MongoDB Inc (MDB) 2019 Q3 法說會逐字稿

Good day, everyone, and welcome to the MongoDB Third Quarter Fiscal 2019 Earnings Call. Today's call is being recorded.

At this time, I would like to turn the call over to Brian Denyeau, ICR, Investor Relations. Please go ahead, sir.

Thank you, Anne. Good afternoon, and thank you for joining us today to review MongoDB's third quarter fiscal 2019 financial results, which we announced in our press release issued after the close of the market today. Joining me on the call today are Dev Ittycheria, President and CEO of MongoDB; and Michael Gordon, MongoDB's CFO.

During this call, we may make statements related to our business that are forward-looking under Federal Securities Laws. These statements are made pursuant to the safe harbor provisions of the Private Securities Litigation Reform Act of 1995, including statements related to our financial guidance for the fourth quarter and full year fiscal 2019; our market opportunity; our go-to-market and growth strategies; the potential advantages of our new products and enhancements; our ability to expand our leadership position and drive revenue growth; and the anticipated benefits from our acquisition of mLab.

The words anticipate, continue, estimate, expect, intend, will and similar expressions are intended to identify forward-looking statements or similar indications of future expectations. These statements reflect our views only as of today and should not be reflected upon as representing our views as of any subsequent date. We do not have plans to update these statements except as required by law. These statements are subject to a variety of risks and uncertainties that could cause actual results to differ materially from expectations.

For a discussion of material risks and other important factors that could affect our actual results, please refer to those contained in our annual report on Form 10-K filed with the SEC on March 30, 2018, our quarterly report on Form 10-Q filed on September 7, 2018, and our other periodic filings with the SEC. These documents are available in the Investor Relations section of our website at www.mongodb.com. A replay of this call will also be available there for a limited time.

Additionally, non-GAAP financial measures will be discussed in this conference call. Please refer to the tables in our earnings release on the Investor Relations portion of our website for a reconciliation of these measures to the most directly comparable GAAP financial measure.

And with that, I'd like to turn the call over to Dev.

Thanks, Brian, and thanks to all of you for joining us today to review our third quarter results. We delivered terrific performance in third quarter across all aspects of our business. Our ability to generate best-in-class growth at significant scale, which now exceeds $0.25 billion revenue run rate, reflects our strengthening position as the next-generation database platform of choice.

We serve one of the largest markets in enterprise software. IDC now forecasts that the database market is $59 billion in size in 2018, growing to $84 billion in 2022, reflecting a 9% compounded annual growth rate. We believe MongoDB is uniquely positioned to capture growing share of this market given the power of MongoDB's sophisticated document model, large developer mindshare and compelling customer proof points across nearly every industry and geography.

To quickly summarize our third quarter results, we generated revenue of $65 million, a 57% year-over-year increase, which was above the high end of our guidance. We grew subscription revenue 59% year-over-year. Atlas revenue grew more than 300% year-over-year and now represents 22% of revenue. We ended the quarter with over 8,300 customers, up 69% compared to a year ago and we saw strong growth across all major regions including North America, EMEA and Asia. I do want to note that our results do not include any impact from the acquisition of mLab which closed on November 1.

Our success of customers is based on the strength of our database platform. At our recent MongoDB Europe event, we announced the general availability of MongoDB Mobile, which gives developers a uniform way to work with data wherever the data resides. MongoDB Mobile allows developers to build faster, more responsive applications that can take advantage of the document model for data held on a mobile device. Previously, developers were stuck using a feature-limited tablet database for a mobile app.

We also included several updates to MongoDB Stitch, our serverless platform, including the MongoDB Stitch SDK, which allows data access regardless of whether it is held on the mobile client or the back end. This allows developers to create mobile and IoT data any way they want with local reads and writes, native JSON storage, indexing and aggregations. With Stitch Mobile Sync now available in beta, you just can automatically synchronize changes between data held on an end device and a back end database even after the device has been offline.

Expanded integration options including more than 20 AWS services and additional authentication providers, which can all be exploited without learning new APIs or adding back end code. Last week, we also introduced updates to MongoDB Atlas including high throughput clusters in AWS that will enhance realtime workload processing to accelerate new and existing applications. These clusters are optimized for low latency, very high I/O performance for large data sets. Support for cross-region VPC peering, which allows Atlas to communicate with a customers' virtual private cloud across regions via private networking. This will enable customers to connect data between disparate database nodes without utilizing the public Internet, significantly reducing security risks and operational work overhead and improving performance.

We also introduced the Server Side Public License or SSPL, a new licensing model for open source development in the cloud era. We are in a unique position to decide how our software is licensed because we are full copyright holders of all underlying MongoDB database code. We created this new license for Community Server, our free-to-download offering to make explicit the conditions for providing MongoDB as a service. We are strong believers in open source and believe it leads to more valuable, robust and secure software. By introducing SSPL, we have given the community a new licensing model to build on the spirit of AGPL.

