks. I will finish up on a positive note, than, So Doug, on the license performance you guys talked a lot on this call about the ASP decline and tough compare. It almost sounded like you were explaining a miss but on your license side it easily beat your implied guide. License growth was -- it accelerated compared to last quarter, despite a tougher comp. Maybe I will finish up by just asking you where you think the one or two areas of outperformance were on the license side, whether it is by use case, vertical, geo, what surprised you? Thank you.

Thanks, Karl. So I’d say by far what we continue to be excited about is that the -- cloud exceeding our expectations. It's -- it's important for our customers, it's important for us, the long-term value with our cloud customers winds up being a benefit to us as well as the speed of adoption and speed of value our customers get, so it's really nice to continue traction with clouds and our need to make continued investments there so that we can meet our customers’ expectations. And then despite the lower average order size, the 40% growth and seven-figure transactions and it was another really nice thing to see that that was a -- a good in that increase versus what we see in other quarters, and even though the average order size was lower most because of these mega deals, I think that’s part of that trend that we are looking for of these EAAs and customers using us as this broad-based horizontal platform with lots of data sources and lots of use cases. And again, when I look at us versus any other technology that's out there, that that's where we just continue to shine. If you’ve got a simple use case, a few data sources, their more structured nature, you can ask the same question many times, there's lots of technologies you can use. But that's not the nature of big data. Big data is messy, some of its structured, some of it isn't, there's a lot of lots of sources that are only growing and the possible use cases are only limited by your imagination, and so whatever we can do to help our customers understand what they can do with big data is a really along with cloud, those are -- there are two, those are two exciting things for us.

Okay, good. Congrats.

Thanks, Karl.

Karl, I want to pile on there. One thing that I was really excited about in the quarter. And you guys have heard us talk about EAAs for a long time, but translation of that familiarity to customers hasn't been nearly as great. So Doug was very explicit to announce that our comp, that we've got this program and now in the customer visits the awareness that there's a vehicle to get broad adoption, like that’s really exciting to me. So as we think about number of EAAs, it's and the number of our customers, are we going to standardize on Splunk right you know that's really exciting, I'm really really looking forward to it.

Okay, good thanks guys, congrats.

Thanks, Karl.

Thanks, Karl.

Thank you. This does conclude the question and answer session of today’s program. I’d like to hand the program back to Ken Tinsley for any further remarks.

Great. Thanks Jonathan, I really appreciate your help today. And thanks everybody for joining. Hope everyone has a wonderful holiday season and we look forward to seeing you at the Analyst Meeting in January. Thanks very much.

Thank you ladies and gentlemen for your participation in today's conference. This does conclude the program. You may now disconnect. Good day.