Total revenue for the quarter was $159.6 million, up 20% sequentially and 119% year-over-year growth. Product revenue, which is based on platform consumption, was $148.5 million, up 115% year over year
Remaining performance obligation (RPO) increased 240% year over year to $928 million
Net revenue retention rate was 162%, up from 158% in 2Q
Added 437 new customers in 3Q to take total customer count to 3,554, up 84% year over year
Ended the quarter with 65 customers with over $1M in trailing 12 month revenue, up 110% year over year.
Expects 4Q21 product revenue between $162-167 million, representing year over year growth between 113-115%
Earnings Call Highlights
Impact from Covid-19 has been neutral.
The pandemic has been more or less neutral to our business. Some businesses were negatively affected in terms of demand sentiment, but others stepped up their data strategy given the new complexities of the health crisis and economic effects.
Management emphasized that Snowflake's consumption-based business model is different from SaaS.
It bears repeating that Snowflake is not a SaaS business model. We're a consumption company, and our reported revenue has a direct relationship with the consumption of our platform during the period. The consumption model is variable, not fixed, meaning our model places no limits on how much of our platform a customer can consume, and this contributes to our strong revenue retention rate.
We are not a SaaS model. We are a consumption model. Our business model is a key differentiator for us and is designed to drive customer success. Our customers purchase credits, and when those credits are consumed, we recognize the revenue. Unlike a ratable model, we only recognize revenue if the customer uses our platform. For this reason, there is no shelfware in our revenue.
Management said that as a consumption-based model, product revenue and RPO are key metrics that best represent the company's performance.
For these reasons, we do not focus on the same metrics that a SaaS business would. We focus on product revenue and remaining performance obligation. Product revenue, which excludes professional services and other revenue, is the most transparent disclosure we offer. It gives full insight into how our customers are actually using our product in the period reported. If a customer purchases credits and does not consume, their revenue will be $0.
Unlike most SaaS businesses, billings is not a meaningful metric for us because it is less correlated to product revenue due to the variability of consumption.
On average, customers are consuming more than expected
we do see in certain industries, like we have some customers that are in the travel industry, we see their consumption down. But we have a lot of customers that are kind of more in the online consumer world that their business is booming, and their consumption is much higher than we're forecasting. So it all depends upon the industry they're in, but on average, we're seeing our customers consume more than we would expect. That's why we ended up beating by what we did. It was a higher beat than I was expecting for the quarter.
Gross margin improved from 62% in 2Q to 70% in the third quarter due to:
Renegotiated prices with cloud vendors
Increased scale due to global deployments
Better discipline in discounting
Management expects the gross margin to be 70% in 4Q, and improve to mid-70s in the long term.
So as you saw for the most recent quarter, we just did 70% gross margin. And the implied for the full year is 68%, but we're actually going to do 70% margin in Q4. As we talked about when we were going public, I do think longer term, in terms of model, we can get to the mid-70s.
One is better pricing with our cloud vendors. We did just renegotiate deals with 2 of our cloud vendors, AWS and Azure, who are the majority of virtually all of our businesses running there today. And I do think as we continue to grow, we'll be able to renegotiate those again.
Scale, we have a lot of deployments around the world where we're not even close to being at scale. And especially as EMEA is starting to take off for us, we'll get more scale in those, which will drive better margins for us.
And then also, we're seeing a lot better discipline in our field around discounting. And if you see the average price per credit we're getting, it continues to increase.
Better discipline in sales force and improved product is driving better pricing.
I would say a lot of it is our better discipline in our sales force, but then also given the reference customers we have and what customers are seeing, they're more willing to move to Snowflake as well too, which makes the procurement process easier.
But what I will also tell you today, the product we're delivering today is so much better than what it was 3 years ago or 2 years ago, continues, and it becomes more -- the performance on it every year gets better and better. And as a result, customers should pay more for it.
Accelerated RPO growth being driven by larger, multiyear deals.
As we're moving more into large enterprises, large enterprises really want to do multiyear deals. They're not interested in having to go to procurement every year. And so we saw a number of large enterprises commit to multiyear deals with us, and we think that is going to be a trend that will continue. And I will say it wasn't until this year that our sales force really started pushing more 3-year deals, and we're going to continue to do that going forward in the future.
Snowflake recently announced support for unstructured data. Management said there there has been strong customer demand for this new offering.
there is ferocious appetite in our customer base for us expanding the scope of our capabilities, both in terms of workloads and in terms of data types, our ability to use external services. So it's a very broadly capable platform. Customers don't want a multitude of platforms in their environment. They're very, very intent of bringing as much data as they possibly can onto Snowflake and running as many workloads as they can on Snowflake, and we're running hard to enable that.
They explained that unstructured data such as videos and images are used in machine learning, which should drive additional compute loads for Snowflake. Moreover, these data types are usually very large so they should drive increased storage usage on the platform as well.
image data, video data, I mean, PDFs, social media. And by the way, this is also where ML services are going to become -- machine learning services are going to be really interesting because, say, I have image scans from webcams somewhere in the city, and I'm able to shoot that over to an external service, have it scanned, too, whether that's a recognizable, known person. There's going to be enormous potential for this. And we're...getting pushed by our customers to really bring these different data types onto our platform because they're just part of the types of analytical processes that they want to run. So that's sort of where we see these initially going. And these data types are large. This will definitely add to the storage side of the equation as well.
Snowflake has also recently announced Snowpark, a developer service which enables users to write code in their preferred language to build data transformations and score machine learning models, all processed on the platform.
So what Snowpark will do is we will start hosting the language run times inside the Snowflake platform so it becomes a completely optimized experience. It dramatically expands the workload scope of Snowflake, so it's a very, very important direction for us.
Management expects Snowpark to drive increased platform usage.
in terms of how does this help. Our entire business model, as we've said over and over, is based on consumption. So anything that drives consumption of our query engine and our platform accrues to our business model. So our whole strategy is driven to move the dial on consumption. And this is definitely going to do it.
Snowflake wants to remain a neutral platform when it comes to machine learning algorithms because that is the best strategy to drive increased usage.
So we're not going to be a company where we have our own flavor of everything and our partners are all going to be secondary. We want to actively encourage development on our platform, participation on our platform because, again, we're a consumption company. So the way we drive our strategy is to drive activity on our platform, right? I mean if we get activity on our platform, whether our product drives it or a partner drives it, it yields the same result, right? So we'd be crazy to -- because the way our model works, we don't have to prioritize our own products.
Product revenue increased $23 million sequentially in 2Q and 3Q, but the 4Q guide implies an increase of just $16 million, when 4Q is usually the strongest. How should investors think about that?
There's also a number of our large customers that we've given that are getting more mature right now, too, and their growth has slowed down somewhat with some of these large customers. But there is seasonality within our customer base. And I would not assume Q4 is what's driving their consumption at all. Quite frankly, in Q4, there's a lot of holidays. You have Thanksgiving, you have Christmas, and you do see decrease in the New Year. You do see decreases because employees aren't in the office driving analytics.