We will continue to provide the same freedom to the open source community to use, review, modify or redistribute the software. The only substantive change is an explicit condition that any organization attempting to offer MongoDB as a third party service must open-source the software it uses to offer such a service.

Turning to our third quarter results in more detail. We again saw a healthy mix of new wins and strong upsell activity including customers migrating existing workloads from legacy databases to MongoDB. Our performance in the quarter exemplified the power of our land, expand and expand model and our multiple vectors for growth.

We saw success across-the-board. A number of exciting customer wins came from outside the United States, including countries where we are either beginning to establish a direct sales presence or where we have partners that are expanding our market reach. We're also seeing a growing number of wins with customers in traditionally conservative industries and markets, reflecting the growing popularity and mass appeal of MongoDB. We believe both of these trends demonstrate the mainstream adoption of MongoDB and growing recognition by the industry that we are the modern database platform of choice. We also believe that illustrates the strong result returns we're getting on our sales and marketing investments.

I would like to spend a few minutes reviewing the details of some key wins in the quarter. An important indication of MongoDB's strength and positioning within the ISV ecosystem is an agreement we signed this quarter with SAP. This agreement will enable our joint customers to rapidly build and deploy new applications, leveraging MongoDB within the SAP Cloud Platform, its enterprise platform as a service offering.

Her Majesty's Revenue & Customs, or HMRC, the U.K.'s tax payments and customs authority, has increased its use of MongoDB this past quarter. HMRC has enhanced its digital platform by using MongoDB as the core database and MongoDB Atlas is helping to further improve the experience of citizens, who submit tax self-assessments online. We signed one of our first Atlas enterprise transactions in South Korea. Rainist, a leading financial asset management mobile application company achieved increasing developer velocity and flexible support for the sophisticated financial requirements in a secure environment by using MongoDB Atlas enterprise.

We signed our first Atlas win in Argentina with a large vertically integrated energy company that wanted to migrate its existing logistics platform from a relational database to a technology that enabled their multi-cloud strategy without having to rewrite a line of code. The Maryland Health Exchange, which provides different insurance plans for residents, chose MongoDB for complex requirements like interfacing with multiple government agencies and being able to quickly determine if the consumer is eligible for a subsidy.

One of the largest banks in China chose MongoDB to build a new loan processing platform, which is expected to process thousands of loan approvals daily and significantly reduce loan processing times. Michelin, a global tire manufacturer, chose MongoDB Atlas to serve as a single highly performing database to manage the content of hundreds of websites in 70 countries around the world, replacing their legacy database. Realtime replication and continuous backups were required to deliver a compelling user experience in every region.

A large Italian government institution chose MongoDB to create a single view of their customers as part of its digital transformation initiative. Using MongoDB, they were able to bring together multiple previously siloed business units to create a holistic view of their customers and unlock new opportunities to better serve their constituents.

One of Europe's largest manufacturers and distributors of gas and electricity recently moved critical billing data from multiple services from Oracle to MongoDB. This allowed for significant performance improvements, lower TCL and enabled the organization to capture new business more quickly. Oasi, a leading Italian anti-money laundering and supervisory reporting software company replatformed their applications from legacy relational databases to MongoDB to quickly identify anomaly indicators for suspicious transactions and to improve the performance and scalability of the overall application.

Buffer, a social media management platform that consolidates social media publishing and engagement in realtime, moved multiple applications to MongoDB Atlas to significantly improve performance by eliminating queued reads and writes and dramatically reducing disk storage requirements.

These wins were customers of various sizes across different industries and countries but all were driven to use MongoDB rather than alternative database technologies due to a pressing need to innovate more quickly to better serve their customers and to take advantage of new business opportunities. These wins showcase the growing acceptance of MongoDB as a general-purpose database and demonstrate how we're winning in a database market forecasted to be $84 billion by 2022.

To further extend our leadership in the cloud, we recently acquired mLab, a pioneer in the database-as-a-service market since its founding 7 years ago. mLab has been particularly successful focusing on developer-centric start-up communities via self-serve model. We're excited to bring that self-serve and database-as-a-service expertise into MongoDB and believe mLab is an efficient way to capture a large number of customers. We will shortly begin the process of moving mLab customers to Atlas.

Before I wrap up, I want to provide a quick update on our leadership team. I'm pleased to announce that we have appointed Michael Gordon to the newly created role of Chief Operating Officer, which will be in addition to his current CFO responsibilities. This well-earned promotion reflects Michael's significant contributions and active role across all aspects of the company.

To wrap up, we delivered outstanding results in the third quarter. We are seeing increased market momentum as a growing number of customers recognize the value MongoDB can enable for their businesses. Our products have never been stronger and our efforts to establish MongoDB as a strategic vendor for enterprises are paying off. We are well positioned to carry this momentum into the fourth quarter and beyond.

