Beyond Subscriptions: Understanding Usage-Based Pricing Models

While usage-based models have been the norm in infrastructure (ex: AWS, Azure, GCP) for a long time, they’re now also found in horizontal and vertical application software.

Let’s understand the reason behind this:

Public companies that use consumption see 38% faster revenue growth over their SaaS peers, 50% higher revenue multiples, and top-quartile net dollar retention is 120%, compared to 110% for the broader SaaS index.

Due to this — Usage-Based pricing seemed inevitable. Then, last year, it stalled.

Usage-based pricing growth rates overreact when businesses are spending, and when businesses cut. In a year of efficiency — 2023 — UBP, the model itself, also saw cuts.

While building a usage-based pricing model, there are at least 3 unique models:

• Value metric based
• Success-based
• Credit Drawdowns

Check out the deep dive with the inimitable Kyle Poyar if you want to learn more about usage-based pricing.

𝗩𝗮𝗹𝘂𝗲 𝗺𝗲𝘁𝗿𝗶𝗰 𝗯𝗮𝘀𝗲𝗱

Traditional usage-based pricing revolves around finding a value metric that:

• The customer perceives as important and wants more of
• Helps you price discriminate
• Scales with your costs

For example, Stripe’s metric is transaction volume.

They charge a percentage of the transaction amount along with a fixed fee per transaction.

• Customers want more transaction volume and the ones with more willingness to pay have higher transaction volume.
• Stripe’s costs scale by transaction volume, as well.

𝗦𝘂𝗰𝗰𝗲𝘀𝘀-𝗯𝗮𝘀𝗲𝗱 𝗽𝗿𝗶𝗰𝗶𝗻𝗴

A special variant of value metric is a success. Intercom is one of the most well-known examples of this. Being a customer support technology provider, it charges per successful resolution.

It’s a risk.

Successful resolutions may not scale with their costs — if they aren’t solving as many customer issues.

But, it’s more aligned to customer value. It sets the right incentives for internal teams, and encourages large customers to grow with you.

The biggest users of usage-based pricing are actually Facebook and Google ads, which do hundreds of billions yearly on pay per conversion and pay per click models. It scales really well.

𝗖𝗿𝗲𝗱𝗶𝘁 𝗗𝗿𝗮𝘄𝗱𝗼𝘄𝗻𝘀

The final common model in usage-based pricing is to have a concept of a ‘credit’ (there may even be multiple credit types) that you drawdown.

Snowflake is one of the most successful users of this credit drawdown model. They have 150%+ NDR based on a model that requires a 13-page PDF to fully understand. It depends on the warehouse you use, whether it’s snowspark optimized, and a whole lot more.

And that’s just on the compute side. There’s also the storage side.

The takeaway is: a credit drawdown program doesn’t have to be super simple, but users have to be able to understand it, and be able to take predictable steps to optimize it — if that’s their objective.