Since Snowflake’s inception, we’ve had the needs of our customers as our North Star, with a clear focus on security and governance of data. Early on we also committed to continuous innovations to improve performance and reduce latencies, and by virtue of our business model continuously improve the economics for our customers. We want to share our view on why a consumption-based business model is good for customers and our philosophy on continuing to improve performance and pass savings directly on to customers.
Moving from constrained to near-unlimited capacity
From the beginning, Snowflake’s goal has been to enable every organization to be data-driven, resulting in better business outcomes for our customers. We do this by delivering a cloud-built data platform with instant elasticity, per-second pricing, and global availability across the three major clouds. Our success is predicated on our customers being successful when using Snowflake— this means that we continually strive to provide the optimal performance at the best price.
With our collective experience of designing and building on-premises databases, we understood the difficulties that these systems imposed on customers. On-premises systems were built in a world where compute and storage were tightly coupled and the amount available for use was constrained. Because there was a fixed amount of capacity, companies prioritized and limited which data they could most gain insight and business value from. Any insight and business value from the remainder of their data was lost because of these limits. Worse yet, this model resulted in customers purchasing systems that far outstripped their current needs. They were required to purchase systems that they estimated would meet their needs for expected peak capacity 3+ years into the future, resulting in significant upfront costs and wasted capacity. With the rapid growth in number and size of data, customers increasingly found this model unsustainable.
We wanted our customers to take full advantage of their data at a cost aligned with business impact when they adopted Snowflake. We built Snowflake to take advantage of the cloud’s near-unlimited scale, so that our customers can run as many workloads as needed. Teams within a company no longer need to fight for slices from a fixed amount of compute. This enables them to gain insight and business value from all their data on demand. And in this world of near-unlimited capacity, we wanted to ensure that customers pay for only the precise resources used on Snowflake—and no more. They can dynamically add and remove workloads, focusing on the needs of the business and not limited by the availability of infrastructure resources.
With the ability to use near-unlimited capacity, we’ve shifted the dynamics for customers to managing abundant resources as opposed to scarce infrastructure. We know that this requires a new way of operating, and to help with this we want to ensure that customers have detailed visibility and control of their Snowflake usage. Companies can clearly understand resource usage and make fine-grained decisions on the spend associated with every workload they run on Snowflake. Using this visibility and control, companies can empower individual organizations or business units to manage their spend on those workloads with high business value. Our larger customers have leveraged this information to implement departmental chargebacks. And they can monitor their usage and billing on demand to understand their spend for each workload.
Fast keeps getting faster
For as long as databases have existed, there has been a relentless focus on improving the performance of these systems. Without that continual improvement, any database solution would quickly become irrelevant. In the on-premises world, these performance improvements were delivered to customers in newer, updated versions of the database system (along with many other feature improvements). Customers had to spend significant resources to upgrade and migrate to newer versions of the database engine to take advantage of any performance improvement. Because these new versions were delivered every few years, this meant customers’ workloads only gained the benefits after extremely long periods of time.
At Snowflake, improving the performance of our product, and especially the core database engine, is central to our product philosophy. With Snowflake’s weekly release process, we regularly release performance improvements to our customers that require no effort or cost to adopt. Any workload using functionality we’ve improved gains the benefits of these changes seamlessly and transparently. Those workloads immediately run faster with zero effort from customers. And because of our consumption-based pricing, these faster workloads often result in direct cost savings to our customers. This relationship between our continued focus on improving performance and the reduced costs which improves our customers satisfaction highlights how our consumption-based business model aligns our incentives with those of our customers. In the end, more performant workloads, timelier transformations, and faster decision-making correlate to increased business value for both our customers and Snowflake.
To many this may seem counter-intuitive. We invest in performance improvements incurring research and development costs which, in turn, result in improved functionality, increased speed, and reduced costs to our customers, which reduces consumption of our platform in the short term. Conventional wisdom would equate this to Snowflake investing to lose revenue! However, Snowflake has been successful in challenging conventional wisdom. Our persistent focus on the needs of our customers means increasing business value for them—and they respond by using more of the performance, flexibility, and scalability available within the Snowflake Data Cloud, which can increase consumption of our platform in the long term.And, they contribute to our industry-leading Net Promoter Score (NPS) of 72,1 a score more than three times the industry average of 21, based on the Qualtrics 2021 NPS Industry Benchmarking Report.
Our continuing commitment
Snowflake is committed to continuing to improve the performance of our product and deliver direct cost savings for our customers. What’s more, by having one unified product, we can more efficiently deliver improvements that span all parts of the system, including the core engine, storage, hardware optimizations, and more. This has resulted in significant positive impact to our customers. In our last several Snowflake Summits we’ve announced the latest set of performance improvements and efficiencies and we will continue to do so. This philosophy has yielded a >20% reduction in the average cost of warehouse queries for customers over the last 3 years. And this number does not include taking advantage of new hardware advances – we pass those savings on to customers as well.
Here are three examples from 2022 which demonstrate our approach in action and which have yielded significant simultaneous performance and cost improvements for our customers. Those who use the Search Optimization Service are spending on average 25% less on Search Optimization compute. This is because we optimized the background maintenance tasks to consume less resources. On the compilation side, common table expressions (CTEs), a convenient way to simplify and manage complex SQL queries, are now faster, with up to 30% reduced compilation time, and are correspondingly less expensive. We also announced improvements, based on new hardware capabilities we leveraged, that are projected to save customers roughly $97 million in aggregate over the next year. Taking advantage of new hardware advances and passing those savings onto customers enables them to both reduce their existing bill with Snowflake and move additional workloads over to Snowflake, likely for even less than their original bill. For details on selected specific improvements, visit the Inside the Data Cloud Blog. The ongoing performance and cost improvements we deliver in Snowflake can always be found here.
We are committed to continuing to improve the value that customers gain from using Snowflake. We’ve found that partnering closely with our customers to understand their needs enables us to build functionality and deliver improvements that have a direct impact on helping them more effectively run their business. This ensures that our customers continue to gain more value, for less cost, with Snowflake.
1 As of June 2022. If a customer fails to (i) respond to each required question in the survey or (ii) submit a complete set of responses by the end of the survey period, we consider that customer’s survey incomplete. Starting with our NPS as of June 2022, we exclude incomplete survey responses from the calculation.
Originally posted on November 15, 2022 @ 2:04 am