With that, let me turn the call over to Michael to review the financials.

Thanks, Dev. As mentioned, we're very pleased with our third quarter performance, which reflects strong growth at significant scale. I'll begin with the detailed review of our third quarter results and then finish with our outlook for the fourth quarter and full fiscal year 2019.

Total revenue in the quarter was $65 million, up 57% year-over-year. Subscription revenue was $60.1 million, up 59% year-over-year and professional services revenue was $4.9 million, up 36% year-over-year. The strong performance of the business was broad based during the quarter, with healthy new logo and upsell activity among enterprise customers as well as continued rapid adoption of Atlas.

Atlas represented 22% of revenue during the quarter, up from 8% in the third quarter last year and up from 18% last quarter. As a reminder, with the introduction of Enterprise Atlas -- Enterprise Advanced features into Atlas, we expect the product distinctions between the 2 will blur. Instead, customers will choose between -- based on which consumption mode best suits their needs.

We continue to see strong global demand for our offerings. During the third quarter, we grew our customer base by approximately 900 customers, bringing our total customer count to over 8,300 which is up from over 4,900 in the year ago period and over 7,400 at the end of last quarter. Of our total customer count, over 1,700 are direct sales customers, which compares to over 1,600 at the end of the prior quarter and over 1,400 in the year ago period. The growth in our total customer count is being driven in large part by Atlas, which had over 6,200 customers at the end of the quarter compared to over 5,300 at the end of the second quarter.

The growth in total customers includes growth in our Enterprise Advanced customers as well as new Atlas customers. It is important to keep in mind that the growth in our Atlas customer count reflects both new customers to MongoDB as well as existing Enterprise Advanced customers adding incremental Atlas workloads.

We also continue to see healthy expansion from existing customers. Our net ARR expansion rate in the third quarter remained above 120% for the 15th consecutive quarter. We ended the quarter with 490 customers with at least $100,000 in annual recurring revenue and annualized MRR, which is up from 438 in the second quarter and 320 in the year ago period. Driving expanded adoption and spend among existing customers is a key component of our growth strategy and has been a consistent area of success. We believe MongoDB comprises a small fraction of the total database spend of most of our largest customers, which represents a large and attractive market opportunity for us over time.

Moving down the P&L, I will be discussing our results in a non-GAAP basis unless otherwise noted. Gross profit in the third quarter was $48.1 million, representing a gross margin of 74%, which was consistent with the year ago period. We saw gross margin improvement on a quarter-over-quarter basis for Atlas, for subscription revenue not related to Atlas and for professional services. We are pleased with the gross margin performance in the quarter, particularly in light of Atlas's growth. Atlas includes the underlying infrastructure and our success in reducing the infrastructure and support costs related to Atlas that resulted in improved Atlas gross margins. However, we continue to expect we will see some modest reduction in overall margin as Atlas continues to become a bigger portion of our revenue.

Our operating loss was $16.8 million or negative 26% operating margin for the third quarter compared to a negative 44% margin in the year ago period. The more than 1,800 basis point improvement in operating margin indicates the operating leverage we are driving in the business while investing in our continued growth. Net loss in the third quarter was $16 million or $0.30 per share based on 52.7 million weighted average shares outstanding.

Turning to the balance sheet and cash flow. We ended the quarter with $522.7 million in cash, cash equivalents, short-term investments and restricted cash. Subsequent to the end of the quarter, we closed on the acquisition of mLab for approximately $68 million in net cash consideration.

Operating cash flow in the third quarter was negative $7.6 million. After taking into consideration approximately $2.1 million in capital expenditures, free cash flow is negative $9.7 million in the quarter.

Short-term deferred revenue is $136.6 million, up 48% year-over-year, while total deferred revenue of $153.8 million was up 34% year-over-year. Keep in mind that Atlas is a usage-based model and generates less deferred revenue than Enterprise Advanced. Also as a reminder, Atlas is often billed monthly in arrears versus the annual and advanced billing terms we see in our Enterprise Advanced customers. Lastly, quarter-to-quarter comparisons of deferred revenue can also have some level of variability due to timing.

Before I turn to guidance, I would like to review the financial impact of the mLab acquisition, which closed on November 1. As Dev mentioned, mLab substantially enhances our self-service knowledge and provides additional database-as-a-service expertise. From a revenue perspective, our current expectation is mLab will generate approximately $5 million of revenue in the fourth quarter and approximately $18 million of revenue in fiscal 2020.

From a profitability perspective, we expect mLab to be roughly breakeven in the fourth quarter and in fiscal 2020. Our intention is to migrate the mLab customer base to Atlas in the coming quarters.

