For the last few years, usage-based pricing has been an excellent strategy for SaaS companies. But there's a question of how it will affect companies in a downturn. Everyone was looking toward Snowflake, one of the largest SaaS companies with usage-based pricing. One should note that at high contract values, usage-based pricing looks more like subscription-based pricing. Committed spend and negotiated discounts help companies have more predictable spend at scale.

However, sometimes usage-based pricing can lead to unpredictable and surprising bills. An unoptimized query or scanning a large table might be costly in a product like Snowflake or BigQuery. Collecting metrics that accidentally explode in cardinality can lead to shocking Datadog bills.

As I wrote in The Cloud Cost Era, many of these services are severely unoptimized by customers. For example, a 5-minute scheduled query may run on a Snowflake cluster with a 15-minute automatic shutdown. If nothing else is running, you'd only be utilizing 25% of your usage-spend. In addition, customers don't have any incentive to optimize with subscription pricing – they pay the same either way. This could be good (customers that would churn in subscription but simply reduce spend in usage) or bad (perceived value).

It boils down to: Is there more bloat in seats or usage?

There's an analogy to companies with network effects – low friction can lead to fast growth. Network effect companies accrue exponential value as more people join the network (e.g., Facebook 2006). However, they also suffer the reverse effect when people leave the platform (e.g., Facebook today). Will companies shrink their SaaS usage just as quickly as they grew it?  

On the other hand, usage pricing will most likely continue to be a great pricing strategy for bottoms-up adoption. Subscription pricing is sometimes prohibitive to potential customers, even with a free tier. As customers become more price-sensitive when adopting SaaS, they will most likely look toward usage-based products. Low usage and small headcount make it easy for a startup to monitor spend.