Our expectation for a modest decline in revenue in fiscal 2020 is driven by 2 assumptions. First, Atlas and mLab have different pricing models. We intend to honor existing pricing for migrated mLab customers when advantageous to them. For customers for whom Atlas is priced at a discount to current mLab pricing, we'll realize lower revenue as those customers benefit from the new lower pricing package.

Second, we anticipate the potential for some customer churn due to the disruption from switching platforms and a large number of small customers in mLab's customer base. Importantly, these assumptions were part of our acquisition thesis and incorporated into our purchase price. We should also call out that mLab customers spend significantly less per year than self-serve Atlas customers have experienced to date, reflecting the breadth of appeal of offering MongoDB as a service to the long tail of the MongoDB community.

I'd now like to turn to our outlook for the fourth quarter and our full year fiscal 2019. Beginning with the fourth quarter, we expect revenue to be in the range of $73 million to $74 million. Non-GAAP loss from operations is expected to be in the range of negative $21 million to negative $20 million and non-GAAP net loss per share in the range of negative $0.39 to negative $0.38 based on 53.9 million weighted average shares outstanding. Again, as mentioned, this includes $5 million in revenue from mLab.

For the full year fiscal 2019, we are raising our revenue outlook to $243.7 million to $244.7 million. We are also improving our profitability outlook with non-GAAP loss from operations now expected to be in the range of negative $81.1 million to negative $80.1 million and non-GAAP net loss per share to be in the range of negative $1.53 to negative $1.52 per share based on 52 million weighted average shares outstanding.

Given the size of our market opportunity, the strength of our product market fit and the attractiveness of our unit economics, we believe it makes sense to continue investing in R&D and sales and marketing for the long-term benefit of the company. As a result, we will continue to invest in the business in fiscal '20 with the goal of driving continued high growth while demonstrating continued operating leverage.

In closing, MongoDB continued to perform at a very high level in the third quarter. We are successfully executing against our strategic priorities, which is driving strong growth in fiscal 2019 and beyond.

With that, we'd like to open up to questions. Operator?

(Operator Instructions) We'll take our first question from Raimo Lenschow with Barclays Capital.

This is actually David Rainville on for Raimo today. Congrats on another impressive quarter here. Maybe we'll start with Atlas, I guess that's what everybody wants to talk about and the numbers continue to be so strong. Can you give us maybe a little bit more color on the types of workload that you are seeing now versus what you were seeing a year ago? And maybe an update on where you feel you are in your journey of bringing Enterprise Advanced features on to Atlas as you're trying to play this ASP game going forward?

Sure. So just as a reminder, I mean, Atlas was introduced about 2.5 years ago. So as you can imagine, the workloads that customers first started with were probably tier 2 and tier 3 workloads and maybe even development test workloads. We're clearly seeing now large sophisticated customers moving mission-critical workloads to Atlas, and that's reflected in both the size of the deals as well as the broad-based set of use cases the customers are using for. So I think people are viewing Atlas as truly a viable platform to run mission-critical applications.

That makes sense. And maybe just a quick follow-up on SSPL. Any early customer feedback? Or feedback from the community from the changes that you've done?

In general, the feedback has been quite positive. I would say for customers, it was a nonevent because most customers do not plan to offer MongoDB as a third party service so that was really a nonissue. And I think, generally, the feedback from the community has been quite positive. There'll always be some people who don't like any change but most people recognize that given we are now in the cloud era, there has to be a way for open source companies and open source projects to thrive in the cloud era. And the SSPL, a lot of people said, is the best way to do that.

We'll go next to Sanjit Singh with Morgan Stanley.

Congrats, Michael on the promotion, well-deserved. I have 2 questions, actually, it's a two-parter on just getting update on go-to-market. So Dev, I think when you updated us last quarter, you said that you haven't seen a whole lot of contribution yet from the IBM and Accenture partnerships and that could be a driver in the second half. So I want to see if you've got an update there for us on sort of the partner ecosystem contribution. And then secondly, in terms of penetrating that long tail, you guys were up to, I think, 45 million downloads. That equals around 30 million downloads at the time of IPO, so a big increase there. Can you give us a sense on how you guys are intending to capture some of that long tail and what mLab could bring in terms of that part of the equation?

Sure. In terms of your partner question, last quarter, we did do a few deals with IBM. We feel like this is a start of a very healthy long-term partnership. Our pipeline with IBM is growing quite significantly. And while I don't want to commit to any specific numbers, we feel very good about the relationship not just here in the U.S. but actually in Europe and also in Asia, so that relationship is really working well. And I speak to the GM of the business, who's most closely tied to us on a monthly basis. So we have a pretty deep relationship up and down the organization. With the growth with Accenture, we actually -- I'm not sure I said that things were not taking off yet with Accenture. We actually are doing quite a few deals with Accenture. In fact, some of the pretty material deals we did this quarter were through partners like Accenture. And Accenture is also helping expand our reach into markets where we may not have a lot of in-market coverage. And so that relationship is going well. We feel very good about that relationship. They've actually built a bunch of products on top of MongoDB and they've enabled their people to think about using MongoDB for a variety of use cases all the way from helping customers migrate to the cloud, helping customers migrate off legacy platforms, et cetera. So we feel really good about that relationship as well. And then in terms of your -- the second part of your question was -- just remind me again was...

The long tail adoption...

The long tail.

Just the long tail, yes.

Yes, the long tail. So one of the things that really impressed us with mLab was that they built their business without hiring a single sales rep. And so they basically have to build their self-serve business. Now we also started off building a business on Atlas with self-serve but they obviously have been doing this a lot longer than we have and that expertise we're inculcating into our own organization and really expanding our self-serve business. So we believe the best way to go after the long tail is through self-serve, at least in the initial customer acquisition. And then where appropriate, we will get our sales team to follow up and help expand those customers depending on the size and scale of those customers. But the long tail will be our natural go-to-market motion for going after the long tail -- sorry, the self-serve motion will be a natural sales motion to go after the long tail.

We'll go next to Brad Reback with Stifel.

Michael, quickly on mLab, how should we think about the revenue breaking out between subscription and services?

There's really no services. Think of all $5 million as subscription and analogous to the self-service bucket. Although as we called out, it's lower spend per customer.

Got it. And then quickly, Dev, on your commentary around all these really impressive enterprise wins around the country, in the world, that is, how should we think about initial order entry from new customers today versus the size they were, maybe, 2 years ago? How much bigger can you start with the customer nowadays?

Well, we have seen our initial lands getting bigger and bigger. I think that's a function of people moving and getting comfortable moving, engaging with us with more and more mission-critical workforce even through our Enterprise Advanced solution or through Atlas. But we're also seeing very significant expansion. As Michael mentioned, this is yet another quarter where expansion rates were north of 120%. And so we don't try -- this is truly a land, expand, expand business. So we don't try and close -- go elephant hunting and try and close the entire business upfront. It makes no sense for the customer, and, frankly, we'd rather engage with them in a way that they feel very natural. It creates less friction. And once they get more and more comfortable, they start coming back and buying more. And that's great for us as a business. So that's the typical way we engage with customers. But customers, our lands are getting bigger and bigger as customers get more and more comfortable with using MongoDB for more and more critical workloads.

We'll go next to Ted Lin with Goldman Sachs.

Congrats on the new role, Michael. I guess -- so in terms of the competitive landscape against both the public cloud guys, which you noted that you have a technological advantage against, in the past, but especially kind of the new NoSQL players or the pure-play NoSQL players, any color there on win rates, who you're seeing? And then are people standardizing on a single NoSQL offering? Or are they choosing still the best database for every use case?

So obviously, there is some talk by the cloud providers about having a database, a different database for every solution. We don't think that, that makes complete sense. One, it's very hard for an organization to train, support and build applications across all these disparate databases. Moreover, you now have to spend a lot of time making sure all that data is synced across all these different databases, and so that becomes pretty expensive and time-consuming. So we do believe while there may not be one database for everything, that people want a general-purpose database to serve a majority of their use cases. And we believe we're well positioned to do that. The document model is a superset of a key-value store. It supports our document model, supports graph functionality. We support joins and the term -- a feature called $lookup. So the breadth -- and we announced transactions in the middle of this past year, so the breadth and scale of the use cases our customers are running on MongoDB is quite profound. So we believe that we're well positioned to go after that market. And that's a function of why we believe our developer mindshare is really second to none when it comes to next-generation databases. And that's referenced by all the external market data out there.

Great. And I guess as a follow-up, digging in on the SSPL versus the AGPL. I guess, what drove kind of the change in strategy there? I think you mentioned before that the AGPL kind of prevents people from running Mongo as a service. And so are you seeing demand for third parties that want to offer MongoDB as a service? And what are the implications there for Atlas?

Right. So SSPL was built on the spirit of AGPL. We believe that it was quite clear that as people wanted to offer MongoDB as a service, they couldn't do so given the licensing around AGPL. When we started seeing some international providers starting to test the lines about what was and what was not possible, and we thought it would just make sense to be -- make it very explicit, the conditions required to offer MongoDB as a service. It's not a restriction. Anyone can offer MongoDB as a service, but there are conditions that you need to conform to, to be able to do that. And we want to make that very clear. And the reaction has been incredibly positive from customers. They think it makes sense. They want to see us continue to grow and be a healthy company, from users have been very positive. And I would say, in general, the community has received the feedback quite well as you can see in some of the press reports that have come out since our announcement.

We'll go next to Brent Bracelin with KeyBanc Capital Markets.

Start out here with Dev here. mLab's, obviously, contribution, $5 million here in Q4. Have you talked to some of their larger customers? And if so, what's been kind of early response after acquiring that asset so far with some of their, maybe, larger customers?

So the short answer is yes, we have talked to some of their larger customers, and the response has been generally very positive. And as Michael mentioned, there are some certain SKUs that are more expensive than what we offer at Atlas. And in some cases, there are situations, where they have a cheaper price. So we plan to make it be very customer friendly and honor the most advantageous price for customers. So we expect that to have a little bit of pricing compression. And we also expect that there will be some natural churn because of some really tiny customers, who, just in the process of switching platforms, may decide to bring that back in-house, and so -- but in general, the feedback has been very positive. We're very excited to basically acquire the entire mLab team except 1 person. And they're super excited to be part of MongoDB. And we feel that, together, we really can build a very large database as a service offering.

Great, helpful color there. And then in the script, you mentioned conservative industries starting to kind of embrace Mongo. You also called it out here on the call. What type of conservative industries are there? And maybe just drill down into kind of the dialogue you're having there and why you're kind of calling that out this quarter.

Yes, I mean, like, one example I'll give you was HMRC which is really the IRS of the U.K. They were existing MongoDB paying customers and they recently expanded their use of MongoDB by using Atlas. So if you had told me 6 months ago that we'd have a large government entity in the U.K. using Atlas, that, frankly, would have surprised me because those organizations -- we have another large government institution in Italy decide to use us, use MongoDB to serve their constituents better. We have people in Latin America in markets like Argentina, where we don't have any of our own people there today, who are now using Atlas. We only -- we have a very small team in Korea, and all of a sudden, we have people using MongoDB pretty quickly. So we feel like what that really speaks to is the popularity of MongoDB as truly the database, the modern database platform of choice and the mass appeal for all these wide and disparate use cases. And so we feel that's a strong indicator of the opportunity we have at Atlas.

Right, got it. And then 2 quick ones for Michael, if I could, here. On deferred revenue, short-term deferred actually accelerated in the quarter, even with that less mix shift kind of which I thought would be a little bit of a headwind. What drove the acceleration in short-term deferred? And how should we kind of look at that metric versus the mix shift to Atlas which has little to no deferred revenue component?

Yes, so I think the headline trends are, obviously, it's a very strong quarter. Billings, certainly not something we guide to and has a lot of quarter-to-quarter variability. And you're absolutely correct in the commentary around how does Atlas behave. And Atlas doesn't add nearly as much from a deferred revenue perspective as compared to Enterprise Advanced. So I wouldn't go extrapolating out from the quarterly results especially as we expect Atlas, which is growing faster than the rest of the business, which is growing quite quickly, to be an even larger portion of the business. So I think it's mostly just the fact that it was a very strong quarter across the board.

Got it. And then last, on the gross margins, particularly the subscription gross margin, it looks like this was the first quarter in over a year that, that actually increased even with, again, Atlas mix shift. Walk through the drivers of improving subscription gross margins.

Yes, sure. So we saw subscription gross margins improvements both for Atlas as well as for sort of Enterprise Advanced and the rest of the non-Atlas revenue, so again, kind of strong across the board. I think the Atlas progress is probably the most notable because that's where we've been actively working. And as Atlas continues to become a larger portion of the revenue, we've been very pleased with our progress on that front. A lot of it relates to infrastructure, but we've also started to see very successful internal efforts and the initial fruits from driving lower support costs as a percent of revenue. And so that's been very valuable. I continue to think that we'll have sort of the shallow view that we've talked about because the business is growing faster than we can optimize every dial and every lever. And then, I guess, I'd also call out, in the short term, mLab's business, while small in the grand scheme of MongoDB, on a kind of run-rate basis based off of the numbers and the guidance we've given, is close to a quarter of the Atlas business. And that was at lower gross margins. And so it will take a little bit of time when we -- to integrate them and kind of harmonize those margins. So I think that will be a Q4 headwind for sure, and probably at the beginning of the year in Q1 and Q2 as we kind of work through the integration plan. And we can certainly update our outlook on that as we get out to March in the Q1 guidance, but want to call that out for folks as well just so people won't get overly carried away in extrapolating numbers and things like that.

We'll go next to Pat Walravens with JMP Securities.

First of all, on the Service Side Public License, so was mLab itself one of the vendors that were sort of testing the limits of the AGPL?

We did have a relationship with mLab in the early days. As we -- when we introduced Atlas, as you guys can imagine, that relationship terminated. So while technically mLab did not have an agreement with us to offer MongoDB as a service, that ultimately brought the 2 parties together. And I think the deal we struck was both good for mLab shareholders and good for MongoDB shareholders because it allowed us to accelerate our MongoDB as a -- our Atlas business, brought some new competencies around self-serve and allows us to, essentially, position us for long-term growth. So it's a transaction that made sense for both parties.

Okay, great, just wondering. And then did -- can people get around this just by using versions of the Community Server from before October of -- before, I guess, October 16?

So any version before our SSPL announcement was licensed under AGPL, which we still also believe has pretty broad restrictions around what you can do when offering MongoDB as a service. That being said, if someone want to really test this using AGPL, they'd have to fork the code base prior to our announcement and then have to manage that themselves, which, frankly, would be a pretty Herculean task. Secondly, the Community, they'd have a tough time getting a large following because as we add features and new capabilities into our platform, the gap between the fourth version and us would just continue to increase, so that would not be a very attractive kind of option for most people.

All right. And then this is a totally different subject. So going back in history, as I look at a really old model, I think Oracle went from roughly $320 million revenue to $612 million from '90 to '91 in just the database side of business for getting the apps. Can the environment support that sort of thing happening again or is it different this time?

Well, one, I would say, it was a very different licensing model. That was a perpetual license model so you can write it down as revenue. So that's a big difference. And obviously, the market was much more nascent, and so I think it became clear that Oracle emerged as the leader. The market kind of gravitated towards them. We believe that those same dynamics are applying here, where I think people are sort of gravitating to MongoDB. But given our revenue recognition model and so forth, I think it will be -- I don't want to give -- start setting any guidance, but I think, in general, trying to build a subscription business at the same rate as a perpetual upfront business is 2 very different challenges.

(Operator Instructions) We'll go next to Tyler Radke with Citi.

My question was related to the licensing changes. You talked about some positive responses from your customers. I'm just curious what the feedback, if at all, you've had from kind of the major cloud service providers on the licensing change.

Candidly, we haven't had much feedback at all but I think the announcement probably hasn't been lost on anyone because it was pretty widely covered. I don't think -- I think there is a lot of debate in the industry today about how much do cloud providers really contribute back to the open source community. I mean they're very good at taking open source projects, plug it into their cloud platforms and then monetizing it, but they're not so good at giving back to the community, especially when you measure it on a percentage of R&D. About half of our R&D goes towards our free Community Server, which really helps everyone. And there's no cloud provider that comes even close to that, I would say, even spend maybe 1% of their R&D in giving back to community. So I think that speaks to the differences in the contributions back to community. And I think that's become a burning issue for a lot of people. And when it comes to customers, it was really a nonevent because 99.99% of our customers have no interest in offering MongoDB as a service, so it's a moot issue for them.

Great. And as we think about just the mix of the business that's tied to legacy replacement, was there any change there? I think historically it's been about around 25%. And just related to that, how have the use cases evolved with the transaction support, if that's driven any different type of customer, different use case? Just those 2 questions.

Yes, sure. So as I mentioned earlier in one of the earlier questions, we're definitely seeing an increase of mission critical workloads moving to MongoDB. And I mentioned one of the use case, in the prepared remarks, was an energy company moving a lot of its billing data to MongoDB. What's more important in a utility than its billing data? And we're seeing customers start using our capabilities around transactions. We've had a bunch of deals close where that drove the deal. And also gives a lot of confidence to customers, who may not need transactions today but know that it's available for them in the future, to really double down on MongoDB. And so we're definitely seeing that people are viewing MongoDB as truly a viable mission-critical platform to run the most sophisticated demanding workloads. And I'll let Michael answer the question about the mix of legacy to Mongo.

Yes. So given that backdrop, we continue to see very strong increases in the applications that are migrating off of legacy relational workloads. As a reminder, I think we've called this out before, but Atlas probably sees a little bit less than what we've historically seen in terms of the percent migration. And if you look at sort of, like, the overall business, it's probably down on a percentage of new business but if you look at the absolute dollars of migrations that we're driving, they're up significantly.

We'll go next to Jack Andrews with Needham.

It's actually Khanh Ngo filling in for Jack. With the rise of DevOps, we're seeing other employees become more important in the business in the buying cycle. So how are you guys leveraging some of these changes? And it's helped easing motions in sales efficiency, especially if we consider the self-service nature of Atlas.

Yes. So I mean, so we have multiple modes of going to market to customers. One is obviously self-serve, so if a customer just wants to basically engage with us without having to talk to a sales person, they can easily do so and sign up for Atlas in our website, pick out a cloud platform in instance and they literally get a connection string and they're up and running. Then you have smaller customers engage with us through our inside sales channel. That channel has been really effective in pushing Atlas because the product market fit for Atlas is incredibly strong for those customers. And they range from very early stage companies to venture-backed companies. The companies are growing very, very quickly. And we have people around the world who help us close business through that means. And then we have our field organization who work with the largest set of customers. They could be large banks, media, telco, tech companies, et cetera, insurance companies. And there, where the buying process is far more complicated, there's multiple stakeholders that need to be engaged. Typically, the deal sizes are bigger, sometimes it may be a business case required. That tends to be a pretty rigorous technical validation phase where we need to go in and prove the technology really works as advertised or really addresses the problems they're trying to address. That's where you need a field organization, which also includes both salespeople and technical people in the field working with our customers. And yes, I mean, people are moving to more of a DevOps model, where there's a merging of the development resources and operational resources. But in large organizations, there still tends to be a fairly discrete sort of stakeholders that need to be brought on. Above and beyond that, even people like -- people who have a security background, because obviously they're storing their data on MongoDB and they need to validate that we pass all the stress tests that they have around the security of our architecture.

Okay, that make sense. And there are some vendors out there making a bigger push into data warehouses, specifically with Hive and HBase. Although you guys have a different use case and richer query language, are you seeing any changes in the competitive landscape as these competitors make a bigger push in terms of sales and marketing?

No. We're not seeing -- in fact, we're seeing the reverse. We're seeing our customers bring us into more of these analytic workloads and operationalizing a lot of the data lake use cases that they had tried with alternative vendors, where they found it was either too complex, too costly or just took too long to get the time to value and they're bringing us -- and part of the value of using MongoDB is your most up-to-date and accurate information sits in your transactional database. So being able to run analytics using MongoDB, it really allows people to get immediate insights into their business. And that's incredibly valuable to customers.

We'll go next to Chris Eberle with Nomura.

Can you give us a little idea on where we are in the actual approval process with OSI for SSPL?

Sure. So we submitted, in conjunction with our public announcement, we made our first submission to OSI on October 16. And their process tends to be an open and transparent process, so you tend to get a lot of feedback in terms of commentary and so forth. And we had obviously sought out advice and counsel from a lot of open source domain expertise, and we have obviously, as you can imagine, a lot of expertise in-house. And so that drove the issuance of the SSPL. We did get some feedback that we thought merited some deliberation and we've come back with some adjustments, I believe it's about 2 weeks ago. And we'll see what happens from there. I want to be very clear, though, that part of our motivation to work with the OSI is to really get SSPL adopted not just by MongoDB but by the industry as a whole. So that's what our motivation was with the OSI. No matter what the OSI does, we own 100% of the copyright to the code of MongoDB. So like any license, any owner of the IP, we have the right to define the terms of use with our license. So there's no risk for us if, for some reason, OSI doesn't ratify this, or in some cases, rejects or if it ends up rejecting this, this will not change our posture about the use of SSPL. The purpose of you going to the OSI was really to get this established as an industry standard so that others could use this license as we think about their own business models.

We'll go next to Rishi Jaluria with D. A. Davidson.

This is actually Hannah on for Rishi. Just first off, I was wondering if you've received any specific feedback from Atlas customers on any functionalities they wish Atlas had?

Yes, I mean, as you can imagine, we got -- we have lots of customers, and frankly, that's how we run our business. I mean, we constantly listen very, very carefully to what our customers are doing, what new features or capabilities they're asking for or anticipating what new things they may want based on the usage patterns we see. So that's, frankly, part and parcel with how we run our business. So that's kind of ingrained to the core of who we are. We run a lot of customer advisory boards in North America and Europe and soon in Asia to just get that -- continue to make sure we get all the feedback that we think is appropriate along with, obviously, our direct engagements with those customers. I would say one big feature that is on the horizon is customers asking to run the same application across different cloud providers. And that's not just for pure lock-in concerns, but it's also based on customers who want to take advantage of different capabilities on different cloud providers. Because every cloud provider is now obviously trying to compete against other cloud providers. And so that's an area that you'll probably see us make some announcements early next year, and that's something that we're looking at very carefully. And again, we have so many different customers, different -- there's customers pushing on things like performance, scalability and geographic distribution and so forth because we're one of the few platforms, probably the only platform, that really enables customers to deploy a globally distributed database. And I should also point out that given that we run across all the 3 major cloud providers, we have the widest geographic coverage of any database-as-a-service offering, which is a very compelling value proposition to customers. And so that's a big reason why MongoDB is winning.

That makes a lot of sense. And then second, regarding operating margins going forward, are there any specific changes you feel you still need to make? Or do you feel the improvement will just come from top line growth and scaling the business and driving towards more self-serve?

I don't think that there's, like, a particular magic bullet other than just sort of the scaling and operating the business in the way that we have and we'll continue to do. So I wouldn't say there's, like, a specific thing per se, that will suddenly result in a step-function change. I think it's more just sort of the operational blocking and tackling and scaling of the business.

And with no further questions in the queue, I would like to turn the call back over to Dev Ittycheria for any additional or closing remarks.

Well, I'd like to thank everyone for their time today. We're very pleased with our performance in the third quarter, and we will get back to work. So thank you very much for your time. Take care.

And this does conclude today's conference. We thank you for your participation. You may now disconnect.